diff --git a/.github/workflows/publish-skills.yml b/.github/workflows/publish-skills.yml new file mode 100644 index 00000000..f6202691 --- /dev/null +++ b/.github/workflows/publish-skills.yml @@ -0,0 +1,78 @@ +name: publish-skills + +on: + push: + branches: [feat/bailian-docs-update] + paths: ['skills/**'] + +permissions: + contents: write + pull-requests: write + +concurrency: + group: publish-skills + cancel-in-progress: false + +jobs: + publish: + runs-on: ubuntu-latest + steps: + - name: Checkout + uses: actions/checkout@v4 + with: + fetch-depth: 0 + + - name: Poke FC publisher (publish-skills) + run: | + # FC uses reconciliation-based publishing: request body is not trusted. + # curl -f only catches HTTP errors; FC returns 200 + success:false on + # business failure, so we also parse the response body to surface it. + RESP=$(curl -fsS --retry 3 --retry-delay 5 -X POST \ + "${{ vars.FC_TRIGGER_URL }}/publish-skills") + echo "$RESP" | jq . + if [ "$(echo "$RESP" | jq -r '.success')" != "true" ]; then + echo "::error::publish-skills failed: $(echo "$RESP" | jq -r '.error // "unknown"')" + exit 1 + fi + + echo "### publish-skills result" >> "$GITHUB_STEP_SUMMARY" + echo '```json' >> "$GITHUB_STEP_SUMMARY" + echo "$RESP" | jq '{upserted, deleted, unchanged: (.unchanged | length), durationMs}' >> "$GITHUB_STEP_SUMMARY" + echo '```' >> "$GITHUB_STEP_SUMMARY" + + - name: Create or update PR to main + id: pr + env: + GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} + run: | + BRANCH="${{ github.ref_name }}" + TITLE="Auto PR: merge ${BRANCH} into main" + + EXISTING_PR=$(gh pr list --head "$BRANCH" --base main --state open --json number --jq '.[0].number') + + if [ -n "$EXISTING_PR" ]; then + echo "PR #$EXISTING_PR already exists, updating..." + gh pr edit "$EXISTING_PR" --title "$TITLE" + PR_NUMBER="$EXISTING_PR" + else + echo "Creating new PR..." + PR_NUMBER=$(gh pr create \ + --base main \ + --head "$BRANCH" \ + --title "$TITLE" \ + --body "Automated PR created on push to \`${BRANCH}\`." \ + --no-maintainer-edit) + fi + + echo "number=$PR_NUMBER" >> "$GITHUB_OUTPUT" + + - name: Auto-merge PR + env: + GH_TOKEN: ${{ secrets.GITHUB_TOKEN }} + run: | + PR_NUMBER="${{ steps.pr.outputs.number }}" + echo "Merging PR #$PR_NUMBER..." + gh pr merge "$PR_NUMBER" --squash --delete-branch=false + + echo "### PR #$PR_NUMBER merged" >> "$GITHUB_STEP_SUMMARY" + echo "Branch \`${{ github.ref_name }}\` has been merged into \`main\`." >> "$GITHUB_STEP_SUMMARY" diff --git a/skills/bailian-docs-llm-wiki/SKILL.md b/skills/bailian-docs-llm-wiki/SKILL.md index 7ec690da..4c8b7d7e 100644 --- a/skills/bailian-docs-llm-wiki/SKILL.md +++ b/skills/bailian-docs-llm-wiki/SKILL.md @@ -148,22 +148,6 @@ description: >- | **按家族筛选**：按 primaryCapability / providers / itemCount / maxContextWindow 找家族 | `models/families.jsonl`（一行一家族，含 items[] 摘要） | | 模型家族总览 / 按能力分桶浏览 | `models/index.md` | | 主题页 / API 文档（按功能领域查找） | `wiki/index.md`（完整索引入口） | -| OpenAI 兼容接口 | `wiki/concepts/openai-compatible-interface.md` | -| API Key 鉴权 | `wiki/concepts/api-key.md` | -| 函数调用（Function Calling） | `wiki/concepts/function-calling.md` | -| 检索增强生成（RAG） | `wiki/concepts/rag.md` | -| 异步调用与任务轮询 | `wiki/concepts/async-invocation.md` | -| Token 与计费 | `wiki/concepts/token-and-billing.md` | -| 流式输出 | `wiki/concepts/streaming-output.md` | -| 业务空间（Workspace） | `wiki/concepts/workspace.md` | -| 模型调优与部署 | `wiki/concepts/fine-tuning-and-deployment.md` | -| MCP 与工具扩展 | `wiki/concepts/mcp-and-tools.md` | -| 模型微调、压缩与部署对比 | `wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md` | -| 模型评估与模型监控对比 | `wiki/comparisons/model-evaluation-vs-monitoring.md` | -| 图像、视频与 3D 生成对比 | `wiki/comparisons/image-vs-video-vs-3d-generation.md` | -| 应用评估与应用监控对比 | `wiki/comparisons/app-evaluation-vs-monitoring.md` | -| 知识库与记忆库对比 | `wiki/comparisons/knowledge-base-vs-memory-library.md` | -| 托管智能体：指南与 API 对比 | `wiki/comparisons/managed-agents-guide-vs-api.md` | > 实际文件名以 `wiki/index.md` 为准；上表若有出入应回到索引页查找。 diff --git a/skills/bailian-docs-llm-wiki/llms.txt b/skills/bailian-docs-llm-wiki/llms.txt index f711c7c8..62c019a4 100644 --- a/skills/bailian-docs-llm-wiki/llms.txt +++ b/skills/bailian-docs-llm-wiki/llms.txt @@ -4,25 +4,32 @@ ## 模型使用指南 -- **模型体验** - - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) - - [图片生成与编辑](raw/model-user-guide/model-experience/image-model.md) - - [视频生成与编辑](raw/model-user-guide/model-experience/video-generate-edit-model.md) - - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) - - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - - [语音合成](raw/model-user-guide/model-experience/tts-model.md) - - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) - - [语音识别](raw/model-user-guide/model-experience/asr-model.md) - - [语音转语音](raw/model-user-guide/model-experience/s2s-model.md) - - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) - - [全模态](raw/model-user-guide/model-experience/omni.md) - **开始使用** - [什么是阿里云百炼](raw/model-user-guide/get-started-with-models/what-is-model-studio.md) - [首次调用千问API](raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) - [选择模型](raw/model-user-guide/get-started-with-models/models.md) - [Base URL总览](raw/model-user-guide/get-started-with-models/base-url.md) - - [选择地域、服务部署范围和接入域名](raw/model-user-guide/get-started-with-models/regions.md) - [限流](raw/model-user-guide/get-started-with-models/rate-limit.md) + - [地域及接入域名](raw/model-user-guide/get-started-with-models/regions.md) +- **Token Plan** + - **团队版** + - [概述](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md) + - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md) + - [团队管理](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md) + - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md) + - **个人版** + - [概述](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md) + - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md) + - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md) + - **最佳实践** + - [接入 Harness 工具](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md) + - [接入多模态生成模型](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) + - [添加视觉理解能力](raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md) + - [联网搜索](raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md) + - **Coding Plan** + - [Coding Plan概述](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) + - [常见问题](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) + - [Token Plan 概述](raw/model-user-guide/token-plan-guide/token-plan-overview.md) - **产品计费** - [新人免费额度](raw/model-user-guide/test-1/new-free-quota.md) - [模型训练与部署计费](raw/model-user-guide/test-1/model-training-and-deployment-billing.md) @@ -30,87 +37,434 @@ - [账单查询与成本管理](raw/model-user-guide/test-1/bill-query-and-cost-management.md) - [模型调用价格](raw/model-user-guide/test-1/model-pricing.md) - **接入客户端/开发工具** - - [Hermes Agent](raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [OpenClaw](raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) + - [Hermes Agent](raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [Claude Code](raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) - [OpenCode](raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - [Cursor](raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) - [Codex](raw/model-user-guide/use-chat-client-or-development-tool/codex.md) + - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - [QwenPaw](raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) - - [Cline](raw/model-user-guide/use-chat-client-or-development-tool/cline.md) - [Cherry Studio](raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) - - [Chatbox](raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) - [Qoder](raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) + - [Cline](raw/model-user-guide/use-chat-client-or-development-tool/cline.md) + - [Chatbox](raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) - [Qoder CN（原 Lingma）](raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md) - [Kilo CLI](raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [使用Postman或cURL调用图像/视频生成API](raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) - [Dify](raw/model-user-guide/use-chat-client-or-development-tool/dify.md) - [更多工具](raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) -- **Token Plan（团队版）** - - **最佳实践** - - [工具调用](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md) - - [接入多模态生成模型](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) - - **Coding Plan** - - [Coding Plan概述](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) - - [联网搜索](raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md) - - [添加视觉理解能力](raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md) - - [常见问题](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) - - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) - - [Token Plan（团队版）概述](raw/model-user-guide/token-plan-guide/token-plan-overview.md) - - [团队管理](raw/model-user-guide/token-plan-guide/token-plan-team.md) - - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-faq.md) +- **模型体验** + - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) + - [图片生成与编辑](raw/model-user-guide/model-experience/image-model.md) + - [视频生成与编辑](raw/model-user-guide/model-experience/video-generate-edit-model.md) + - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) + - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) + - [语音合成](raw/model-user-guide/model-experience/tts-model.md) + - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) + - [语音识别](raw/model-user-guide/model-experience/asr-model.md) + - [全模态](raw/model-user-guide/model-experience/omni.md) + - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) + - [语音转语音](raw/model-user-guide/model-experience/s2s-model.md) - **模型推理** - [TPM 预留](raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - [快速模式](raw/model-user-guide/model-high-speed-inference/fast-mode.md) -- **模型部署** - - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - - [预置吞吐长输入与缓存](raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) - - [模型导入](raw/model-user-guide/model-deployment-1/model-import.md) - - [使用 API或命令行进行模型部署](raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) - **模型调优** - **千问模型调优** - [模型调优简介](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md) - [在控制台进行模型调优](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md) - - [0 代码强化大模型安全合规能力](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - [使用 API 或命令行进行模型调优](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md) + - [0 代码强化大模型安全合规能力](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - **语音合成模型调优** - [CosyVoice模型调优](raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) - [微调图像生成模型](raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md) - [微调视频生成模型](raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md) + - [强化学习训练概述](raw/model-user-guide/fine-tuning/rl-training-overview.md) +- **模型部署** + - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) + - [预置吞吐长输入与缓存](raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) + - [模型导入](raw/model-user-guide/model-deployment-1/model-import.md) + - [使用 API或命令行进行模型部署](raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) - **模型评测** - [模型评测](raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) - [评测维度](raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) -- **模型压缩** - - [模型压缩](raw/model-user-guide/model-compression/model-compression-introduction.md) - **用量统计与性能监控** - [模型用量](raw/model-user-guide/model-monitoring/model-usage-statistics.md) - [模型监控](raw/model-user-guide/model-monitoring/model-telemetry.md) +- **模型压缩** + - [模型压缩](raw/model-user-guide/model-compression/model-compression-introduction.md) - **模型数据** - [训练集与评测集](raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) - [数据清洗或增强](raw/model-user-guide/model-data-overview/data-processing.md) + - [日志回流](raw/model-user-guide/model-data-overview/model-log-backflow.md) - **安全合规** + - **传输安全** + - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) + - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) + - [通过终端节点私网访问阿里云百炼模型或应用 API](raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) - **安全存储** - [配置终端节点并发起连接](raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) - - [配置可用区IP](raw/model-user-guide/security-and-compliance/secure-storage/configure-zone-ip.md) - [配置MSE云原生网关](raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) + - [配置可用区IP](raw/model-user-guide/security-and-compliance/secure-storage/configure-zone-ip.md) - [配置私有网络中的资源](raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md) - - **传输安全** - - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) - - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) - - [通过终端节点私网访问阿里云百炼模型或应用 API](raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) - [权限管理](raw/model-user-guide/security-and-compliance/permission-management-overview.md) - - [输⼊输出AI安全护栏](raw/model-user-guide/security-and-compliance/content-security.md) - [模型备案信息公示](raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) + - [输⼊输出AI安全护栏](raw/model-user-guide/security-and-compliance/content-security.md) - [千问大模型应用上架及合规备案](raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) - [合规资质与隐私说明](raw/model-user-guide/security-and-compliance/privacy-notice.md) +- **服务支持** + - **模型列表** + - **文本生成** + - [qwen3.7-max](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-7-max.md) + - [qwen3.7-max-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-7-max-us.md) + - [qwen3.6-max-preview](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-6-max.md) + - [qwen-max](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-max.md) + - [qwen3-max](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/model-qwen3-max.md) + - [qwen3.6-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-6-plus.md) + - [qwen3.7-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-7-plus.md) + - [qwen-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-plus.md) + - [qwen3.5-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-5-plus.md) + - [qwen-plus-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-plus-us.md) + - [qwen3.6-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-6-flash.md) + - [qwen3.7-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-7-flash.md) + - [qwen3.6-flash-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-6-flash-us.md) + - [qwen3.5-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-5-flash.md) + - [qwen-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-flash.md) + - [qwen-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-turbo.md) + - [qwq-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwq-plus.md) + - [qwen-plus-character](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-plus-character.md) + - [qwen-plus-character-ja](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-plus-character-ja.md) + - [qwen-flash-character](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-flash-character.md) + - [qwen-long](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-long.md) + - [qwen-math-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-math-plus.md) + - [qwen-math-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-math-turbo.md) + - [qwen3-coder-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-coder-flash.md) + - [qwen3-coder-next](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-coder-next.md) + - [qwen3-coder-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-coder-plus.md) + - [qwen-flash-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-flash-us.md) + - [qwen-coder-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-coder-plus.md) + - [qwen-mt-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-mt-plus.md) + - [qwen-coder-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-coder-turbo.md) + - [qwen-mt-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-mt-flash.md) + - [qwen-mt-lite](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-mt-lite.md) + - [qwen-mt-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-mt-turbo.md) + - [qwen-mt-lite-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-mt-lite-us.md) + - [qwen-doc-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen-doc-turbo.md) + - [tongyi-intent-detect-v3](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/tongyi-intent-detect-v3.md) + - [qwen3.7-plus-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-7-plus-us.md) + - [farui-plus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/farui-plus.md) + - [tongyi-xiaomi-analysis-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/tongyi-xiaomi-analysis-flash.md) + - [qwen3-coder-30b-a3b-instruct](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-coder-30b-a3b-instruct.md) + - [qwen3-coder-480b-a35b-instruct](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-coder-480b-a35b-instruct.md) + - [qwen3.6-27b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-6-27b.md) + - [qwen3.6-35b-a3b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-6-35b-a3b.md) + - [qwen3.5-122b-a10b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-5-122b-a10b.md) + - [qwen3.5-27b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-5-27b.md) + - [qwen-deep-research](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/model-qwen-deep-research.md) + - [qwen3.5-397b-a17b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-5-397b-a17b.md) + - [tongyi-xiaomi-analysis-pro](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/tongyi-xiaomi-analysis-pro.md) + - [qwen3-235b-a22b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-235b-a22b.md) + - [qwen3-235b-a22b-instruct-2507](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-235b-a22b-instruct-2507.md) + - [qwen3-235b-a22b-thinking-2507](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-235b-a22b-thinking-2507.md) + - [qwen3-30b-a3b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-30b-a3b.md) + - [qwen3-30b-a3b-instruct-2507](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-30b-a3b-instruct-2507.md) + - [qwen3-32b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-32b.md) + - [qwen3-14b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-14b.md) + - [qwen3-8b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-8b.md) + - [qwen3.5-35b-a3b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-5-35b-a3b.md) + - [qwen3-next-80b-a3b-instruct](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-next-80b-a3b-instruct.md) + - [deepseek-v4-pro](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-pro.md) + - [deepseek-v4-pro-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-pro-us.md) + - [deepseek-v4-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-flash.md) + - [deepseek-v4-flash-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-flash-us.md) + - [deepseek-v3.2](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-3.md) + - [deepseek-v3.2-exp](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-2-exp.md) + - [vanchin/deepseek-v3.2-think](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-2-think.md) + - [qwen3-next-80b-a3b-thinking](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/qwen3-next-80b-a3b-thinking.md) + - [vanchin/deepseek-v3.1-terminus](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-1-terminus.md) + - [deepseek-v3](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3.md) + - [deepseek-r1](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1.md) + - [deepseek-r1-distill-qwen-1.5b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-1-5b.md) + - [deepseek-r1-distill-qwen-14b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-14b.md) + - [deepseek-r1-distill-qwen-32b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-32b.md) + - [deepseek-r1-distill-qwen-7b](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-7b.md) + - [vanchin/deepseek-ocr](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-ocr.md) + - [deepseek-v3.1](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-1.md) + - [kimi/kimi-k3](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k3.md) + - [kimi-k2.7-code](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-7-code.md) + - [kimi/kimi-k2.7-code-highspeed](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-7-code-highspeed.md) + - [kimi-k2.6](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-6.md) + - [kimi-k2.5](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-5.md) + - [Moonshot-Kimi-K2-Instruct](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/moonshot-kimi-k2-instruct.md) + - [kimi-k2-thinking](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-thinking.md) + - [glm-5.2-fast-preview](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2-fast.md) + - [glm-5.2](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2.md) + - [glm-5.1](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-1.md) + - [glm-5.2-us](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2-us.md) + - [glm-4.7](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-7.md) + - [glm-5](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-9.md) + - [glm-4.6](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-6.md) + - [glm-4.5](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4.md) + - [glm-4.5-air](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-5-air.md) + - [MiniMax/MiniMax-M3](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m3.md) + - [MiniMax/MiniMax-M2.7](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2-7.md) + - [MiniMax-M2.5](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2-5.md) + - [MiniMax-M2.1](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2.md) + - [xiaomi/mimo-v2.5-pro](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/mimo-v2-5-pro.md) + - [stepfun/step-3.7-flash](raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/step-3-7-flash.md) + - **视觉理解** + - [qwen3-vl-flash](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-flash.md) + - [qwen3-vl-flash-us](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-flash-us.md) + - [qwen3-vl-plus](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-plus.md) + - [qwen-vl-max](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen-vl-max.md) + - [qwen-vl-ocr](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwenvl-ocr.md) + - [qwen-vl-plus](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen-vl-plus.md) + - [qwen3.5-ocr](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-5-ocr.md) + - [gui-plus](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/gui-plus.md) + - [qvq-max](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qvq-max.md) + - [qvq-plus](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qvq-plus.md) + - [qwen3-vl-235b-a22b-thinking](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-235b-a22b-thinking.md) + - [qwen3-vl-235b-a22b-instruct](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-235b-a22b-instruct.md) + - [qwen3-vl-30b-a3b-instruct](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-30b-a3b-instruct.md) + - [qwen3-vl-30b-a3b-thinking](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-30b-a3b-thinking.md) + - [qwen3-vl-32b-instruct](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-32b-instruct.md) + - [qwen3-vl-32b-thinking](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-32b-thinking.md) + - [qwen3-vl-8b-instruct](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-8b-instruct.md) + - [qwen3-vl-8b-thinking](raw/model-user-guide/support/model-studio-model-list/model-list-visual-understanding/qwen3-vl-8b-thinking.md) + - **全模态** + - [qwen3.5-omni-flash](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-flash.md) + - [qwen3-omni-flash](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-omni-flash.md) + - [qwen-omni-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen-omni-turbo.md) + - [qwen3.5-omni-plus-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-plus-realtime.md) + - [qwen3.5-omni-flash-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-flash-realtime.md) + - [qwen3-omni-flash-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-omni-flash-realtime.md) + - [qwen-omni-turbo-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen-omni-turbo-realtime.md) + - [qwen2.5-omni-7b](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen2-5-omni-7b.md) + - [qwen3.5-omni-plus](raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-plus.md) + - **图像生成** + - [qwen-image-3.0-pro](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-3-0-pro.md) + - [qwen-image-2.0-pro](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-2-0-pro.md) + - [qwen-image-2.0](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-2-0.md) + - [qwen-image-edit-max](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-edit-max.md) + - [qwen-image-edit-plus](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-edit-plus.md) + - [qwen-image-edit](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-qwen-image-edit.md) + - [qwen-image-max](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-max.md) + - [qwen-image-plus](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-plus.md) + - [qwen-image](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image.md) + - [qwen-mt-image](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-mt-image.md) + - [wan2.7-image-pro](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-7-image-pro.md) + - [wan2.7-image](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-7-image.md) + - [wan2.6-image](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-6-image.md) + - [wan2.5-i2i-preview](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-5-i2i.md) + - [wan2.6-t2i](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-6-t2i.md) + - [wan2.5-t2i-preview](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-5-t2i.md) + - [wan2.2-t2i-plus](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-2-t2i-plus.md) + - [wan2.2-t2i-flash](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-2-t2i-flash.md) + - [wanx2.1-imageedit](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-imageedit.md) + - [wanx2.1-t2i-plus](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-t2i-plus.md) + - [wan2.1-t2i-plus](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-1-t2i-plus.md) + - [wan2.1-t2i-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-1-t2i-turbo.md) + - [wanx2.1-t2i-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-t2i-turbo.md) + - [wanx2.0-t2i-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-0-t2i-turbo.md) + - [wanx-background-generation-v2](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-background-generation-v2.md) + - [wanx-poster-generation-v1](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-poster-generation-v1.md) + - [wanx-style-repaint-v1](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-style-repaint-v1.md) + - [wanx-sketch-to-image-lite](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-sketch-to-image-lite.md) + - [wanx-v1](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-v1.md) + - [wanx-x-painting](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-x-painting.md) + - [wanx-virtualmodel](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-virtualmodel.md) + - [wordart-semantic](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wordart-semantic.md) + - [wordart-texture](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wordart-texture.md) + - [z-image-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/z-image-turbo.md) + - [shoemodel-v1](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/shoemodel-v1.md) + - [virtualmodel-v2](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/virtualmodel-v2.md) + - [aitryon-parsing-v1](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-parsing-v1.md) + - [aitryon](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon.md) + - [aitryon-refiner](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-refiner.md) + - [aitryon-plus](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-plus.md) + - [facechain-facedetect](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/facechain-facedetect.md) + - [image-erase-completion](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/image-erase-completion.md) + - [image-instance-segmentation](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-image-instance-segmentation.md) + - [facechain-generation](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-facechain-generation.md) + - [image-out-painting](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/image-out-painting.md) + - [kling/kling-v3-image-generation](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/kling-v3-image-generation.md) + - [kling/kling-v3-omni-image-generation](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/kling-v3-omni-image-generation.md) + - [vidu/vidu-image_reference2image](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/vidu-image-reference2image.md) + - [vidu/viduq2-fast_reference2image](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq2-fast-reference2image.md) + - [vidu/viduq2-pro_reference2image](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq2-pro-reference2image.md) + - [vidu/viduq3-fast_reference2image](raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq3-fast-reference2image.md) + - **视频生成** + - [happyhorse-1.0-t2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/happyhorse-1-0-t2v.md) + - [happyhorse-1.1-t2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/happyhorse-1-1-t2v.md) + - [happyhorse-1.1-i2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/happyhorse-1-1-i2v.md) + - [happyhorse-1.1-r2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/happyhorse-1-1-r2v.md) + - [happyhorse-1.0-video-edit](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/happyhorse-1-0-video-edit.md) + - [happyhorse-1.0-r2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/happyhorse-1-0-r2v.md) + - [wan2.7-t2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-7-t2v.md) + - [wan2.6-t2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-6-t2v.md) + - [wan2.5-t2v-preview](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-5-t2v.md) + - [wan2.6-t2v-us](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-6-t2v-us.md) + - [wan2.2-t2v-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-2-t2v-plus.md) + - [wanx2.1-t2v-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wanx2-1-t2v-plus.md) + - [wan2.1-t2v-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-1-t2v-plus.md) + - [wanx2.1-t2v-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wanx2-1-t2v-turbo.md) + - [wan2.1-t2v-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-1-t2v-turbo.md) + - [wan2.7-i2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-7-i2v.md) + - [wan2.6-i2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-6-i2v.md) + - [wan2.6-i2v-us](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-6-i2v-us.md) + - [wan2.6-i2v-flash](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-6-i2v-flash.md) + - [wan2.2-i2v-flash](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-2-i2v-flash.md) + - [wan2.5-i2v-preview](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-5-i2v.md) + - [wan2.2-i2v-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-2-i2v-plus.md) + - [wan2.1-i2v-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-1-i2v-plus.md) + - [wanx2.1-i2v-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wanx2-1-i2v-plus.md) + - [wanx2.1-i2v-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wanx2-1-i2v-turbo.md) + - [wan2.1-i2v-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-1-i2v-turbo.md) + - [happyhorse-1.0-i2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/happyhorse-1-0-i2v.md) + - [wan2.7-r2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-7-r2v.md) + - [wan2.6-r2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-6-r2v.md) + - [wan2.6-r2v-flash](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-6-r2v-flash.md) + - [wan2.2-s2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-2-s2v.md) + - [wan2.2-s2v-detect](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-2-s2v-detect.md) + - [wan2.2-kf2v-flash](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-2-kf2v-flash.md) + - [wanx2.1-kf2v-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wanx2-1-kf2v-plus.md) + - [wan2.1-kf2v-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-1-kf2v-plus.md) + - [wan2.2-animate-mix](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-2-animate-mix.md) + - [wan2.2-animate-move](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-2-animate-move.md) + - [wanx2.1-vace-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wanx2-1-vace-plus.md) + - [wan2.7-videoedit](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-7-videoedit.md) + - [wan2.1-vace-plus](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/wan2-1-vace-plus.md) + - [animate-anyone-detect-gen2](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/animate-anyone-detect-gen2.md) + - [animate-anyone-gen2](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/animate-anyone-gen2.md) + - [animate-anyone-template-gen2](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/animate-anyone-template-gen2.md) + - [emo-detect-v1](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/emo-detect-v1.md) + - [emo-v1](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/emo-v1.md) + - [emoji-v1](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/emoji-v1.md) + - [liveportrait](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/smart-portrait-liveport.md) + - [emoji-detect-v1](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/emoji-detect-v1.md) + - [liveportrait-detect](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/liveportrait-detect.md) + - [video-style-transform](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/video-style-transform.md) + - [videoretalk](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/acoustic-portraits-videoretalk.md) + - [kling/kling-v3-video-generation](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/kling-v3-video-generation.md) + - [kling/kling-v3-omni-video-generation](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/kling-v3-omni-video-generation.md) + - [pixverse/pixverse-lipsync](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-lipsync.md) + - [pixverse/pixverse-upscale](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-upscale.md) + - [pixverse/pixverse-motioncontrol](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-motioncontrol.md) + - [pixverse/pixverse-v6-it2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-v6-it2v.md) + - [pixverse/pixverse-v6-kf2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-v6-kf2v.md) + - [pixverse/pixverse-v6-t2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-v6-t2v.md) + - [pixverse/pixverse-v6-r2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-v6-r2v.md) + - [pixverse/pixverse-v5.6-it2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-v5-6-it2v.md) + - [pixverse/pixverse-v5.6-kf2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-v5-6-kf2v.md) + - [pixverse/pixverse-v5.6-t2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-v5-6-t2v.md) + - [pixverse/pixverse-v5.6-r2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-v5-6-r2v.md) + - [pixverse/pixverse-c1-kf2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-c1-kf2v.md) + - [pixverse/pixverse-c1-it2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-c1-it2v.md) + - [vidu/viduq3-ad_reference2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-ad-reference2video.md) + - [pixverse/pixverse-c1-r2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-c1-r2v.md) + - [vidu/viduq3-pro-fast_img2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-pro-fast-img2video.md) + - [vidu/viduq3-drama_reference2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-drama-reference2video.md) + - [vidu/viduq2-pro-fast_img2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq2-pro-fast-img2video.md) + - [vidu/viduq2-pro_img2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq2-pro-img2video.md) + - [vidu/viduq2-pro_start-end2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq2-pro-start-end2video.md) + - [vidu/viduq2-pro_reference2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq2-pro-reference2video.md) + - [vidu/viduq2-turbo_img2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq2-turbo-img2video.md) + - [vidu/viduq2-turbo_start-end2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq2-turbo-start-end2video.md) + - [vidu/viduq2_reference2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq2-reference2video.md) + - [vidu/viduq2_text2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq2-text2video.md) + - [vidu/viduq3-mix_reference2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-mix-reference2video.md) + - [vidu/viduq3-pro_img2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-pro-img2video.md) + - [vidu/viduq3-pro_start-end2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-pro-start-end2video.md) + - [vidu/viduq3-pro_text2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-pro-text2video.md) + - [vidu/viduq3-turbo_img2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-turbo-img2video.md) + - [vidu/viduq3-turbo_start-end2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-turbo-start-end2video.md) + - [vidu/viduq3-turbo_reference2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-turbo-reference2video.md) + - [vidu/viduq3-turbo_text2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-turbo-text2video.md) + - [vidu/viduq3_reference2video](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/viduq3-reference2video.md) + - [pixverse/pixverse-c1-t2v](raw/model-user-guide/support/model-studio-model-list/model-list-video-generation/pixverse-c1-t2v.md) + - **3D生成** + - [Tripo/Tripo-H3.1](raw/model-user-guide/support/model-studio-model-list/model-list-3d-generation/tripo-h3-1.md) + - [Tripo/Tripo-P1.0](raw/model-user-guide/support/model-studio-model-list/model-list-3d-generation/tripo-p1-0.md) + - **语音合成** + - [qwen3-tts-flash](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-flash.md) + - [qwen3-tts-instruct-flash](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-instruct-flash.md) + - [qwen3-tts-vc-2026-01-22](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vc.md) + - [qwen3-tts-vd-2026-01-26](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vd.md) + - [qwen-tts](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/model-qwen-tts.md) + - [qwen3-tts-flash-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-flash-realtime.md) + - [qwen3-tts-instruct-flash-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-instruct-flash-realtime.md) + - [qwen3-tts-vc-realtime-2026-01-15](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vc-realtime.md) + - [qwen3-tts-vd-realtime-2026-01-15](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vd-realtime.md) + - [qwen-tts-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/model-qwen-tts-realtime.md) + - [qwen-voice-design](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen-voice-design.md) + - [cosyvoice-v3.5-flash](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-5-flash.md) + - [qwen-voice-enrollment](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen-voice-enrollment.md) + - [cosyvoice-v3.5-plus](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-5-plus.md) + - [cosyvoice-v3-flash](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-flash.md) + - [cosyvoice-v3-plus](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-plus.md) + - [cosyvoice-v2](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v2.md) + - [cosyvoice-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v1.md) + - [cosyvoice-clone-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-clone-v1.md) + - [Sambert语音合成](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/sambert.md) + - [fun-music-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/fun-music-v1.md) + - [fun-music-preview](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/music-generation-preview.md) + - [voice-enrollment](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/voice-enrollment.md) + - [MiniMax/speech-2.8-hd](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-2-8-hd.md) + - [MiniMax/speech-2.8-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-2-8-turbo.md) + - [MiniMax/speech-02-hd](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-02-hd.md) + - [MiniMax/speech-02-turbo](raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-02-turbo.md) + - **语音识别** + - [qwen3.5-livetranslate-flash-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-6.md) + - [qwen3-livetranslate-flash](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/model-qwen3-livetranslate-flash.md) + - [qwen3-livetranslate-flash-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/model-qwen3-livetranslate-flash-realtime.md) + - [qwen3-asr-flash](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash.md) + - [qwen3-asr-flash-us](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-us.md) + - [qwen3-asr-flash-filetrans](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-filetrans.md) + - [qwen3-asr-flash-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-realtime.md) + - [fun-asr](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr.md) + - [fun-asr-flash-2026-06-15](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-flash.md) + - [fun-asr-flash-8k-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-flash-8k-realtime.md) + - [fun-asr-realtime](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-realtime.md) + - [fun-asr-mtl](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-mtl.md) + - [gummy-chat-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/gummy-chat-v1.md) + - [gummy-realtime-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/gummy-realtime-v1.md) + - [paraformer-8k-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-8k-v1.md) + - [paraformer-mtl-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-mtl-v1.md) + - [paraformer-realtime-8k-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-8k-v1.md) + - [paraformer-realtime-8k-v2](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-8k-v2.md) + - [paraformer-realtime-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-v1.md) + - [paraformer-realtime-v2](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-v2.md) + - [paraformer-v1](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-v1.md) + - [paraformer-v2](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-v2.md) + - [speech-biasing](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/speech-biasing.md) + - [qwen3-omni-30b-a3b-captioner](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-omni-30b-a3b-captioner.md) + - [paraformer-8k-v2](raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-8k-v2.md) + - **向量与重排序** + - [qwen3.7-text-embedding](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-7-text-embedding.md) + - [qwen3-vl-embedding](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-vl-embedding.md) + - [qwen3-vl-rerank](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-vl-rerank.md) + - [qwen2.5-vl-embedding](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen2-5-vl-embedding.md) + - [qwen3-rerank](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-rerank.md) + - [gte-rerank-v2](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/gte-rerank-v2.md) + - [text-embedding-v4](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v4.md) + - [text-embedding-v3](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v3.md) + - [text-embedding-v2](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v2.md) + - [text-embedding-v1](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v1.md) + - [text-embedding-async-v2](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-async-v2.md) + - [text-embedding-async-v1](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-async-v1.md) + - [multimodal-embedding-v1](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/multimodal-embedding-v1.md) + - [tongyi-embedding-vision-flash](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/tongyi-embedding-vision-flash.md) + - [tongyi-embedding-vision-plus](raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/tongyi-embedding-vision-plus.md) + - [常见问题](raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) + - [相关协议](raw/model-user-guide/support/related-agreements.md) + - [阿里云百炼平台售后服务范围说明](raw/model-user-guide/support/after-sales-service-scope.md) - **实践教程** - **三方模型调用教程** - - [DeepSeek-硅基流动](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek-阿里云](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) + - [DeepSeek-硅基流动](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) + - [Kimi-月之暗面](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - [DeepSeek](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md) - [Kimi](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md) - - [Kimi-月之暗面](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [GLM-智谱](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) - [GLM](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md) @@ -118,37 +472,31 @@ - [Vidu视频生成Prompt指南](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [MiMo-小米](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) - [Stepfun-阶跃星辰](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md) - - [HappyHorse 打造一站式影视创作平台](raw/model-user-guide/use-cases/infinite-canvas.md) - [高效搭建 AI 智能体与工作流应用](raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) + - [HappyHorse 打造一站式影视创作平台](raw/model-user-guide/use-cases/infinite-canvas.md) - [深度研究：生成你的独家洞察报告](raw/model-user-guide/use-cases/deep-research.md) - [AI 解题 + 批改：推动课程教学智变](raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生文Prompt指南](raw/model-user-guide/use-cases/prompt-engineering-guide.md) - [文生图Prompt指南](raw/model-user-guide/use-cases/text-to-image-prompt.md) - - [基于LlamaIndex构建RAG应用](raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [文生视频/图生视频Prompt指南](raw/model-user-guide/use-cases/text-to-video-prompt.md) + - [基于LlamaIndex构建RAG应用](raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [自定义模型调优、部署与评测](raw/model-user-guide/use-cases/model-training-best-practices.md) - [借助大模型将文档转换为视频](raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md) - [限流应对最佳实践 ](raw/model-user-guide/use-cases/rate-limiting-best-practices.md) - [显式缓存最佳实践](raw/model-user-guide/use-cases/explicit-cache-guide.md) -- **服务支持** - - [常见问题](raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) - - [相关协议](raw/model-user-guide/support/related-agreements.md) - **产品动态** - - [模型平台功能更新](raw/model-user-guide/release-notes/model-release-notes.md) - [模型上下架与更新](raw/model-user-guide/release-notes/newly-released-models.md) + - [模型平台功能更新](raw/model-user-guide/release-notes/model-release-notes.md) ## 应用使用指南 -- **开始使用** - - [0代码构建私有知识问答应用](raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) - - [应用功能动态](raw/application-user-guide/start-using/application-release-notes.md) - **应用开发** - [应用类型介绍](raw/application-user-guide/llm-application/application-introduction.md) - [智能体应用](raw/application-user-guide/llm-application/single-agent-application.md) - - [新版智能体应用（Agent 2.0）](raw/application-user-guide/llm-application/new-single-agent-application.md) + - [新版智能体应用](raw/application-user-guide/llm-application/new-single-agent-application.md) + - [工作流应用](raw/application-user-guide/llm-application/workflow-application.md) - [高代码应用](raw/application-user-guide/llm-application/rich-code-application.md) - [文件问答](raw/application-user-guide/llm-application/file-q-a.md) - - [工作流应用](raw/application-user-guide/llm-application/workflow-application.md) - **Managed Agents** - [概述](raw/application-user-guide/managed-agents/managed-agents-introduction.md) - [快速开始](raw/application-user-guide/managed-agents/managed-agents-quick-start.md) @@ -156,64 +504,66 @@ - [配置 Agent 环境](raw/application-user-guide/managed-agents/managed-agents-environment.md) - [委派任务给 Agent](raw/application-user-guide/managed-agents/managed-agents-session.md) - [Agent 上下文管理](raw/application-user-guide/managed-agents/managed-agents-context.md) +- **开始使用** + - [0代码构建私有知识问答应用](raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) + - [应用功能动态](raw/application-user-guide/start-using/application-release-notes.md) +- **记忆库** + - [记忆库](raw/application-user-guide/memory-library-overview/memory-library.md) + - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) + - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) - **Prompt** - [Prompt模板概述](raw/application-user-guide/prompt/prompt-template.md) + - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) - [Prompt自动优化](raw/application-user-guide/prompt/optimize-prompt.md) - [使用Prompt样例库优化模型输出](raw/application-user-guide/prompt/prompt-sample-optimization.md) - [基于大模型输入输出样例的Prompt自动优化](raw/application-user-guide/prompt/prompt-feedback-optimization.md) - - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) -- **记忆库** - - [记忆库](raw/application-user-guide/memory-library-overview/memory-library.md) - - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) +- **Skill** + - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) +- **数据连接** + - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) - **知识库（RAG）** - [知识库](raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - - [RAG效果优化](raw/application-user-guide/knowledge-base/rag-optimization.md) - [知识库日志与监控](raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) + - [RAG效果优化](raw/application-user-guide/knowledge-base/rag-optimization.md) + - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) - [知识库配额与限制](raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识库计费说明](raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) + - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识问答](raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) -- **数据连接** - - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) -- **Skill** - - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) -- **插件** - - [插件概述](raw/application-user-guide/plug-in/plug-in-overview.md) - - [官方和第三方插件](raw/application-user-guide/plug-in/plugins.md) - - [自定义插件](raw/application-user-guide/plug-in/custom-plug-ins.md) - **MCP** - [模型上下文协议（MCP）](raw/application-user-guide/model-context-protocol/mcp-introduction.md) - [自定义 MCP 服务](raw/application-user-guide/model-context-protocol/custom-mcp.md) - - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) - - [外部调用](raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - [MCP 常见问题](raw/application-user-guide/model-context-protocol/mcp-faq.md) + - [外部调用](raw/application-user-guide/model-context-protocol/mcp-external-calls.md) + - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) +- **插件** + - [插件概述](raw/application-user-guide/plug-in/plug-in-overview.md) + - [官方和第三方插件](raw/application-user-guide/plug-in/plugins.md) - **应用发布与分享** - - [使用智能体或工作流作为组件](raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) - [分享智能体应用](raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) + - [使用智能体或工作流作为组件](raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) - [UI设计器](raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) - **应用调用** - - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) - - [调用工作流应用](raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) - [调用智能体应用](raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) + - [调用工作流应用](raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) + - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) - **应用评测** - **新版应用评测** - [新版评测集](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md) - - [评测任务](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md) - [标签管理](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md) + - [评测任务](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md) - [评估器](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md) + - [自动评测](raw/application-user-guide/application-evaluation/application-auto-evaluation.md) - [手动评测](raw/application-user-guide/application-evaluation/evaluate-manual-application.md) - [评测集](raw/application-user-guide/application-evaluation/application-evaluation-dataset.md) - - [自动评测](raw/application-user-guide/application-evaluation/application-auto-evaluation.md) - **应用广场** - **官方应用-通义拍照解题辅导** - **API参考** - **API目录** - [CutQuestions - 试卷切题](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-dir/api-edututor-2025-07-07-cutquestions.md) - [AnswerSSE - 解题辅导](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-dir/api-edututor-2025-07-07-answersse.md) - - [API概览](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-overview.md) - [服务接入点](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-endpoint.md) + - [API概览](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-overview.md) - [授权信息](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-ram.md) - [通义拍照解题辅导产品介绍](raw/application-user-guide/application-gallery/edu-tutor/brief-introduction-of-edu-tutor.md) - **官方应用-通义音频播客生成** @@ -228,24 +578,24 @@ - **官方应用-多模态交互开发套件** - **使用指南** - [应用创建](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-creation.md) - - [应用配置](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-configuration.md) - [应用体验与发布](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-experience-and-publishing.md) + - [应用配置](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-configuration.md) - [百炼应用推荐模板](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/agent-template.md) - [指令列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/instruction-list.md) - - [对话日志接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-chatlog.md) - [多语言对话](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multi-language-dialogue.md) - - [音色列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md) + - [对话日志接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-chatlog.md) - [三方Agent接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-integration-a2a.md) + - [音色列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md) - **SDK安装** - [服务端Java SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-java.md) - [服务端Python SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-python.md) - [服务端 Go SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/server-go-sdk.md) - - [移动端iOS SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios.md) - [移动端Android SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android.md) + - [移动端iOS SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios.md) - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) - [移动端iOS Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios-lite.md) - - [Linux C++ SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-linux.md) - [RTOS C SDK（License模式）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/mmi-rtos-sdk.md) + - [Linux C++ SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-linux.md) - **API参考** - [实时多模态交互协议（WebSocket）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md) - [HTTP协议](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-http-protocol.md) @@ -254,13 +604,16 @@ - [调用插件](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/call-plugins.md) - [管理热词](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/management-hot-words.md) - [多模态对话结果 extra_info 说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/extra-info-description.md) - - [多模态交互套件-错误码](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md) - [长期记忆开放接口](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/long-term-memory-api.md) + - [多模态交互套件-错误码](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md) - **最佳实践** - **接入百炼及三方Agent** - [百炼及三方Agent直连调用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/agent-direct-call.md) - [接入百炼智能体应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-app.md) - [接入百炼工作流应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-workflow.md) + - **接入拍照问答Agent** + - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) + - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) - **接入图像生成Agent** - [通过HTTP协议接入图像生成Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/image-agent.md) - [语音请求直通图像生成Agent(websocket)](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/audio-to-generateimgagent.md) @@ -268,87 +621,40 @@ - [录音纪要Agent使用教程](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/recording-summary-agent-tutorial.md) - [快速集成智能纪要Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/fast-integrate-offline-tingwu-meeting-agent.md) - [实时转写能力集成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/realtime-tingwu-meeting-agent-integration.md) - - **接入拍照问答Agent** - - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) - - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) - [接入多模态备忘录Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/multimodal-memo-agent.md) - [接入音乐电台Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/music-agent.md) - [接入视频通话Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/live-api-integration.md) - [动作情绪控制实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/action-emotion-control-practice.md) - - [自定义指令实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-directive.md) - [自定义对话角色实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-role.md) - - [声音复刻及声音设计实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/voice-cloning-and-voice-design.md) + - [自定义指令实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-directive.md) - [音频采集和播放说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/audio-capture-and-playback-instructions.md) - [基于RTOS SDK (License模式) 实现聊天能力](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/chat-capability-based-on-rtos-sdk.md) + - [声音复刻及声音设计实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/voice-cloning-and-voice-design.md) - [产品概述](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-overview.md) - [产品计费](raw/application-user-guide/application-gallery/multimodal-products/product-billing.md) - [多模态交互开发套件常见问题](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-faq.md) - - **官方应用-伶鹊CCAI-对话分析AIO** - - **使用指南** - - [如何开通伶鹊CCAI-对话分析AIO](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md) - - [如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/application-management.md) - - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) - - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) - - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) - - [知识库的使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/using-the-knowledge-base.md) - - **API参考** - - **API目录** - - **热词管理** - - [CreateVocab - 创建热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md) - - [UpdateVocab - 修改热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md) - - [ListVocab - 获取热词列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md) - - [DeleteVocab - 删除热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md) - - [GetVocab - 获取热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md) - - **不推荐或白名单开放** - - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) - - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) - - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) - - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) - - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) - - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) - - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) - - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) - - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) - - **最佳实践** - - [客服服务质检最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md) - - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) - - [摘要生成（含摘要/标题/关键词）最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md) - - **接口调用示例** - - [通过原生Prompt调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md) - - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) - - [通过任务类型调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md) - - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) - - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) - - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) - - [伶鹊CCAI-对话分析RAM子账号使用方式和授权操作](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md) - - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) - - [产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/product-overview-1.md) - - [伶鹊CCAI-对话分析AIO产品计费](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/billing-description-magpie-ccai-dialogue-analysis-aio.md) - - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md) - **官方应用-全妙轻应用系列** - **计费说明（全妙轻应用）** - - [电商零售推广文案写作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-retail-promotion-copywriting-billing.md) - [电商文案智能可控生成计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-copy-intelligent-controllable-generation-billing.md) - - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) - [影视互娱剧本创作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-mutual-entertainment-script-creation-billing.md) - - [泛企业VOC挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-voc-mining-billing.md) + - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) - [车机网络热点信息互动问答计费文档](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/car-machine-network-hot-information-interactive-question-and-answer-billing.md) + - [泛企业VOC挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-voc-mining-billing.md) - [泛企业线索挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-lead-mining-billing.md) - [网络内容安全审核计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/network-content-security-audit-billing.md) - [作文批改计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/composition-correction-billing.md) - [视频智能拆条计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/video-smart-strip-billing.md) + - [电商零售推广文案写作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-retail-promotion-copywriting-billing.md) - **使用指南** - [电商文案智能可控生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/intelligent-and-controllable-generation-of-e-commerce-copywriting.md) - - [传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-retail-article-style-and-format-learning.md) - - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - [影视互娱剧本创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/film-and-television-script-creation.md) + - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - [车机网络热点信息互动问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/car-machine-content-platform-news-hot-list-interaction.md) - - [泛企业VOC挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-voc-mining.md) - [泛企业线索挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-clue-mining.md) - - [作文批改助手](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/composition-correction-assistant.md) - [网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/network-content-security-audit.md) + - [作文批改助手](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/composition-correction-assistant.md) + - [传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-retail-article-style-and-format-learning.md) + - [泛企业VOC挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-voc-mining.md) - **开发文档** - **最佳实践** - [应用视频理解和一键成片的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-applying-video-understanding-and-one-click-film.md) @@ -359,35 +665,35 @@ - [ModelUsage](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-struct-dir/api-quanmiaolightapp-2024-08-01-struct-modelusage.md) - **API目录** - **电商零售推广文案写作** - - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) - [RunMarketingInformationExtract - 电商零售内容实体抽取](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationextract.md) + - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) - **传媒/零售文章风格与格式学习** - [RunStyleWriting - 传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-media-retail-article-style-and-format-learning/api-quanmiaolightapp-2024-08-01-runstylewriting.md) - **影视互娱剧本创作** - [RunScriptRefine - 影视互娱剧本创作-剧本整理](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptrefine.md) - - [RunScriptPlanning - 影视互娱剧本创作-剧本策划](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptplanning.md) - [RunScriptChat - 影视互娱剧本创作-交互式创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptchat.md) + - [RunScriptPlanning - 影视互娱剧本创作-剧本策划](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptplanning.md) - [RunScriptContinue - 影视互娱剧本创作-剧本续写](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptcontinue.md) - **影视传媒视频理解** - - [SubmitVideoAnalysisTask - 视频理解-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-submitvideoanalysistask.md) - [GetVideoAnalysisTask - 视频理解-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md) + - [SubmitVideoAnalysisTask - 视频理解-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-submitvideoanalysistask.md) - [UpdateVideoAnalysisConfig - 视频理解-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysisconfig.md) - [GetVideoAnalysisConfig - 视频理解-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysisconfig.md) - [RunVideoAnalysis - 视频理解-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-runvideoanalysis.md) - [UpdateVideoAnalysisTask - 视频理解-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistask.md) - [UpdateVideoAnalysisTasks - 视频理解-批量取消任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistasks.md) - **影视传媒智能拆条** - - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) - [SubmitVideoDetectShotTask - 智能拆条-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-submitvideodetectshottask.md) + - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) - [UpdateVideoDetectShotConfig - 智能拆条-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshotconfig.md) - [GetVideoDetectShotConfig - 智能拆条-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshotconfig.md) - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) + - **泛企业VOC挖掘** + - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) - **车机网络热点信息互动问答** - [RunHotTopicChat - 播报单（热榜）问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicchat.md) - [RunHotTopicSummary - 播报单热点自定义摘要生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicsummary.md) - - **泛企业VOC挖掘** - - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) - **泛企业线索挖掘** - [GenerateOutputFormat - 获取输出格式示例](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-generateoutputformat.md) - [RunTagMiningAnalysis - 标签挖掘分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-runtagmininganalysis.md) @@ -400,137 +706,194 @@ - [GetEssayCorrectionTask - 获取作文批改任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-getessaycorrectiontask.md) - **其他** - [GenerateBroadcastNews - 播报单（热榜）热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-generatebroadcastnews.md) - - [GetTagMiningAnalysisTask - 获取标签挖掘分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettagmininganalysistask.md) - [ListHotTopicSummaries - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listhottopicsummaries.md) - [SubmitTagMiningAnalysisTask - 提交标签挖掘分析任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submittagmininganalysistask.md) - - [HotNewsRecommend - 新闻热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-hotnewsrecommend.md) + - [GetTagMiningAnalysisTask - 获取标签挖掘分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettagmininganalysistask.md) - [GetFileContent - 获取文件内容](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getfilecontent.md) + - [HotNewsRecommend - 新闻热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-hotnewsrecommend.md) - [BatchQueryTaskStatus - 批量查询异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchquerytaskstatus.md) - - [CancelAsyncTask - 根据任务ID取消异步任务的执行](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-cancelasynctask.md) - [BatchCancelTasks - 批量取消异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchcanceltasks.md) - - [ExportAnalysisTagDetailByTaskId - 根据任务ID导出分析明细](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-exportanalysistagdetailbytaskid.md) + - [CancelAsyncTask - 根据任务ID取消异步任务的执行](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-cancelasynctask.md) - [GetEnterpriseVocAnalysisTask - 获取企业VOC分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getenterprisevocanalysistask.md) - - [GetTaskExecutionStatistics - 查询任务执行情况统计](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettaskexecutionstatistics.md) + - [ExportAnalysisTagDetailByTaskId - 根据任务ID导出分析明细](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-exportanalysistagdetailbytaskid.md) - [ListAnalysisTagDetailByTaskId - 获取挖掘结果明细列表](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listanalysistagdetailbytaskid.md) - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC挖掘异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submitenterprisevocanalysistask.md) - - [API概览](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-overview.md) + - [GetTaskExecutionStatistics - 查询任务执行情况统计](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettaskexecutionstatistics.md) - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-endpoint.md) + - [API概览](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-overview.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-ram.md) - [版本说明](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-changeset.md) - [全妙轻应用更新公告](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-update-announcement.md) - [常见问题](raw/application-user-guide/application-gallery/quanmiao-light-application-series/quanmiao-lightapp-faq.md) - - **官方应用-伶鹊CCAI-客服对话Agent** + - **官方应用-伶鹊CCAI-对话分析AIO** + - **使用指南** + - [如何开通伶鹊CCAI-对话分析AIO](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md) + - [如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/application-management.md) + - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) + - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) + - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) + - [知识库的使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/using-the-knowledge-base.md) + - **最佳实践** + - [客服服务质检最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md) + - [摘要生成（含摘要/标题/关键词）最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md) + - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) - **API参考** - **API目录** - - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) - - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) - - [计费说明（客服对话Agent）](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/billing-description-beebot-agent.md) - - [使用指南](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/guidelines-for-use.md) + - **热词管理** + - [CreateVocab - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md) + - [UpdateVocab - 修改热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md) + - [ListVocab - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md) + - [DeleteVocab - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md) + - [GetVocab - 获取热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md) + - **不推荐或白名单开放** + - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) + - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) + - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) + - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) + - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) + - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) + - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) + - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) + - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) + - **接口调用示例** + - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) + - [通过原生Prompt调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md) + - [通过任务类型调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md) + - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) + - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) + - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) + - [伶鹊CCAI-对话分析RAM子账号使用方式和授权操作](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-and-authorize-ram-users-for-ccai-dialogue-analysis.md) + - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) + - [产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/product-overview-1.md) + - [伶鹊CCAI-对话分析AIO产品计费](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/billing-description-magpie-ccai-dialogue-analysis-aio.md) + - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md) - **官方应用-伶鹊CCAI-语音对话机器人** - **API参考** - **API目录** - **MQ消息订阅配置** - - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) - - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) - - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) + - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) + - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) + - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) - **变量管理** - - [DeleteVariable - 删除变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md) - - [ListVariable - 获取变量列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md) - - [UpdateVariable - 更新变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md) - - [CreateVariable - 创建变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md) + - [DeleteVariable - 删除变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md) + - [ListVariable - 获取变量列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md) + - [CreateVariable - 创建变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md) + - [UpdateVariable - 更新变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md) - **三方语音配置** - - [UpdateVoiceAccessProfile - 更新三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md) - - [ListVoiceEngines - 获取三方语音引擎列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md) - - [ListVoiceAccessProfile - 获取三方语音配置列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md) - - [DeleteVoiceAccessProfile - 删除三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md) - - [CreateVoiceAccessProfile - 创建三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md) + - [ListVoiceEngines - 获取三方语音引擎列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md) + - [UpdateVoiceAccessProfile - 更新三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md) + - [ListVoiceAccessProfile - 获取三方语音配置列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md) + - [DeleteVoiceAccessProfile - 删除三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md) + - [CreateVoiceAccessProfile - 创建三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md) - **热词管理** - - [UpdateVocabulary - 更新热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md) - - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) - - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) - - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) - - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) - - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) - - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) + - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) + - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) + - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) + - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) + - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) + - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) + - [UpdateVocabulary - 更新热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md) - **克隆音管理** - - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) - - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) - - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) - - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) - - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) + - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) + - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) + - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) + - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) + - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) - **应用管理** - - [DeleteApplication - 删除语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md) - - [ListNluModels - 获取对话大模型列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md) - - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) - - [PreviewVoice - TTS合成试听](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md) - - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) - - [CreateApplicationVersion - 创建语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md) - - [ListApplications - 查询语音机器人应用列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md) - - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) - - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) - - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) - - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) - - [UpdateApplicationVersion - 修改语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md) - - [BridgeWebCall - 软电话测试通话](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md) - - [GetDataChannelCredential - 获取数据通道凭证](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md) - - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) - - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md) + - [DeleteApplication - 删除语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md) + - [ListNluModels - 获取对话大模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md) + - [PreviewVoice - TTS合成试听](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md) + - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) + - [ListApplications - 查询语音机器人应用列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md) + - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) + - [CreateApplicationVersion - 创建语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md) + - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) + - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) + - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) + - [UpdateApplicationVersion - 修改语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md) + - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) + - [BridgeWebCall - 软电话测试通话](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md) + - [GetDataChannelCredential - 获取数据通道凭证](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md) + - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) + - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md) + - [语音对话机器人产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md) - [伶鹊CCAI-语音对话机器人产品计费说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/billing-information-lingque-ccai-voice-dialogue-robot.md) - - [语音对话机器人产品概述](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md) - [语音对话机器人操作指南](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/operation-guide.md) - - **通义点金** + - **官方应用-伶鹊CCAI-客服对话Agent** - **API参考** - **API目录** - - **平台能力-文档库** - - [GetAppConfig - 获取配置信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getappconfig.md) + - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) + - [计费说明（客服对话Agent）](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/billing-description-beebot-agent.md) + - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) + - [使用指南](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/guidelines-for-use.md) + - **官方应用-通义数据挖掘** + - **API参考** + - **API目录** + - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) + - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) + - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) + - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) + - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) + - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) + - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) + - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) + - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) + - **通义点金** + - **API参考** + - **API目录** + - **平台能力-文档库** - [UpdateDocumentChunk - 更新文档块内容](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocumentchunk.md) - [CreateLibrary - 创建文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createlibrary.md) - - [GetLibraryList - 获取文档库列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrarylist.md) + - [GetAppConfig - 获取配置信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getappconfig.md) - [GetLibrary - 获取文档库详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrary.md) + - [GetLibraryList - 获取文档库列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrarylist.md) - [UploadDocument - 上传文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-uploaddocument.md) - [GetDocumentUrl - 获取文档的下载链接](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumenturl.md) - [GetFilterDocumentList - 按元信息过滤查询文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getfilterdocumentlist.md) - - [GetDocumentList - 获取文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentlist.md) - [PreviewDocument - 预览文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-previewdocument.md) + - [GetDocumentList - 获取文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentlist.md) + - [CreatePredefinedDocument - 创建预定义文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createpredefineddocument.md) - [DeleteDocument - 删除文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletedocument.md) - [UpdateDocument - 更新文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocument.md) - - [CreatePredefinedDocument - 创建预定义文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createpredefineddocument.md) - [GetDocumentChunkList - 获取文档块列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentchunklist.md) - [RecallDocument - 文档召回](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-recalldocument.md) - - [ReIndex - 重建索引](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-reindex.md) - [GetParseResult - 获取文档解析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getparseresult.md) - - [DeleteLibrary - 删除文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletelibrary.md) + - [ReIndex - 重建索引](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-reindex.md) - [UpdateLibrary - 更新文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatelibrary.md) + - [DeleteLibrary - 删除文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletelibrary.md) + - [RunLibraryChatGeneration - 文档库会话生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-runlibrarychatgeneration.md) - [GetHistoryListByBizType - 根据业务类型获取对话历史记录](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-gethistorylistbybiztype.md) - [InvokePlugin - 调用插件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-invokeplugin.md) - - [RunLibraryChatGeneration - 文档库会话生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-runlibrarychatgeneration.md) - **平台能力-应用** - [EndToEndRealTimeDialog - 语音实时对话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-endtoendrealtimedialog.md) - [RunDialogAnalysis - 会话分析结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rundialoganalysis.md) - - [RunAgent - 运行智能体](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runagent.md) - [CreateDialog - 创建外呼会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialog.md) + - [RunAgent - 运行智能体](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runagent.md) - [RealTimeDialog - 实时会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialog.md) - - [GetDialogDetail - 获取会话详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialogdetail.md) - [RealtimeDialogAssist - 实时会话辅助](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialogassist.md) + - [GetDialogDetail - 获取会话详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialogdetail.md) - [GetDialogLog - 获取对话日志](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoglog.md) - - [CreateDialogAnalysisTask - 创建会话分析任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialoganalysistask.md) - [GetDialogAnalysisResult - 获取会话分析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoganalysisresult.md) + - [CreateDialogAnalysisTask - 创建会话分析任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialoganalysistask.md) - [RebuildTask - 重建任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rebuildtask.md) - - [GetTaskStatus - 获取任务状态](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskstatus.md) - [EvictTask - 取消任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-evicttask.md) + - [GetTaskStatus - 获取任务状态](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskstatus.md) - [CreateDocsSummaryTask - 创建多文档总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdocssummarytask.md) - - [CreateAnnualDocSummaryTask - 创建按年份总结文档任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createannualdocsummarytask.md) - [CreatePdfTranslateTask - 创建pdf文档翻译任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createpdftranslatetask.md) - [CreateFinReportSummaryTask - 创建财报总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createfinreportsummarytask.md) + - [CreateAnnualDocSummaryTask - 创建按年份总结文档任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createannualdocsummarytask.md) - [GetSummaryTaskResult - 获取财报总结任务结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getsummarytaskresult.md) - - [GetTaskResult - 获取结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskresult.md) - [CreateQualityCheckTask - 创建质检任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createqualitychecktask.md) - - [GetQualityCheckTaskResult - 获取质检结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getqualitychecktaskresult.md) + - [GetTaskResult - 获取结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskresult.md) - [GenDocQaResult - 根据文档解析问答QA](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gendocqaresult.md) + - [GetQualityCheckTaskResult - 获取质检结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getqualitychecktaskresult.md) - [RecognizeIntention - 意图识别](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-recognizeintention.md) - [UpdateQaLibrary - 更新QA问答库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-updateqalibrary.md) - [SubmitChatQuestion - 提交问题列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-submitchatquestion.md) @@ -539,23 +902,10 @@ - **其他** - [DashscopeAsyncTaskFinishEvent - Dashscope异步任务完成回调事件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-other/api-dianjin-2024-06-28-dashscopeasynctaskfinishevent.md) - [服务接入点](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-endpoint.md) - - [API概览](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-overview.md) - [授权信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-ram.md) + - [API概览](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-overview.md) - [版本说明](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-changeset.md) - [产品简介](raw/application-user-guide/application-gallery/tongyi-dianjin/tongyi-dianjin-overview.md) - - **官方应用-通义数据挖掘** - - **API参考** - - **API目录** - - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) - - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) - - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) - - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) - - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) - - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) - - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) - - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) - - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) - **官方应用-通义多模态翻译** - **API参考** - **API目录** @@ -564,28 +914,36 @@ - [TextTranslate - 文本翻译接口](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-texttranslate.md) - [SubmitLongTextTranslateTask - 提交长文本翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submitlongtexttranslatetask.md) - [SubmitHtmlTranslateTask - 提交html翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submithtmltranslatetask.md) + - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) - [GetHtmlTranslateTask - 获取html翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-gethtmltranslatetask.md) - [TermEdit - 术语库编辑](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termedit.md) - [TermQuery - 术语库查询](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termquery.md) - - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) - **图片翻译** - [GetImageTranslateTask - 获取图片翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-getimagetranslatetask.md) - [SubmitImageTranslateTask - 提交图片翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-submitimagetranslatetask.md) - **文档翻译** - - [SubmitDocTranslateTask - 文档翻译任务提交](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-submitdoctranslatetask.md) - [GetDocTranslateTask - 文档翻译结果获取](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-getdoctranslatetask.md) + - [SubmitDocTranslateTask - 文档翻译任务提交](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-submitdoctranslatetask.md) - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-endpoint.md) - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-overview.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md) - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md) - [通义多模态翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/official-application-tongyi-translate-overview.md) - [网页翻译JSSDK](raw/application-user-guide/application-gallery/official-application-tongyi-translate/web-page-translation-jssdk.md) + - **官方应用-千问联网检索Agent** + - **API参考** + - [API概览](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/service-access-point.md) + - [多模态文件操作](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat-multimodal-file.md) + - [生成对话](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat.md) + - [错误码-千问联网检索Agent](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-error-code.md) + - [千问联网检索Agent产品简介](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-guide.md) - **官方应用-通义深度搜索** - **API参考** - **API目录** - [生成对话](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-chat-generate.md) - - [对话文件管理](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-session-file-management.md) - [上传文件](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-file-upload.md) + - [对话文件管理](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-session-file-management.md) - [生成报告导出](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-report-export.md) - [对接自有知识库](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/docking-self-built-database.md) - [API概览](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-overview.md) @@ -593,14 +951,6 @@ - [错误码-通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-error-code.md) - [通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/tongyi-deep-search-introduction.md) - [操作指南](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-guide.md) - - **官方应用-千问联网检索Agent** - - **API参考** - - [API概览](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/service-access-point.md) - - [生成对话](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat.md) - - [多模态文件操作](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat-multimodal-file.md) - - [千问联网检索Agent产品简介](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-guide.md) - - [错误码-千问联网检索Agent](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-error-code.md) - **通义 UI Agent** - [通义 UI Agent](raw/application-user-guide/application-gallery/ui-agent/ui-agent-api.md) - **官方应用-全妙解决方案类产品** @@ -610,14 +960,14 @@ - **功能界面** - [妙笔-分布生成创作文章](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/step-by-step-generation.md) - [直接生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/direct-generation.md) - - [搜索素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/search-materials.md) - [智能配图](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/smart-image-generation.md) - - [AI妙笔产品概述](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/product-overview-for-amb.md) + - [搜索素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/search-materials.md) - [妙笔首页概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/amb-homepage-overview.md) + - [AI妙笔产品概述](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/product-overview-for-amb.md) - [AI工具箱](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/ai-toolbox.md) - [素材库](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/material-library.md) - - [系统配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/system-configuration.md) - [文章风格和格式学习](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/style-imitation.md) + - [系统配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/system-configuration.md) - [AI妙策](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/ai-miaoce.md) - [智能审校](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/article-review.md) - [深度写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/deep-writing.md) @@ -628,8 +978,8 @@ - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/generate-titles-summaries-media-text.md) - **政务公文写作指导** - [快速写一篇政务稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/quick-gov-writing-prompt.md) - - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/generate-titles-summaries-gov-text.md) + - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) - [常见FAQ](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/faq-for-using-quanmiao-series-products.md) - **更新公告** - **功能更新** @@ -637,35 +987,27 @@ - [2025年1月24日更新-全妙解决方案类产品](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/2025-1-24-function-update-announcement-quanmiao-saas.md) - [2024年3月11更新-AI全妙系列 V2.2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-11-ai-quanmiao-v2-2.md) - [2024年3月1更新-AI全妙系列 V2.2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-01-ai-quanmiao-v2-2.md) - - [2023年12月19更新-AI妙笔V2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2023-12-19-amb-v2.md) - [2024年2月28更新-AI全妙系列 V2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-02-28-ai-quanmiao-v2.md) + - [2023年12月19更新-AI妙笔V2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2023-12-19-amb-v2.md) - [计费说明（政务公文配套工具）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/government-document-tool-billing.md) - - [计费说明（妙笔）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/miaobi-billing.md) - [计费说明（PPT生成）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/ppt-generation-billing.md) - [计费说明（妙策-自定义数据源）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/billing-document-miaoce-custom-data-source.md) - [计费说明（视频混剪）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/billing-description-video-mixing.md) + - [计费说明（妙笔）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/miaobi-billing.md) - **妙搜和妙读** - **使用指南** - [妙搜](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaodou-and-miaodu-guidelines-for-use/ai-miaosou.md) - [计费说明（妙搜和妙读）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaosou-miaodu-api-billing.md) - **开发文档** - - **最佳实践** - - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) - - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) - - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) - - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) - - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) - - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) - - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) - **API参考** - **数据结构** - [GenerateTraceability](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-generatetraceability.md) - [OutlineSearchResult](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinesearchresult.md) - [HottopicNews](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-hottopicnews.md) - [OutlineWritingArticle](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinewritingarticle.md) - - [TopicSelection](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-topicselection.md) - [WritingOutline](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingoutline.md) - [WritingStyleTemplateDefine](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatedefine.md) + - [TopicSelection](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-topicselection.md) - [WritingStyleTemplateField](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatefield.md) - **API目录** - **通用接口** @@ -673,65 +1015,59 @@ - [ListDialogues - 生成历史列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listdialogues.md) - [ListVersions - 获取版本信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listversions.md) - [GetProperties - 获取配置信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-getproperties.md) + - **通用接口-文件上传下载** + - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) + - [GenerateFileUrlByKey - 生成文件URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generatefileurlbykey.md) - **通用接口-异步任务管理** - - [SubmitAsyncTask - 提交异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-submitasynctask.md) - [CancelAsyncTask - 取消异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-cancelasynctask.md) - - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) + - [SubmitAsyncTask - 提交异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-submitasynctask.md) - [QueryAsyncTask - 查询异步任务明细](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-queryasynctask.md) - - **通用接口-文件上传下载** - - [GenerateFileUrlByKey - 生成文件URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generatefileurlbykey.md) - - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) + - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) - **妙笔-创作文章** - - [RunAiHelperWriting - AI帮写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runaihelperwriting.md) - [RunWritingV2 - 智能写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritingv2.md) - - [RunTranslateGeneration - 中英翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtranslategeneration.md) + - [RunAiHelperWriting - AI帮写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runaihelperwriting.md) - [RunWriting - 直接写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwriting.md) - [RunStepByStepWriting - 分步骤写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runstepbystepwriting.md) + - [RunTranslateGeneration - 中英翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtranslategeneration.md) - [RunTextPolishing - 润色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtextpolishing.md) - - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) - [RunContinueContent - 内容续写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runcontinuecontent.md) + - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) - [RunWriteToneGeneration - 文风改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritetonegeneration.md) - [RunTitleGeneration - 标题生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtitlegeneration.md) - - [RunExpandContent - 内容扩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runexpandcontent.md) - [RunSummaryGenerate - 摘要生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runsummarygenerate.md) - - [SearchNews - 信息检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-searchnews.md) + - [RunExpandContent - 内容扩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runexpandcontent.md) - [RunAbbreviationContent - 内容缩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runabbreviationcontent.md) - - [RunQuickWriting - 快速写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runquickwriting.md) + - [SearchNews - 信息检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-searchnews.md) - [GenerateImageTask - 生成智能配图任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-generateimagetask.md) - [ListBuildConfigs - 获取系统自定义预设](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-listbuildconfigs.md) - - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) + - [RunQuickWriting - 快速写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runquickwriting.md) - [FeedbackDialogue - 反馈对话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-feedbackdialogue.md) - - **通用接口-通用配置** - - [CreateGeneralConfig - 通用配置-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-creategeneralconfig.md) - - [ListGeneralConfigs - 通用配置-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-listgeneralconfigs.md) - - [GetGeneralConfig - 通用配置-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-getgeneralconfig.md) - - [DeleteGeneralConfig - 通用配置-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-deletegeneralconfig.md) - - [UpdateGeneralConfig - 通用配置-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-updategeneralconfig.md) + - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) - **妙笔-文体仿写** - - [ListStyleLearningResult - 获取文体学习分析结果列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-liststylelearningresult.md) - [RunStyleFeatureAnalysis - 内容特点分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-runstylefeatureanalysis.md) + - [ListStyleLearningResult - 获取文体学习分析结果列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-liststylelearningresult.md) + - [SaveStyleLearningResult - 保存文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-savestylelearningresult.md) - [GetStyleLearningResult - 获取文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-getstylelearningresult.md) - [DeleteStyleLearningResult - 删除自定义文体](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-deletestylelearningresult.md) - - [SaveStyleLearningResult - 保存文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-savestylelearningresult.md) - [ListWritingStyles - 获取写作文体列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-listwritingstyles.md) - - **妙笔-视频审校** - - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) - - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) - **妙笔-文章审校-规则库管理** - [SubmitAuditNote - 提交自定义规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-submitauditnote.md) - - [ConfirmAndPostProcessAuditNote - 确认提交规则库用于审核](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-confirmandpostprocessauditnote.md) - [DownloadAuditNote - 下载规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-downloadauditnote.md) + - [ConfirmAndPostProcessAuditNote - 确认提交规则库用于审核](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-confirmandpostprocessauditnote.md) - [DeleteAuditNote - 删除规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-deleteauditnote.md) - [GetAuditNotePostProcessingStatus - 获取规则库后处理进度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnotepostprocessingstatus.md) - - [GetAuditNoteProcessingStatus - 查询规则库上传状态](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnoteprocessingstatus.md) - [GetAvailableAuditNotes - 查询可用规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getavailableauditnotes.md) + - [GetAuditNoteProcessingStatus - 查询规则库上传状态](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnoteprocessingstatus.md) + - **妙笔-视频审校** + - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) + - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) - **妙笔-文章审校-词库管理** - - [ListAuditTerms - 获取自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-listauditterms.md) - [AddAuditTerms - 添加自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-addauditterms.md) + - [ListAuditTerms - 获取自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-listauditterms.md) - [EditAuditTerms - 编辑自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-editauditterms.md) + - [FetchImportTermsTask - 获取导入词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchimporttermstask.md) - [DeleteAuditTerms - 删除指定词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-deleteauditterms.md) - [SubmitImportTermsTask - 提交导入词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitimporttermstask.md) - - [FetchImportTermsTask - 获取导入词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchimporttermstask.md) - [SubmitExportTermsTask - 提交导出词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitexporttermstask.md) - [FetchExportTermsTask - 获取导出词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchexporttermstask.md) - **妙笔-文章审校-事实性审核** @@ -739,35 +1075,32 @@ - [GetFactAuditUrl - 获取事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-getfactauditurl.md) - [DeleteFactAuditUrl - 删除事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-deletefactauditurl.md) - **妙笔-文章审校** - - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md) - [SubmitSmartAudit - 提交智能审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitsmartaudit.md) - - [SubmitAuditTask - 提交审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitaudittask.md) - - [CancelAuditTask - 取消审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-cancelaudittask.md) - [GetSmartAuditResult - 查询智能审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-getsmartauditresult.md) - - [ListAuditContentErrorTypes - 获取审校维度列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-listauditcontenterrortypes.md) - [ExportAuditContentResult - 导出智能审校报告](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-exportauditcontentresult.md) + - [ListAuditContentErrorTypes - 获取审校维度列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-listauditcontenterrortypes.md) + - **妙笔-文档管理** + - [GenerateExportWordTask - 生成导出文档任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-generateexportwordtask.md) + - [CreateGeneratedContent - 保存文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-creategeneratedcontent.md) + - [FetchExportWordTask - 获取导出文档任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-fetchexportwordtask.md) + - [GetGeneratedContent - 获取文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-getgeneratedcontent.md) + - [DeleteGeneratedContent - 删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-deletegeneratedcontent.md) + - [UpdateGeneratedContent - 更新文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-updategeneratedcontent.md) + - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) + - [ExportGeneratedContent - 导出文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-exportgeneratedcontent.md) - **妙笔-素材库** - [SaveMaterialDocument - 保存素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-savematerialdocument.md) - - [DeleteMaterialById - 删除素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-deletematerialbyid.md) - [UpdateMaterialDocument - 更新素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-updatematerialdocument.md) + - [DeleteMaterialById - 删除素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-deletematerialbyid.md) - [GetMaterialById - 获取素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-getmaterialbyid.md) - [ListMaterialDocuments - 获取素材列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-listmaterialdocuments.md) - **妙笔-素材库-自定义文本** - [GetCustomText - 获取自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-getcustomtext.md) - [UpdateCustomText - 更新自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-updatecustomtext.md) - [ListCustomText - 获取自定义文本列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-listcustomtext.md) + - [DeleteCustomText - 删除自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-deletecustomtext.md) - [SaveCustomText - 保存自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-savecustomtext.md) - [DocumentExtraction - 文档提取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-documentextraction.md) - - [DeleteCustomText - 删除自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-deletecustomtext.md) - - **妙笔-文档管理** - - [GenerateExportWordTask - 生成导出文档任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-generateexportwordtask.md) - - [FetchExportWordTask - 获取导出文档任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-fetchexportwordtask.md) - - [CreateGeneratedContent - 保存文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-creategeneratedcontent.md) - - [DeleteGeneratedContent - 删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-deletegeneratedcontent.md) - - [UpdateGeneratedContent - 更新文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-updategeneratedcontent.md) - - [GetGeneratedContent - 获取文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-getgeneratedcontent.md) - - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) - - [ExportGeneratedContent - 导出文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-exportgeneratedcontent.md) - **妙笔-视频混剪** - [GetClipsBuildInResource - 获取智能混剪内置资源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getclipsbuildinresource.md) - [AsyncCreateClipsTimeLine - 创建剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstimeline.md) @@ -776,49 +1109,49 @@ - [GetAutoClipsTaskInfo - 获得剪辑任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getautoclipstaskinfo.md) - [AsyncCreateClipsTask - 创建剪辑任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstask.md) - [ListAutoClipsTask - 智能混剪任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-listautoclipstask.md) - - **公文库检索** - - [ListDocumentRetrieve - 公文库检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-public-library-retrieval/api-aimiaobi-2023-08-01-listdocumentretrieve.md) + - **妙策-自定义数据源** + - [ExportCustomSourceAnalysisTask - 导出自定义源-话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-exportcustomsourceanalysistask.md) + - [GetCustomSourceTopicAnalysisTask - 获取自定义源话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-getcustomsourcetopicanalysistask.md) + - [SubmitCustomSourceTopicAnalysis - 提交自定义源话题选题分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-submitcustomsourcetopicanalysis.md) + - **通用接口-通用配置** + - [CreateGeneralConfig - 通用配置-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-creategeneralconfig.md) + - [ListGeneralConfigs - 通用配置-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-listgeneralconfigs.md) + - [GetGeneralConfig - 通用配置-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-getgeneralconfig.md) + - [UpdateGeneralConfig - 通用配置-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-updategeneralconfig.md) + - [DeleteGeneralConfig - 通用配置-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-deletegeneralconfig.md) - **妙策-选题热点** - [RunTopicSelectionMerge - 选题热点融合](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-runtopicselectionmerge.md) + - [ListHotTopics - 获取热点话题列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhottopics.md) - [ListHotNewsWithType - 获取选题热点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotnewswithtype.md) - [ListHotSources - 获取三方热榜源列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotsources.md) - - [ListHotTopics - 获取热点话题列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhottopics.md) - - [ListHotViewPoints - 获取热门视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotviewpoints.md) - [GetTopicById - 获取热点对象](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-gettopicbyid.md) + - [ListHotViewPoints - 获取热门视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotviewpoints.md) - [ListTimedViewAttitude - 获取时效性视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listtimedviewattitude.md) - [ListWebReviewPoints - 获取网友视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listwebreviewpoints.md) - - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) - [ListPlanningProposal - 获取选题策划列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listplanningproposal.md) - [ExportHotTopicPlanningProposals - 导出选题策划文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-exporthottopicplanningproposals.md) + - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) + - **公文库检索** + - [ListDocumentRetrieve - 公文库检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-public-library-retrieval/api-aimiaobi-2023-08-01-listdocumentretrieve.md) - **妙策-自定义话题** - [DeleteCustomTopicByTopic - 删除自定义热点事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicbytopic.md) - - [ListTopicRecommendEventList - 获取热点推荐事件列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicrecommendeventlist.md) - [ListTopicViewPointRecommendEventList - 获取热点事件推荐观点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicviewpointrecommendeventlist.md) + - [ListTopicRecommendEventList - 获取热点推荐事件列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicrecommendeventlist.md) - [RunCustomHotTopicAnalysis - 自定义热点话题分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicanalysis.md) - [RunCustomHotTopicViewPointAnalysis - 自定义选题视角分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicviewpointanalysis.md) - [ListCustomViewPoints - 获取自定义视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listcustomviewpoints.md) - [DeleteCustomTopicViewPointById - 删除自定义选题视角](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicviewpointbyid.md) - - **妙策-自定义数据源** - - [GetCustomSourceTopicAnalysisTask - 获取自定义源话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-getcustomsourcetopicanalysistask.md) - - [SubmitCustomSourceTopicAnalysis - 提交自定义源话题选题分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-submitcustomsourcetopicanalysis.md) - - [ExportCustomSourceAnalysisTask - 导出自定义源-话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-exportcustomsourceanalysistask.md) - **妙策-openapi** - [SubmitDocClusterTask - 提交内容聚合任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitdocclustertask.md) - - [GetDocClusterTask - 获取内容聚合任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getdocclustertask.md) - [SubmitTopicSelectionPerspectiveAnalysisTask - 提交选题热点分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submittopicselectionperspectiveanalysistask.md) - - [GetTopicSelectionPerspectiveAnalysisTask - 获取选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-gettopicselectionperspectiveanalysistask.md) + - [GetDocClusterTask - 获取内容聚合任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getdocclustertask.md) - [SubmitCustomTopicSelectionPerspectiveAnalysisTask - 提交自定义热点选题视角分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitcustomtopicselectionperspectiveanalysistask.md) + - [GetTopicSelectionPerspectiveAnalysisTask - 获取选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-gettopicselectionperspectiveanalysistask.md) - [GetCustomTopicSelectionPerspectiveAnalysisTask - 获取自定义选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getcustomtopicselectionperspectiveanalysistask.md) - **妙策-新闻播报** - [SubmitCustomHotTopicBroadcastJob - 提交自定义播报单任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-submitcustomhottopicbroadcastjob.md) - [GetHotTopicBroadcast - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-gethottopicbroadcast.md) - [GetCustomHotTopicBroadcastJob - 获取自定义播报单任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-getcustomhottopicbroadcastjob.md) - - **妙搜-智能搜索** - - [ListSearchTasks - 查询妙搜搜索生成历史任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtasks.md) - - [ListSearchTaskDialogues - 查询妙搜搜索生成任务详情列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialogues.md) - - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) - - [RunSearchGeneration - 妙搜-智能搜索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchgeneration.md) - - [RunSearchSimilarArticles - 妙搜-文搜文](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchsimilararticles.md) - **妙策-企业VOC挖掘** - [ExportAnalysisTagDetailByTaskId - 导出标签挖掘结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-exportanalysistagdetailbytaskid.md) - [ValidateUploadTemplate - 校验VOC上传模板](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-validateuploadtemplate.md) @@ -828,88 +1161,86 @@ - [GetCategoriesByTaskId - 根据任务ID获取分类列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getcategoriesbytaskid.md) - **妙搜-数据源** - [CreateDataset - 数据源-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-createdataset.md) - - [UpdateDataset - 数据源-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md) - [GetDataset - 数据源-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md) - - [DeleteDataset - 数据源-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md) + - [UpdateDataset - 数据源-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md) - [ListDatasets - 数据源-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasets.md) - - [AddDatasetDocument - 数据源-添加文档到数据集](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-adddatasetdocument.md) + - [DeleteDataset - 数据源-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md) - [GetDatasetDocument - 数据源-获取文档详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdatasetdocument.md) - - [UpdateDatasetDocument - 数据源-修改文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedatasetdocument.md) + - [AddDatasetDocument - 数据源-添加文档到数据集](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-adddatasetdocument.md) - [ListDatasetDocuments - 数据源-文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasetdocuments.md) + - [UpdateDatasetDocument - 数据源-修改文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedatasetdocument.md) - [SearchDatasetDocuments - 数据源-搜索文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-searchdatasetdocuments.md) - [DeleteDatasetDocument - 数据源-删除数据集文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedatasetdocument.md) - **系统配置-干预配置** - [ListInterveneCnt - 获得所有干预项的数量](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenecnt.md) - [ListIntervenes - 列出干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenes.md) - - [InsertInterveneGlobalReply - 插入干预全局回复项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertinterveneglobalreply.md) - [ImportInterveneFile - 同步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefile.md) + - [InsertInterveneGlobalReply - 插入干预全局回复项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertinterveneglobalreply.md) - [GetInterveneTemplateFileUrl - 获得干预导入模版文件地址](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getintervenetemplatefileurl.md) - [ImportInterveneFileAsync - 异步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefileasync.md) + - [GetInterveneGlobalReply - 获得干预全局回复内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneglobalreply.md) - [ClearIntervenes - 清除所有干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-clearintervenes.md) - [ListInterveneRules - 列出干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenerules.md) - - [GetInterveneGlobalReply - 获得干预全局回复内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneglobalreply.md) + - [ListInterveneImportTasks - 列出干预项导入任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listinterveneimporttasks.md) - [GetInterveneRuleDetail - 获得干预规则的详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneruledetail.md) - [InsertInterveneRule - 插入干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertintervenerule.md) - - [ListInterveneImportTasks - 列出干预项导入任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listinterveneimporttasks.md) - [DeleteInterveneRule - 删除干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-deleteintervenerule.md) - [ExportIntervenes - 导出干预项内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-exportintervenes.md) - [GetInterveneImportTaskInfo - 获得干预项目导入任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneimporttaskinfo.md) - - **妙读-抽取类** - - [RunHotword - 抽取关键词](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-extraction-class/api-aimiaobi-2023-08-01-runhotword.md) + - **妙搜-智能搜索** + - [RunSearchGeneration - 妙搜-智能搜索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchgeneration.md) + - [ListSearchTasks - 查询妙搜搜索生成历史任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtasks.md) + - [ListSearchTaskDialogues - 查询妙搜搜索生成任务详情列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialogues.md) + - [RunSearchSimilarArticles - 妙搜-文搜文](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchsimilararticles.md) + - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) - **系统配置-信源管理** - - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) - [SaveDataSourceOrderConfig - 保存信源权重配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-savedatasourceorderconfig.md) + - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) - **妙读-基础操作类** - [GetDocInfo - 获取文档信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getdocinfo.md) - [GetFileContentLength - 获取文件长度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getfilecontentlength.md) - [UploadBook - 书籍上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploadbook.md) - [UploadDoc - 文档上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploaddoc.md) - - [ListDocs - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-listdocs.md) - [DeleteDocs - 批量删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-deletedocs.md) + - [ListDocs - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-listdocs.md) + - **妙读-问答类** + - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) + - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) + - **妙读-抽取类** + - [RunHotword - 抽取关键词](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-extraction-class/api-aimiaobi-2023-08-01-runhotword.md) - **妙读-生成类** - - [RunDocIntroduction - 文档导读](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocintroduction.md) - [RunMultiDocIntroduction - 多文档聚合摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runmultidocintroduction.md) + - [RunDocIntroduction - 文档导读](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocintroduction.md) - [RunDocSummary - 文档摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocsummary.md) - [RunDocBrainmap - 全文脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocbrainmap.md) - [RunDocWashing - 改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocwashing.md) - [RunBookIntroduction - 书籍导读（抽取书籍卖点/书籍摘要）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookintroduction.md) - - [RunBookBrainmap - 书籍脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookbrainmap.md) - [RunCommentGeneration - 客户之声预测](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runcommentgeneration.md) - - **妙读-问答类** - - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) - - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) - - **妙读-其他** - - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) - - [RunDocTranslation - 文档翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundoctranslation.md) - - [RunBookSmartCard - 书籍智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-runbooksmartcard.md) + - [RunBookBrainmap - 书籍脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookbrainmap.md) - **深度写作** - - [SubmitDeepWriteTask - 提交深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-submitdeepwritetask.md) - [GetDeepWriteTask - 查询深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetask.md) - - [CancelDeepWriteTask - 取消深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-canceldeepwritetask.md) + - [SubmitDeepWriteTask - 提交深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-submitdeepwritetask.md) - [GetDeepWriteTaskResult - 查询深度写作任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetaskresult.md) - [RunDeepWriting - 查询深度写作事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-rundeepwriting.md) + - [CancelDeepWriteTask - 取消深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-canceldeepwritetask.md) + - **妙读-其他** + - [RunDocTranslation - 文档翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundoctranslation.md) + - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) + - [RunBookSmartCard - 书籍智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-runbooksmartcard.md) - **PPT生成** - [ListEnterprisePptTemplates - 查询企业专属PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listenterpriseppttemplates.md) - - [InitiatePptCreationV2 - 初始化PPT创建操作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreationv2.md) - [GetPptTemplateSelector - 查询PPT模板筛选器](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getppttemplateselector.md) + - [InitiatePptCreationV2 - 初始化PPT创建操作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreationv2.md) - [ListPptTemplates - 查询PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listppttemplates.md) - - [GetPptArtifactExportResult - 查询PPT导出任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifactexportresult.md) - [ExportPptArtifact - 导出PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-exportpptartifact.md) - [GetPptArtifact - 查询PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifact.md) - - [RunPptOutlineGeneration - 生成PPT大纲内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-runpptoutlinegeneration.md) + - [GetPptArtifactExportResult - 查询PPT导出任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifactexportresult.md) - [ListPptArtifacts - 查询PPT作品列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listpptartifacts.md) + - [RunPptOutlineGeneration - 生成PPT大纲内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-runpptoutlinegeneration.md) - [InitiatePptCreation - 初始化用来创建PPT的会话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreation.md) - [GetPptConfig - 获取PPT组件配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptconfig.md) - [BindPptArtifact - 绑定PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-bindpptartifact.md) - [DeletePptArtifact - 删除PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-deletepptartifact.md) - - **标书生成** - - [AsyncUploadTenderDoc - 招标文档解析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncuploadtenderdoc.md) - - [GetBiddingRemainLimitNum - 获得标书写作剩余额度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingremainlimitnum.md) - - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) - - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) - - [DownloadBiddingDoc - 下载标书文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-downloadbiddingdoc.md) - - [AsyncWritingBiddingDoc - 标书写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncwritingbiddingdoc.md) - - [ListBiddingDoc - 列出标书写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-listbiddingdoc.md) - **其他** - [RunVideoScriptGenerate - AI生成视频剪辑脚本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-runvideoscriptgenerate.md) - [GetSmartClipTask - 获取智能剪辑任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getsmartcliptask.md) @@ -920,56 +1251,77 @@ - [ListDataPermissions - 权限-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatapermissions.md) - [GetPptInfo - 查询PPT任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getpptinfo.md) - [GenerateViewPoint - 生成选题视角（已过时，不推荐使用）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-generateviewpoint.md) + - [CancelAuditTask - 取消审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-cancelaudittask.md) + - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-queryaudittask.md) - [FetchParseDocumentLayoutTask - 获取排版任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-fetchparsedocumentlayouttask.md) - - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-endpoint.md) + - [SubmitAuditTask - 提交审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitaudittask.md) + - **标书生成** + - [AsyncUploadTenderDoc - 招标文档解析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncuploadtenderdoc.md) + - [GetBiddingRemainLimitNum - 获得标书写作剩余额度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingremainlimitnum.md) + - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) + - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) + - [DownloadBiddingDoc - 下载标书文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-downloadbiddingdoc.md) + - [AsyncWritingBiddingDoc - 标书写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncwritingbiddingdoc.md) + - [ListBiddingDoc - 列出标书写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-listbiddingdoc.md) - [API概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md) - - [版本说明](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-changeset.md) + - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-ram.md) + - [版本说明](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-changeset.md) - **更多** - [全妙服务关联角色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-slr.md) - [妙笔写作信源对接](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaobi-writing-source-docking.md) - [妙搜数据集管理通过API引入数据源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaosou-introduce-data-source-through-api.md) - [全妙iframe嵌入方案](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/iframe-embedding-scheme.md) - - [全妙Logo定制规范及部署方式](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/logo-customization-specification-and-deployment-method.md) - [全妙云存储（OSS）设置指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/oss-setup-guide.md) + - [全妙Logo定制规范及部署方式](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/logo-customization-specification-and-deployment-method.md) - [全妙PaaS AgentKey 获取指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-paas-agentkey-get-guide.md) + - **最佳实践** + - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) + - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) + - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) + - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) + - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) + - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) + - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) - [官方应用-通义听悟Agent](raw/application-user-guide/application-gallery/official-application-tingwu-agent.md) - - [通义法睿](raw/application-user-guide/application-gallery/tongyi-farui.md) - [官方应用-析言GBI](raw/application-user-guide/application-gallery/xiyan-gbi.md) + - [通义法睿](raw/application-user-guide/application-gallery/tongyi-farui.md) - **应用观测** - [应用观测](raw/application-user-guide/application-monitoring/application-observation.md) - **权限管理** - [权限管理](raw/application-user-guide/application-permission-management/application-permission-management-overview.md) +- **服务支持** + - [常见问题](raw/application-user-guide/application-support/application-faq.md) + - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) + - [阿里云百炼平台售后服务范围说明](raw/application-user-guide/application-support/application-after-sales-service-scope.md) - **实践教程** - [在网站上增加一个AI助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) - [10分钟让微信公众号成为智能客服](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - - [在企业微信中集成一个 AI 助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) - [在钉钉上增加一个AI机器人](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) - [基于本地知识库构建RAG应用](raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) -- **服务支持** - - [常见问题](raw/application-user-guide/application-support/application-faq.md) - - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) + - [在企业微信中集成一个 AI 助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) ## 模型 API 参考 - **使用 API** - - [获取API Key](raw/model-api-reference/preparations/get-api-key.md) - [安装SDK](raw/model-api-reference/preparations/install-sdk.md) + - [获取API Key](raw/model-api-reference/preparations/get-api-key.md) - [使用百炼 CLI](raw/model-api-reference/preparations/use-model-studio-cli.md) - [错误码](raw/model-api-reference/preparations/error-code.md) - **图像生成** - **千问** - - [千问-文生图API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像编辑API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) + - [千问-文生图API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像翻译API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) + - [千问-图像生成与编辑3.0 API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) - **万相** - [万相-文生图V2版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) - [万相-图像生成与编辑2.7 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - - [万相-通用图像编辑2.5](raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-图像生成与编辑2.6 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) - - [万相-通用图像编辑API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) + - [万相-通用图像编辑2.5](raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-涂鸦作画API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) + - [万相-通用图像编辑API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-图像局部重绘API参考](raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) - **Z-Image** - [Z-Image API参考](raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) @@ -984,168 +1336,189 @@ - [鞋靴模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) - [创意海报生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) - [人物实例分割API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - - [图像擦除补全API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) - - [AI试衣OutfitAnyone](raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [图像背景生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) + - [图像擦除补全API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) - [人物写真生成FaceChain](raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) + - [AI试衣OutfitAnyone](raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [创意文字WordArt锦书](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) - [常见问题](raw/model-api-reference/image-generation/image-faq.md) - **3D模型生成** - [Tripo-3D模型生成](raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) - **实时多模态** - [客户端事件](raw/model-api-reference/omni-realtime-api/client-events.md) - - [Python SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) - - [Java SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) - [服务端事件](raw/model-api-reference/omni-realtime-api/server-events.md) + - [Python SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) - [实时多模态交互流程](raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) + - [Java SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) - [声音复刻API参考](raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) +- **Realtime API** + - **最佳实践** + - [通过WebRTC使用多模态交互套件实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) + - [通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md) + - [通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md) + - **快速开始** + - [SDK下载](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md) + - [Token鉴权](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md) + - [实现接通模型/应用](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md) + - **AOQ客户端API** + - **AOQ SDK功能介绍** + - [连接状态管理](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md) + - [音频常用功能介绍](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md) + - [媒体流发送管理](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) + - [自定义音频采集](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md) + - [自定义音频播放](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md) + - [自定义视频输入](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md) + - [视频常用功能介绍](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md) + - [AOQ SDK简介](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md) + - [Realtime API简介](raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md) - **更多模型** - - [通义法睿大语言模型](raw/model-api-reference/more-models/tongyi-farui-api.md) - [意图理解能力](raw/model-api-reference/more-models/intent-detect-capability.md) - - [Qwen-Deep-Research API 参考](raw/model-api-reference/more-models/qwen-deep-research-api.md) - - [Qwen-MT API参考](raw/model-api-reference/more-models/qwen-mt-api.md) - - [GUI-Plus API参考](raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) + - [通义法睿大语言模型](raw/model-api-reference/more-models/tongyi-farui-api.md) - [Qwen-OCR API参考](raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) + - [Qwen-MT API参考](raw/model-api-reference/more-models/qwen-mt-api.md) + - [Qwen-Deep-Research API 参考](raw/model-api-reference/more-models/qwen-deep-research-api.md) - **工具包/框架** - [OpenAI Chat接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) - [OpenAI Responses接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [completions 接口](raw/model-api-reference/toolkits-and-frameworks/completions.md) - [OpenAI Vision接口兼容](raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) - [OpenAI文件接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) + - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI兼容-Batch Chat](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) - [OpenAI Embedding接口兼容](raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI Conversations接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md) - [在LangChain中使用阿里云百炼](raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md) - **模型生产** - - [模型调优](raw/model-api-reference/model-production/fine-tuning-jobs-api.md) - [模型部署](raw/model-api-reference/model-production/deployments-api.md) + - [模型调优](raw/model-api-reference/model-production/fine-tuning-jobs-api.md) - **更多** - [生成临时API Key](raw/model-api-reference/more-about-models/generate-temporary-api-key.md) - [异步任务管理 API](raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - - [通过HTTP回调URL或MQ接收异步任务完成通知](raw/model-api-reference/more-about-models/async-task-api.md) - [子业务空间的模型调用](raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) + - [通过HTTP回调URL或MQ接收异步任务完成通知](raw/model-api-reference/more-about-models/async-task-api.md) - [DashScope SDK连接复用配置](raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - [上传本地文件获取临时URL](raw/model-api-reference/more-about-models/get-temporary-file-url.md) - **视频生成** - **HappyHorse** + - [HappyHorse-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-文生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) - [HappyHorse-参考生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) - - [HappyHorse-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) - **万相** - **万相-早期视频模型（2.1-2.6）** - [万相-图生视频-基于首帧API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) - - [万相-文生视频API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) + - [万相-文生视频API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-首尾帧生视频API参考（2.2）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-视频编辑API参考（2.1）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) - [万相2.7-图生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) - [万相2.7-文生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - - [万相2.7-参考生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) - [万相2.7-视频编辑API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) + - [万相2.7-参考生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) - [万相-图生动作API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) - - [万相-数字人](raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - [万相-视频换人API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) + - [万相-数字人](raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - **爱诗** - - [爱诗-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) - [爱诗-文生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) - - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) + - [爱诗-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) - [爱诗-参考生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) - - [爱诗-视频超清API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) + - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - [爱诗-视频对口型API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) - [爱诗-视频动作模仿API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) + - [爱诗-视频超清API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) - **人像驱动** - [图生舞蹈视频-舞动人像AnimateAnyone](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - - [图生唱演视频-悦动人像EMO](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) - [图生播报视频-灵动人像LivePortrait](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) - - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) + - [图生唱演视频-悦动人像EMO](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) - [视频口型替换-声动人像VideoRetalk](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) - [视频风格重绘API参考](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) + - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) + - **可灵** + - [可灵-视频生成API文档](raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - **Vidu** - [Vidu-文生视频API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) - - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) + - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-参考生视频 API 参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) - - **可灵** - - [可灵-视频生成API文档](raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - **音频** - - **语音识别** - - **实时语音识别（Fun-ASR）** - - [Fun-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-websocket-api.md) - - [实时语音识别（Fun-ASR）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-client-events.md) - - [实时语音识别（Fun-ASR）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-server-events.md) - - [Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md) - - [Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-java-sdk.md) - - [Fun-ASR实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/android-sdk-for-fun-asr-real-time-service.md) - - [Fun-ASR实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/ios-sdk-for-fun-asr-real-time-service.md) - - **实时语音识别（Qwen-ASR-Realtime）** - - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) - - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) - - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) - - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) - - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) - - **实时语音识别（Paraformer）** - - [Paraformer实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/websocket-for-paraformer-real-time-service.md) - - [实时语音识别（Paraformer）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-client-events.md) - - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) - - [Paraformer实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md) - - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) - - [Paraformer实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md) - - [Paraformer实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/ios-sdk-for-paraformer-real-time-service.md) - - **录音文件识别（Fun-ASR）** - - [Fun-ASR录音文件识别HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md) - - [Fun-ASR录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md) - - [Fun-ASR录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md) - - [Fun-ASR录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md) - - [Fun-ASR录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md) - - **录音文件识别（Paraformer）** - - [Paraformer录音文件识别RESTful API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md) - - [Paraformer录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) - - [Paraformer录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md) - - [Paraformer录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md) - - [Paraformer录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md) - - [最佳实践](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-best-practices.md) - - **定制热词** - - [定制热词Python SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-python-sdk.md) - - [定制热词HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-http-api.md) - - [定制热词Java SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-java-sdk.md) - - [录音文件识别（Qwen-ASR）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md) - **语音合成** - **实时语音合成（Qwen-Audio-TTS/CosyVoice）** - [Qwen-Audio-TTS/CosyVoice WebSocket API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md) - [Qwen-Audio-TTS/CosyVoice客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md) - [Qwen-Audio-TTS/CosyVoice服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-server-events.md) - - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) - [实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md) - [语音合成Qwen-Audio-TTS/CosyVoice Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md) + - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) - [语音合成Qwen-Audio-TTS/CosyVoice iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md) - **实时语音合成（Qwen-TTS-Realtime）** - [Qwen-TTS-Realtime WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/interactive-process-of-qwen-tts-realtime-synthesis.md) - [客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-client-events.md) - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) - - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) - [Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-java-sdk.md) + - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) - **实时语音合成（Sambert）** - [Sambert WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-websocket-api.md) - [Sambert客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-client-events.md) - [Sambert服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-server-events.md) - [语音合成Sambert Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-java-sdk.md) - [语音合成Sambert Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-android-sdk.md) - - [语音合成Sambert Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-python-sdk.md) - [语音合成Sambert iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-ios-sdk.md) - - **非实时语音合成（MiniMax）** - - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) + - [语音合成Sambert Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-python-sdk.md) - **非实时语音合成（Qwen-Audio-TTS/CosyVoice）** - - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) - [非实时语音合成Qwen-Audio-TTS/CosyVoice HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md) - [非实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md) + - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) + - **非实时语音合成（MiniMax）** + - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) - **声音复刻** - [声音复刻HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md) - - [声音复刻Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md) - [声音复刻Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md) + - [声音复刻Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md) - [非实时语音合成（Qwen-TTS）API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md) - [声音设计API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/voice-design-api-references.md) + - **语音识别** + - **实时语音识别（Qwen-ASR-Realtime）** + - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) + - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) + - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) + - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) + - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) + - **实时语音识别（Paraformer）** + - [Paraformer实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/websocket-for-paraformer-real-time-service.md) + - [实时语音识别（Paraformer）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-client-events.md) + - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) + - [Paraformer实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md) + - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) + - [Paraformer实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/ios-sdk-for-paraformer-real-time-service.md) + - [Paraformer实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md) + - **非实时语音识别（Fun-ASR）** + - [Fun-ASR非实时语音识别HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md) + - [Fun-ASR非实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md) + - [Fun-ASR非实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md) + - [Fun-ASR非实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md) + - [Fun-ASR非实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md) + - **非实时语音识别（Paraformer）** + - [Paraformer非实时语音识别HTTP API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md) + - [Paraformer非实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md) + - [Paraformer非实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) + - [Paraformer非实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md) + - [Paraformer非实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md) + - [最佳实践](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-best-practices.md) + - **定制热词** + - [定制热词Python SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-python-sdk.md) + - [定制热词HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-http-api.md) + - [定制热词Java SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-java-sdk.md) + - **实时语音识别（Fun-ASR）** + - [实时语音识别（Fun-ASR）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-client-events.md) + - [Fun-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-websocket-api.md) + - [实时语音识别（Fun-ASR）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-server-events.md) + - [Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md) + - [Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-java-sdk.md) + - [Fun-ASR实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/android-sdk-for-fun-asr-real-time-service.md) + - [Fun-ASR实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/ios-sdk-for-fun-asr-real-time-service.md) + - [非实时语音识别（Fun-ASR-Flash）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md) + - [非实时语音识别（Fun-ASR-Realtime）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md) + - [非实时语音识别（Qwen-ASR）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md) - **音乐生成** - [音乐生成Fun-Music API参考](raw/model-api-reference/audio-api-references/music-generation-references/fun-music-api.md) - **语音翻译** @@ -1157,46 +1530,52 @@ - [音视频翻译-通义千问 API 参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/qwen3-livetranslate-flash-api.md) - **语音对话** - **实时语音对话** + - [Qwen-Audio 实时语音对话WebSocket API参考](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-realtime-websocket-api.md) - [Qwen-Audio 实时语音对话客户端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md) - - [Qwen-Audio 实时语音对话WebSocket API参考](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md) - [Qwen-Audio 实时语音对话服务端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md) - **向量与排序** - **通用文本向量** - [同步接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) - [批处理接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - - **排序模型（Rerank）** - - [文本排序](raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) - **多模态向量** - [Multimodal-Embedding API详情](raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) + - **排序模型（Rerank）** + - [文本排序](raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) - [文本生成模型API参考](raw/model-api-reference/qwen-api-reference.md) - [文件管理](raw/model-api-reference/file-management-api.md) ## 应用 API 参考 +- **应用调用** + - **Responses API** + - [同步调用 API 参考](raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) + - [异步调用API参考](raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) + - **DashScope API** + - [新版智能体应用 API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) + - [应用 DashScope API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) + - [获取APP ID和Workspace ID](raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) - **Managed Agents** - [API 总览与认证](raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) - [快速开始](raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - - [Agent](raw/application-api-reference/managed-agents-api/agent-api.md) - [Environment](raw/application-api-reference/managed-agents-api/environment-api.md) - - [Session and Event](raw/application-api-reference/managed-agents-api/session-api.md) + - [Agent](raw/application-api-reference/managed-agents-api/agent-api.md) - [File](raw/application-api-reference/managed-agents-api/files-api.md) + - [Session and Event](raw/application-api-reference/managed-agents-api/session-api.md) - [Skill](raw/application-api-reference/managed-agents-api/skills-api.md) - **应用组件** - **API目录** - **数据连接（原应用数据）** - - [AddCategory - 新增类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) - - [ListCategory - 类目列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [DeleteCategory - 删除类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - - [ApplyFileUploadLease - 申请文件上传租约](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) + - [ListCategory - 类目列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) + - [AddCategory - 新增类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) - [AddFile - 添加文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) - - [DescribeFile - 查询文件状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) + - [ApplyFileUploadLease - 申请文件上传租约](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [ListFile - 文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) + - [DescribeFile - 查询文件状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [UpdateFileTag - 更新文件标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [BatchUpdateFileTag - 批量更新文档标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) - - [DeleteFile - 删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) - [DeleteFiles - 批量删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - - [GetAvailableParserTypes - 获取文件支持的解析器类型](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [GetParseSettings - 获取类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) - [ChangeParseSetting - 修改类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) - [AddTable - 添加表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) @@ -1204,65 +1583,60 @@ - [AddConnector - 新增连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) - [GetConnector - 获取连接器信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [UpdateConnector - 编辑连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) + - [DeleteFile - 删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) + - [GetAvailableParserTypes - 获取文件支持的解析器类型](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - **知识库** - [CreateIndex - 创建知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) - [GetIndexJobStatus - 查询知识库创建任务状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) - [SubmitIndexJob - 提交知识库创建任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) + - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) - [Retrieve - 检索知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) - - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [ListIndexFileDetails - 查询知识库下的文件详情](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - [ListIndexDocuments - 查询知识库下的文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) - - [UpdateIndex - 更新知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) - [ListIndices - 查询知识库列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) + - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [DeleteIndex - 删除知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) - - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) + - [UpdateIndex - 更新知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) + - [AddChunk - 新增切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-addchunk.md) - [DeleteChunk - 删除切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) - [GetIndexMonitor - 获取知识库监控数据](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) + - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) + - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - **Prompt工程** - [CreatePromptTemplate - 创建Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) - [GetPromptTemplate - 获取Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) - [UpdatePromptTemplate - 更新Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) - - [ListPromptTemplates - 获取Prompt模板列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) - [DeletePromptTemplate - 删除Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) + - [ListPromptTemplates - 获取Prompt模板列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) - **其他** - **长期记忆（旧）** - [CreateMemory - 创建长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) - - [UpdateMemory - 更新长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) - [GetMemory - 获取长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - - [DeleteMemory - 删除长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) + - [UpdateMemory - 更新长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) - [ListMemories - 获取长期记忆体列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) - - [UpdateMemoryNode - 更新记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [CreateMemoryNode - 创建记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) + - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) + - [UpdateMemoryNode - 更新记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [DeleteMemoryNode - 删除记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) + - [DeleteMemory - 删除长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [GetAlipayTransferStatus - 查询支付宝打赏状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) - - [GetAlipayUrl - 获取支付宝打赏URL](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) - [ApplyTempStorageLease - 申请临时文件上传许可](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) + - [GetAlipayUrl - 获取支付宝打赏URL](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) - [API概览](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) - - [服务接入点](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [授权信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) + - [服务接入点](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [版本说明](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) -- **应用调用** - - **DashScope API** - - [新版智能体应用 API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - - [应用 DashScope API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) - - **Responses API** - - [同步调用 API 参考](raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) - - [异步调用API参考](raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) - - [获取APP ID和Workspace ID](raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) +- **长期记忆** + - [长期记忆（新）API 参考](raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - **框架** - **Spring AI Alibaba** - - [使用Spring AI Alibaba集成阿里云百炼大模型应用](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) - [通过Spring AI Alibaba检索阿里云百炼知识库](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) + - [使用Spring AI Alibaba集成阿里云百炼大模型应用](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) - [通过LlamaIndex API构建RAG应用](raw/application-api-reference/frameworks/llamaindex.md) -- **长期记忆** - - [长期记忆（新）API 参考](raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - **更多** - - [服务关联角色](raw/application-api-reference/more/bailian-service-linked-role.md) - [生成临时API Key](raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) + - [服务关联角色](raw/application-api-reference/more/bailian-service-linked-role.md) - [知识库SearchFilters](raw/application-api-reference/more/how-to-use-search-filters.md) - [知识检索与问答](raw/application-api-reference/knowledge.md) diff --git a/skills/bailian-docs-llm-wiki/models/families.jsonl b/skills/bailian-docs-llm-wiki/models/families.jsonl index 8bcd4a65..f8c2d385 100644 --- a/skills/bailian-docs-llm-wiki/models/families.jsonl +++ b/skills/bailian-docs-llm-wiki/models/families.jsonl @@ -34,7 +34,7 @@ {"slug":"image-erase-completion","name":"图像擦除补全","description":"图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-erase-completion","name":"图像擦除补全","capabilities":["IG"]}],"detailPath":"groups/image-erase-completion.json"} {"slug":"image-instance-segmentation","name":"人物实例分割","description":"人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-instance-segmentation","name":"人物实例分割","capabilities":["IG"]}],"detailPath":"groups/image-instance-segmentation.json"} {"slug":"image-out-painting","name":"图像画面扩展","description":"图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-out-painting","name":"图像画面扩展","capabilities":["IG"]}],"detailPath":"groups/image-out-painting.json"} -{"slug":"kimi-models-market-place","name":"Kimi","description":"由月之暗面提供的Kimi系列模型的API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["moonshot-ai"],"itemCount":4,"items":[{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","contextWindow":262144,"capabilities":["TG","VU","Reasoning"]},{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/kimi-models-market-place.json","maxContextWindow":262144} +{"slug":"kimi-models-market-place","name":"Kimi","description":"由月之暗面提供的Kimi系列模型的API服务。","primaryCapability":"TG","capabilities":["TG","VU","Reasoning"],"providers":["moonshot-ai"],"itemCount":5,"items":[{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","contextWindow":262144,"capabilities":["TG","VU","Reasoning"]},{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k3","name":"kimi/kimi-k3","contextWindow":1048576,"capabilities":["TG","VU","Reasoning"]}],"detailPath":"groups/kimi-models-market-place.json","maxContextWindow":1048576} {"slug":"kling-models-market-place","name":"可灵AI","description":"由可灵AI提供的高质量视频与图像生成及编辑模型。","primaryCapability":"VG","capabilities":["VG","IG"],"providers":["kling"],"itemCount":4,"items":[{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","capabilities":["IG"]},{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","capabilities":["IG"]},{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","capabilities":["VG"]},{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","capabilities":["VG"]}],"detailPath":"groups/kling-models-market-place.json"} {"slug":"liveportrait-detect","name":"灵动人像LivePortrait-detect","description":"LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","capabilities":["VG"]}],"detailPath":"groups/liveportrait-detect.json"} {"slug":"liveportrait","name":"灵动人像LivePortrait","description":"LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"liveportrait","name":"灵动人像LivePortrait","capabilities":["VG"]}],"detailPath":"groups/liveportrait.json"} @@ -54,18 +54,19 @@ {"slug":"pixverse-v6-market-place","name":"PixVerse V6","description":"由爱诗科技提供的PixVerse V系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-v6-market-place.json"} {"slug":"qvq-max","name":"QVQ-Max","description":"千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-max","name":"QVQ-Max","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-max.json","maxContextWindow":131072} {"slug":"qvq-plus","name":"Qwen-QVQ-Plus","description":"千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-plus","name":"QVQ-Plus","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-plus.json","maxContextWindow":131072} -{"slug":"qwen-audio-realtime-flash","name":"Qwen-Audio-Realtime-Flash","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-flash.json","maxContextWindow":8192} -{"slug":"qwen-audio-realtime-plus","name":"Qwen-Audio-Realtime-Plus","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-plus.json","maxContextWindow":8192} +{"slug":"qwen-audio-realtime-flash","name":"Qwen-Audio-Realtime-Flash","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","contextWindow":40960,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-flash.json","maxContextWindow":40960} +{"slug":"qwen-audio-realtime-plus","name":"Qwen-Audio-Realtime-Plus","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","contextWindow":40960,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-plus.json","maxContextWindow":40960} {"slug":"qwen-audio-tts","name":"Qwen-Audio-TTS","description":"Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。","primaryCapability":"Realtime-Text-to-Speech","capabilities":["Realtime-Text-to-Speech"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","capabilities":["Realtime-Text-to-Speech"]},{"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","capabilities":["Realtime-Text-to-Speech"]}],"detailPath":"groups/qwen-audio-tts.json"} {"slug":"qwen-coder-plus","name":"Qwen-Coder-Plus","description":"千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-plus.json","maxContextWindow":131072} {"slug":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","description":"Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-turbo.json","maxContextWindow":131072} {"slug":"qwen-deep-research","name":"qwen-deep-research","description":"千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-deep-research","name":"qwen-deep-research","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen-deep-research.json","maxContextWindow":1000000} {"slug":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","description":"快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen-doc-turbo.json","maxContextWindow":262144} -{"slug":"qwen-embedding","name":"Qwen-Embedding","description":"基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen-domain-model"],"itemCount":6,"items":[{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","capabilities":["TR"]},{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","capabilities":["TR"]},{"model":"text-embedding-v1","name":"通用文本向量-v1","capabilities":["TR"]},{"model":"text-embedding-v2","name":"通用文本向量-v2","capabilities":["TR"]},{"model":"text-embedding-v3","name":"通用文本向量-v3","capabilities":["TR"]},{"model":"text-embedding-v4","name":"通用文本向量-v4","capabilities":["TR"]}],"detailPath":"groups/qwen-embedding.json"} +{"slug":"qwen-embedding","name":"Qwen-Embedding","description":"基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen","qwen-domain-model"],"itemCount":7,"items":[{"model":"qwen3.7-text-embedding","name":"Qwen3.7-通用文本向量","contextWindow":131072,"capabilities":["TR"]},{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","capabilities":["TR"]},{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","capabilities":["TR"]},{"model":"text-embedding-v1","name":"通用文本向量-v1","capabilities":["TR"]},{"model":"text-embedding-v2","name":"通用文本向量-v2","capabilities":["TR"]},{"model":"text-embedding-v3","name":"通用文本向量-v3","capabilities":["TR"]},{"model":"text-embedding-v4","name":"通用文本向量-v4","capabilities":["TR"]}],"detailPath":"groups/qwen-embedding.json","maxContextWindow":131072} {"slug":"qwen-flash-character","name":"Qwen-Flash-Character","description":"千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash-character","name":"Qwen-Flash-Character","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/qwen-flash-character.json","maxContextWindow":8192} {"slug":"qwen-flash","name":"Qwen-Flash","description":"Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash","name":"Qwen-Flash","contextWindow":1000000,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen-flash.json","maxContextWindow":1000000} {"slug":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","description":"Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","capabilities":["IG"]}],"detailPath":"groups/qwen-image-2.0-pro.json"} {"slug":"qwen-image-2.0","name":"Qwen-Image-2.0","description":"Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","capabilities":["IG"]}],"detailPath":"groups/qwen-image-2.0.json"} +{"slug":"qwen-image-3.0-pro","name":"Qwen-Image-3.0-Pro","description":"内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。\n细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。\n知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。\nQwen-Image-3.0-Pro 不只是在追求\"好看\"，更在追求**“好用”**——让图像生成真正成为可落地的生产力工具。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-3.0-pro","name":"Qwen-Image-3.0-Pro","capabilities":["IG"]}],"detailPath":"groups/qwen-image-3.0-pro.json"} {"slug":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","description":"千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit-max.json"} {"slug":"qwen-image-edit","name":"Qwen-Image-Edit-Plus","description":"千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-image-edit","name":"Qwen-Image-Edit","capabilities":["IG"]},{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit.json"} {"slug":"qwen-image-max","name":"Qwen-Image-Max","description":"千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-max","name":"Qwen-Image-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-max.json"} @@ -117,7 +118,7 @@ {"slug":"qwen3-tts-vd","name":"Qwen3-TTS-VD","description":"Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vd.json"} {"slug":"qwen3-vl-flash","name":"Qwen3-VL-Flash","description":"Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。","primaryCapability":"VU","capabilities":["VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-vl-flash","name":"Qwen3-VL-Flash","contextWindow":262144,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3-vl-flash.json","maxContextWindow":262144} {"slug":"qwen3-vl-plus","name":"Qwen3-VL-Plus","description":"Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。","primaryCapability":"VU","capabilities":["VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-vl-plus","name":"Qwen3-VL-Plus","contextWindow":262144,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3-vl-plus.json","maxContextWindow":262144} -{"slug":"qwen3.5-flash","name":"Qwen3.5-Flash","description":"Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","contextWindow":1000000,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.5-flash.json","maxContextWindow":1000000} +{"slug":"qwen3.5-flash","name":"Qwen3.5-Flash","description":"Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","contextWindow":1000000,"capabilities":["Reasoning","TG","VU"]}],"detailPath":"groups/qwen3.5-flash.json","maxContextWindow":1000000} {"slug":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","description":"Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","contextWindow":53248,"capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json","maxContextWindow":53248} {"slug":"qwen3.5-ocr","name":"Qwen3.5-OCR","description":"Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","contextWindow":65536,"capabilities":["VU"]}],"detailPath":"groups/qwen3.5-ocr.json","maxContextWindow":65536} {"slug":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","contextWindow":262144,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3.5-omni-flash-realtime.json","maxContextWindow":262144} @@ -130,6 +131,7 @@ {"slug":"qwen3.6-max","name":"Qwen3.6-Max","description":"Qwen3.6原生Max模型，相较于此前发布的Qwen3-Max和Qwen3.6-Plus，本模型在vibe coding能力上进一步提升、coding agent执行更加高效、前端编程开发能力显著提升；长尾知识能力进一步升级。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.6-max-preview","name":"Qwen3.6-Max-Preview","contextWindow":262144,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen3.6-max.json","maxContextWindow":262144} {"slug":"qwen3.6-plus","name":"Qwen3.6-Plus","description":"Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.6-plus","name":"Qwen3.6-Plus","contextWindow":1000000,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.6-plus.json","maxContextWindow":1000000} {"slug":"qwen3.6","name":"Qwen3.6开源模型","description":"Qwen3.6系列开源模型，基于混合架构设计的原生视觉语言模型，模型效果相较于3.5系列同尺寸有大幅提升。","primaryCapability":"VU","capabilities":["VU","TG","Reasoning"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3.6-27b","name":"Qwen3.6-27B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","contextWindow":262144,"capabilities":["VU","TG","Reasoning"]}],"detailPath":"groups/qwen3.6.json","maxContextWindow":262144} +{"slug":"qwen3.7-flash","name":"Qwen3.7-Flash","description":"Qwen3.7原生视觉语言系列Flash模型，相较3.6-Flash全面提升多模态理解与Agent执行能力。重点强化多模态基础能力、万物识别能力更强，真实世界感知与空间智能进一步提升，Search Agent、CI Agent等多模态Agent场景能力显著升级、端到端任务执行更稳定，多模态Coding能力优化、vibe coding 体验更加流畅。","primaryCapability":"TG","capabilities":["TG","VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.7-flash","name":"Qwen3.7-Flash","contextWindow":1000000,"capabilities":["TG","VU","Reasoning"]}],"detailPath":"groups/qwen3.7-flash.json","maxContextWindow":1000000} {"slug":"qwen3.7-max","name":"Qwen3.7-Max","description":"Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3.7-max","name":"Qwen3.7-Max","contextWindow":1000000,"capabilities":["Reasoning","TG"]},{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","contextWindow":1000000,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/qwen3.7-max.json","maxContextWindow":1000000} {"slug":"qwen3.7-plus","name":"Qwen3.7-Plus","description":"Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.7-plus","name":"Qwen3.7-Plus","contextWindow":1000000,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/qwen3.7-plus.json","maxContextWindow":1000000} {"slug":"qwen3","name":"Qwen3开源模型","description":"Qwen3系列开源模型，包含混合模型、思考模型与非思考模型，思考能力与通用能力均达到同规模业界SOTA水平。","primaryCapability":"VU","capabilities":["VU","Reasoning","TG"],"providers":["qwen"],"itemCount":20,"items":[{"model":"qwen3-14b","name":"Qwen3-14B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","contextWindow":81920,"capabilities":["Reasoning","TG"]},{"model":"qwen3-32b","name":"Qwen3-32B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-8b","name":"Qwen3-8B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-coder-next","name":"通义千问3-Coder-Next","contextWindow":262144,"capabilities":["TG"]},{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3.json","maxContextWindow":262144} @@ -138,7 +140,7 @@ {"slug":"shoemodel-v1","name":"鞋靴模特","description":"鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"shoemodel-v1","name":"鞋靴模特","capabilities":["IG"]}],"detailPath":"groups/shoemodel-v1.json"} {"slug":"siliconflow-models","name":"SiliconFlow DeepSeek","description":"由硅基流动提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["deepseek"],"itemCount":4,"items":[{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","contextWindow":163840,"capabilities":["TG","Reasoning"]},{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","contextWindow":163840,"capabilities":["TG"]},{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","contextWindow":163840,"capabilities":["TG","Reasoning"]},{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","contextWindow":163840,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/siliconflow-models.json","maxContextWindow":163840} {"slug":"speech-biasing","name":"语音识别热词","description":"热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"speech-biasing","name":"语音识别热词","capabilities":["ASR"]}],"detailPath":"groups/speech-biasing.json"} -{"slug":"stepfun-models-market-place","name":"StepFun推理模型","description":"由阶跃星辰StepFun提供的Step系列推理模型API服务","primaryCapability":"TG","capabilities":["TG"],"providers":["stepfun"],"itemCount":1,"items":[{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/stepfun-models-market-place.json","maxContextWindow":262144} +{"slug":"stepfun-models-market-place","name":"StepFun推理模型","description":"由阶跃星辰StepFun提供的Step系列推理模型API服务","primaryCapability":"TG","capabilities":["TG","VU"],"providers":["stepfun"],"itemCount":1,"items":[{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","contextWindow":262144,"capabilities":["TG","VU"]}],"detailPath":"groups/stepfun-models-market-place.json","maxContextWindow":262144} {"slug":"tongyi-intent-detect-v3","name":"意图分类模型","description":"意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"tongyi-intent-detect-v3","name":"意图分类模型","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/tongyi-intent-detect-v3.json","maxContextWindow":8192} {"slug":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","description":"通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-flash.json","maxContextWindow":32768} {"slug":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","description":"通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-pro.json","maxContextWindow":32768} diff --git a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json index 15fdc4d5..9d280185 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json @@ -23,10 +23,44 @@ "prefix-completion" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi-k2.7-code", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.125", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.65", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -45,6 +79,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "VU", @@ -56,60 +93,23 @@ "contextWindow": 262144, "maxInputTokens": 229376, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "kimi-k2.7-code", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2.7-code\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.7-code\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2.7-code',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2.7-code\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.7-code\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2.7-code',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -s -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation\" \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"kimi-k2.7-code\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n }'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.comapi/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"你是谁\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.7-code',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2.7-code\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -s -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation\" \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"kimi-k2.7-code\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n }'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.comapi/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"你是谁\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.7-code',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2.7-code\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2948482.html" } } @@ -133,10 +133,44 @@ "model-experience" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi-k2.6", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.125", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.65", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -155,6 +189,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU", @@ -166,35 +203,22 @@ "latestOnlineAt": "2026-04-21T09:55:34.000+00:00", "contextWindow": 262144, "maxInputTokens": 229376, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Kimi-K2.6", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.6\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.6\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.6\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.6\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.6\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.6\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.6',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.6\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.6',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.6\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -217,11 +241,45 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi-k2.5", "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -240,6 +298,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU", @@ -250,35 +311,22 @@ "latestOnlineAt": "2026-01-30T01:49:03.000+00:00", "contextWindow": 262144, "maxInputTokens": 229376, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Kimi-K2.5", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.5\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.5\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.5\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.5\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.5\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.5\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.5',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.5\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.5',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.5\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -299,11 +347,31 @@ "function-calling" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi-k2-thinking", "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -322,6 +390,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -333,63 +404,26 @@ "maxInputTokens": 229376, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Kimi-K2-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"kimi-k2-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"kimi-k2-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -410,11 +444,31 @@ "function-calling" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "Moonshot-Kimi-K2-Instruct", "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -433,6 +487,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -441,54 +498,27 @@ "contextWindow": 131072, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Moonshot-Kimi-K2-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"Moonshot-Kimi-K2-Instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"Moonshot-Kimi-K2-Instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"Moonshot-Kimi-K2-Instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"Moonshot-Kimi-K2-Instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json index bd2b5e97..1297226e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json @@ -11,94 +11,37 @@ "Text" ] }, - "description": "MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，以多语言编程与长链 Agent 能力为核心优势。", + "description": "MiniMax-M2.5是MiniMax推出的旗舰级开源大模型，经过数十万个真实复杂环境中的大规模强化学习训练，M2.5 在编程、工具调用和搜索、办公等生产力场景都达到或者刷新了行业的 SOTA。", "features": [ "model-experience", "function-calling", "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax-M2.1", - "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2026-01-23T07:47:49.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 172032, - "inferenceProvider": "bailian", - "name": "MiniMax-M2.1", - "docUrl": "https://help.aliyun.com/document_detail/3017140.html", - "predictConfig": [ + "model": "MiniMax-M2.5", + "prices": [ { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" + "priceUnit": "每百万tokens", + "price": "2.1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" }, { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "priceUnit": "每百万tokens", + "price": "8.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "priceUnit": "每百万tokens", + "price": "0.42", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" } ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiniMax-M2.5是MiniMax推出的旗舰级开源大模型，经过数十万个真实复杂环境中的大规模强化学习训练，M2.5 在编程、工具调用和搜索、办公等生产力场景都达到或者刷新了行业的 SOTA。", - "features": [ - "model-experience", - "function-calling", - "cache" - ], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax-M2.5", "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -117,6 +60,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -127,49 +73,101 @@ "latestOnlineAt": "2026-02-24T15:07:02.000+00:00", "contextWindow": 204800, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "MiniMax-M2.5", "docUrl": "https://help.aliyun.com/document_detail/3017140.html", - "predictConfig": [ + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.5\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.5',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.5\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.5\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，以多语言编程与长链 Agent 能力为核心优势。", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "mini-max", + "model": "MiniMax-M2.1", + "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" + "priceUnit": "每百万tokens", + "price": "2.1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" }, { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "priceUnit": "每百万tokens", + "price": "8.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "priceUnit": "每百万tokens", + "price": "0.42", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" } ], + "priceTimeBands": [ + "standard" + ], + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2026-01-23T07:47:49.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 172032, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "MiniMax-M2.1", + "docUrl": "https://help.aliyun.com/document_detail/3017140.html", "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.5\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.5',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.5\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.5\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json index abeefc7d..f52de10f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json @@ -14,11 +14,24 @@ "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", "features": [], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/speech-2.8-turbo", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "timeBand": "standard", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "2", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -37,6 +50,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -46,7 +62,6 @@ "maxInputTokens": 10000, "inferenceProvider": "mini-max", "name": "speech-2.8-turbo", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -68,11 +83,24 @@ "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", "features": [], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/speech-2.8-hd", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "timeBand": "standard", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "3.5", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -91,6 +119,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -100,7 +131,6 @@ "maxInputTokens": 10000, "inferenceProvider": "mini-max", "name": "speech-2.8-hd", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -122,11 +152,24 @@ "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", "features": [], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/speech-02-turbo", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "timeBand": "standard", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "2", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -145,6 +188,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -154,7 +200,6 @@ "maxInputTokens": 10000, "inferenceProvider": "mini-max", "name": "speech-02-turbo", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -176,11 +221,24 @@ "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", "features": [], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/speech-02-hd", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "timeBand": "standard", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "3.5", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -199,6 +257,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -208,7 +269,6 @@ "maxInputTokens": 10000, "inferenceProvider": "mini-max", "name": "speech-02-hd", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json index 261b3171..ecab22ba 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json @@ -14,49 +14,37 @@ "description": "图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "aitryon-parsing-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "timeBand": "standard", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-15T13:13:06.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "AI试衣OutfitAnyone-图片分割", "docUrl": "https://help.aliyun.com/document_detail/2865249.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/vision/image-process/process' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"aitryon-parsing-v1\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250630/bakbqz/aitryon_parse_model.png\"\n },\n \"parameters\": {\n \"clothes_type\": [\"upper\"]\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/vision/image-process/process' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"aitryon-parsing-v1\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250630/bakbqz/aitryon_parse_model.png\"\n },\n \"parameters\": {\n \"clothes_type\": [\"upper\"]\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json index 2516bb8d..f9447f68 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json @@ -16,36 +16,37 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "aitryon-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-04-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "AI试衣-Plus版", "docUrl": "https://help.aliyun.com/document_detail/2881846.html", - "predictConfig": [ - { - "name": "sample_models" - }, - { - "name": "sample_suits" - }, - { - "name": "sample_tops" - }, - { - "name": "sample_bottoms" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-plus\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-plus\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json index 5fa687a4..cc41aae3 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json @@ -14,36 +14,127 @@ "description": "图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "aitryon-refiner", - "capabilities": [ - "IG" + "priceTimeBands": [ + "standard" ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-28T11:03:13.000+00:00", - "inferenceProvider": "bailian", - "name": "AI试衣OutfitAnyone-图片精修", - "docUrl": "https://help.aliyun.com/document_detail/2796663.html", - "predictConfig": [ + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "图片生成数量<=25", + "prices": [ + { + "priceUnit": "每张", + "price": "0.3", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 25 + }, + { + "rangeStart": 25, + "rangeName": "25<图片生成数量<=125", + "prices": [ + { + "priceUnit": "每张", + "price": "0.275", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 125 + }, { - "name": "sample_models" + "rangeStart": 125, + "rangeName": "125<图片生成数量<=250", + "prices": [ + { + "priceUnit": "每张", + "price": "0.25", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 250 }, { - "name": "sample_suits" + "rangeStart": 250, + "rangeName": "250<图片生成数量<=1250", + "prices": [ + { + "priceUnit": "每张", + "price": "0.225", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 1250 }, { - "name": "sample_tops" + "rangeStart": 1250, + "rangeName": "1250<图片生成数量<=2500", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 2500 }, { - "name": "sample_bottoms" + "rangeStart": 2500, + "rangeName": "2500<图片生成数量<=25000", + "prices": [ + { + "priceUnit": "每张", + "price": "0.175", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 25000 + }, + { + "rangeStart": 25000, + "rangeName": "25000<图片生成数量", + "prices": [ + { + "priceUnit": "每张", + "price": "0.15", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ] } ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-28T11:03:13.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "AI试衣OutfitAnyone-图片精修", + "docUrl": "https://help.aliyun.com/document_detail/2796663.html", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-refiner\",\n \"input\": {\n \"top_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-top.jpg\",\n \"bottom_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-bottom.jpg\",\n \"person_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-person.png\",\n \"coarse_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/result.png\"\n },\n \"parameters\": {\n \"gender\": \"woman\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-refiner\",\n \"input\": {\n \"top_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-top.jpg\",\n \"bottom_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-bottom.jpg\",\n \"person_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-person.png\",\n \"coarse_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/result.png\"\n },\n \"parameters\": {\n \"gender\": \"woman\"\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json index a1747bfc..aac566d9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json @@ -16,36 +16,37 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "aitryon", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-05-24T10:29:05.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "AI试衣-基础版", "docUrl": "https://help.aliyun.com/document_detail/2796626.html", - "predictConfig": [ - { - "name": "sample_models" - }, - { - "name": "sample_suits" - }, - { - "name": "sample_tops" - }, - { - "name": "sample_bottoms" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json index 5f83317b..a52c8c95 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json @@ -12,49 +12,37 @@ "description": "AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "animate-anyone-detect-gen2", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "timeBand": "standard", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-10T06:28:33.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "舞动人像AnimateAnyone-detect", "docUrl": "https://help.aliyun.com/document_detail/2786465.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"animate-anyone-detect-gen2\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {}\n}'" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"animate-anyone-detect-gen2\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {}\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json index 49465b13..37f73866 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json @@ -14,49 +14,37 @@ "description": "AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "animate-anyone-gen2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-10T06:25:42.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "舞动人像AnimateAnyone", "docUrl": "https://help.aliyun.com/document_detail/2786464.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-gen2\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"template_id\": \"AACT.xxx.xxx-xxx.xxx\" \n },\n \"parameters\": {\n \"use_ref_img_bg\": false,\n \"video_ratio\": \"9:16\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-gen2\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"template_id\": \"AACT.xxx.xxx-xxx.xxx\" \n },\n \"parameters\": {\n \"use_ref_img_bg\": false,\n \"video_ratio\": \"9:16\"\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json index a6232ced..d0c3f13b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json @@ -14,49 +14,37 @@ "description": "AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "animate-anyone-template-gen2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-10T06:27:17.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "舞动人像AnimateAnyone-template", "docUrl": "https://help.aliyun.com/document_detail/2807955.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-template-generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-template-gen2\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241210/cwjmsz/1.mp4\"\n },\n \"parameters\": {}\n }'\n" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-template-generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-template-gen2\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241210/cwjmsz/1.mp4\"\n },\n \"parameters\": {}\n }'\n" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json index 62f0c0f7..d7129c86 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json +++ b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json @@ -15,10 +15,16 @@ "collectionTag": "", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-v3.5-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -31,42 +37,23 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-02-27T09:13:23.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音生成CosyVoice-v3.5-flash大模型", "docUrl": "https://help.aliyun.com/document_detail/2842586.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-flash\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", - "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-flash\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", + "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-flash\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", + "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-flash\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -84,10 +71,16 @@ "description": "CosyVoice-v3.5-Plus是通义实验室CosyVoice系列的超高表现力语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-v3.5-plus", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1.5", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -100,21 +93,23 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-02-27T09:13:30.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音生成CosyVoice-v3.5-plus大模型", "docUrl": "https://help.aliyun.com/document_detail/2842586.html", - "predictConfig": [], "samples": { "dashscope": { "default": { - "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-plus\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", - "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-plus\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", + "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-plus\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", + "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-plus\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -132,10 +127,16 @@ "description": "合成能力：CosyVoice-v3-Flash是通义实验室CosyVoice系列最新版高性能的语音合成大模型，较之前版本在自然度、音质、韵律、情感表现力上有更好的表现。该模型支持文本至语音的实时流式合成。克隆能力：CosyVoice-v3-Flash也是通义实验室CosyVoice系列最新版的语音克隆大模型，较之前版本提升了发音准确性、音色相似度，并且增加了更多小语种支持（德、西、法、意、俄）。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-v3-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -148,35 +149,16 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-11-17T06:35:36.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音生成CosyVoice-v3-flash大模型", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], "samples": { "dashscope": { "default": { @@ -198,10 +180,16 @@ "description": "克隆能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音克隆大模型，具有更好的音质和复刻相似度，适用于更专业的场景。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。合成能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音合成大模型，具有更好的音质和表现力，适用于更专业的场景。该模型支持文本至语音的实时流式合成。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-v3-plus", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -214,42 +202,17 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-03T01:45:34.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音生成CosyVoice-v3-plus大模型", "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "情感", - "key": "emotion", - "default": "neutral", - "tip": "合成音频说话的情感" - }, - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], "samples": { "dashscope": { "default": { @@ -271,45 +234,33 @@ "description": "声音复刻Cosyvoice大模型，依托先进的大模型技术进行特征提取，从而完成声音的复刻，且无需训练过程。仅需提供时长较短的音频，即可迅速生成高度相似且听感自然的定制声音。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-clone-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "声音复刻CosyVoice大模型", - "docUrl": "https://help.aliyun.com/document_detail/2861519.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" } - ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "声音复刻CosyVoice大模型", + "docUrl": "https://help.aliyun.com/document_detail/2861519.html" }, { "inferenceMetadata": { @@ -320,50 +271,35 @@ "Text" ] }, - "description": "CosyVoice 是通义实验室依托大规模预训练语言模型，深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", - "features": [ - "model-experience" - ], + "description": "cosyvoice-V2是通义实验室依托大规模预训练语言模型，在深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", + "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v1", + "model": "cosyvoice-v2", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:34.000+00:00", - "inferenceProvider": "bailian", - "name": "语音合成CosyVoice大模型", + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-05-27T10:10:42.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "语音生成cosyvoice-v2大模型", "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v1\"\nvoice = \"longxiaochun\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v1\";\n private static String voice = \"longxiaochun\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v2\"\nvoice = \"longxiaochun_v2\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v2\";\n private static String voice = \"longxiaochun_v2\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" } } } @@ -377,53 +313,43 @@ "Text" ] }, - "description": "cosyvoice-V2是通义实验室依托大规模预训练语言模型，在深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", - "features": [], + "description": "CosyVoice 是通义实验室依托大规模预训练语言模型，深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", + "features": [ + "model-experience" + ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v2", + "model": "cosyvoice-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-05-27T10:10:42.000+00:00", - "inferenceProvider": "bailian", - "name": "语音生成cosyvoice-v2大模型", - "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - }, - { - "name": "字级别时间戳", - "key": "enableWordTimestamp", - "default": false + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:34.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" } - ], + }, + "inferenceProvider": "aliyun-bailian", + "name": "语音合成CosyVoice大模型", + "docUrl": "https://help.aliyun.com/document_detail/2817551.html", "samples": { "dashscope": { "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v2\"\nvoice = \"longxiaochun_v2\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v2\";\n private static String voice = \"longxiaochun_v2\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v1\"\nvoice = \"longxiaochun\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v1\";\n private static String voice = \"longxiaochun\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json index 00cd5780..c7e46d56 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json +++ b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json @@ -24,18 +24,21 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -58,6 +61,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -71,59 +77,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V4-Pro", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 4000, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -165,18 +118,21 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -199,6 +155,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -213,59 +172,6 @@ "name": "DeepSeek-V4-Flash", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 4000, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -308,48 +214,56 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -357,6 +271,7 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -364,22 +279,25 @@ ], "qpmInfo": { "model-default-actual": { - "count_limit_period": 15, + "count_limit_period": 60, "usage_limit": 1200000, "usage_limit_field": "total_tokens", - "count_limit": 3750, + "count_limit": 15000, "usage_limit_period": 60, "type": "model-default" }, "model-default": { - "count_limit_period": 15, + "count_limit_period": 60, "usage_limit": 1200000, "usage_limit_field": "total_tokens", - "count_limit": 3750, + "count_limit": 15000, "usage_limit_period": 60, "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -398,49 +316,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3.2", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -481,12 +356,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -509,6 +386,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -527,49 +407,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Deepseek-V3.2-Exp", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -611,18 +448,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -645,6 +485,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -663,49 +506,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3.1", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -747,30 +547,35 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -793,6 +598,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -811,43 +619,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -888,12 +659,14 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -916,6 +689,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning" ], @@ -933,23 +709,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-0528", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -989,30 +748,35 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -1035,6 +799,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning" ], @@ -1052,23 +819,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1105,12 +855,14 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -1133,6 +885,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -1150,23 +905,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-7B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1203,12 +941,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -1231,6 +971,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -1248,23 +991,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-32B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1301,12 +1027,14 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -1329,6 +1057,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -1346,23 +1077,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-14B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1424,23 +1138,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-1.5B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/embedding.json b/skills/bailian-docs-llm-wiki/models/groups/embedding.json index 387fe2bf..fc01efe1 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/embedding.json @@ -14,10 +14,23 @@ "description": "Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，在视觉向量化上具备极高性价比。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "tongyi-embedding-vision-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -36,45 +49,21 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T09:09:49.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "视觉向量-flash", "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-flash\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "curl": "curl --silent --location --request POST '[workspace-id].cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-flash\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"tongyi-embedding-vision-flash\",\n input=input\n)\n\nprint(resp)\n", "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"tongyi-embedding-vision-flash\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" } @@ -93,10 +82,23 @@ "description": "Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "tongyi-embedding-vision-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -115,45 +117,21 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T09:09:31.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "视觉向量-plus", "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-plus\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "curl": "curl --silent --location --request POST '[workspace-id].cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-plus\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"tongyi-embedding-vision-plus\",\n input=input\n)\n\nprint(resp)\n", "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"tongyi-embedding-vision-plus\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" } @@ -172,10 +150,23 @@ "description": "通义实验室基于预训练多模态大模型构建的多模态向量模型。该模型根据用户的输入生成高维连续向量，这些输入可以是文本、图片或视频。多模态向量在可应用于图片搜索、文搜图、视频搜索、图片分类和视频内容审核等下游任务中。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "multimodal-embedding-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.9", + "timeBand": "standard", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -194,6 +185,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], @@ -202,40 +196,13 @@ "latestOnlineAt": "2024-12-23T11:55:31.000+00:00", "contextWindow": 0, "maxInputTokens": 512, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用多模态向量", "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"multimodal-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"multimodal-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"multimodal-embedding-v1\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"multimodal-embedding-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json index 114ba204..acc736db 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json @@ -12,49 +12,37 @@ "description": "EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "emo-detect-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "timeBand": "standard", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-11-07T14:16:51.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "悦动人像EMO-detect", "docUrl": "https://help.aliyun.com/document_detail/2786463.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emo-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/aejgyj/input_audio.mp3\",\n \"face_bbox\":[302,286,610,593],\n \"ext_bbox\":[71,9,840,778]\n },\n \"parameters\": {\n \"style_level\": \"normal\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emo-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/aejgyj/input_audio.mp3\",\n \"face_bbox\":[302,286,610,593],\n \"ext_bbox\":[71,9,840,778]\n },\n \"parameters\": {\n \"style_level\": \"normal\"\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json index d2b6cb99..5be8940b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json @@ -14,49 +14,44 @@ "description": "EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "emo-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "timeBand": "standard", + "type": "video_duration_1-1", + "priceName": "视频生成（1:1画幅视频）" + }, + { + "priceUnit": "每秒", + "price": "0.16", + "timeBand": "standard", + "type": "video_duration_3-4", + "priceName": "视频生成（3:4画幅视频）" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-11-07T14:16:42.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "悦动人像EMO", "docUrl": "https://help.aliyun.com/document_detail/2786461.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"emo-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\"\n },\n \"parameters\": {\n \"ratio\": \"1:1\"\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"emo-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\"\n },\n \"parameters\": {\n \"ratio\": \"1:1\"\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json index ae9f1a1a..ba35481e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json @@ -12,49 +12,37 @@ "description": "表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "emoji-detect-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "timeBand": "standard", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-16T09:55:26.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "表情包Emoji-detect", "docUrl": "https://help.aliyun.com/document_detail/2865371.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {\n \"ratio\":\"1:1\"\n }\n }'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {\n \"ratio\":\"1:1\"\n }\n }'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json index 5c8c9829..dec649c9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json @@ -14,49 +14,37 @@ "description": "表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "emoji-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-16T09:46:00.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "表情包Emoji", "docUrl": "https://help.aliyun.com/document_detail/2865374.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"driven_id\": \"mengwa_kaixin\",\n \"face_bbox\": [212,194,460,441],\n \"ext_bbox\": [63,30,609,575]\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"driven_id\": \"mengwa_kaixin\",\n \"face_bbox\": [212,194,460,441],\n \"ext_bbox\": [63,30,609,575]\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json index b5948c96..b533698d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json +++ b/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json @@ -12,45 +12,15 @@ "description": "对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "facechain-facedetect", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T08:10:47.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "FaceChain人物图像检测", - "docUrl": "https://help.aliyun.com/document_detail/2712507.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] + "docUrl": "https://help.aliyun.com/document_detail/2712507.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json index d32c1391..6add2e2c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json +++ b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json @@ -14,45 +14,33 @@ "description": "基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "facechain-generation", + "prices": [ + { + "priceUnit": "每张", + "price": "0.18", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T08:12:23.000+00:00", - "inferenceProvider": "bailian", - "name": "FaceChain人物写真生成", - "docUrl": "https://help.aliyun.com/document_detail/2712501.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" } - ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "FaceChain人物写真生成", + "docUrl": "https://help.aliyun.com/document_detail/2712501.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json index ec9bd5b0..3fa87312 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json @@ -16,10 +16,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "farui-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -46,50 +62,29 @@ "latestOnlineAt": "2024-05-14T13:32:40.000+00:00", "contextWindow": 12000, "maxInputTokens": 12000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "通义法睿-Plus-32K", "docUrl": "https://help.aliyun.com/document_detail/2778998.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"farui-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"farui-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"farui-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"farui-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"farui-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"farui-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"farui-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"farui-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"farui-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"farui-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"farui-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"farui-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json index 14412c98..9315f829 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json @@ -14,10 +14,16 @@ "description": "百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr-flash-2026-06-15", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -30,45 +36,21 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-06-17T08:56:52.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR-Flash-2026-06-15", "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header \"Content-Type: application/json\" \\\n --header \"X-DashScope-SSE: enable\" \\\n --data '{\n \"model\": \"fun-asr-flash-2026-06-15\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"input_audio\",\n \"input_audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"format\": \"wav\",\n \"sample_rate\": \"16000\"\n }\n}'", + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header \"Content-Type: application/json\" \\\n --header \"X-DashScope-SSE: enable\" \\\n --data '{\n \"model\": \"fun-asr-flash-2026-06-15\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"input_audio\",\n \"input_audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"format\": \"wav\",\n \"sample_rate\": \"16000\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/2869541.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json index 79fb2a3d..69ca7592 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json @@ -14,10 +14,16 @@ "description": "通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00033", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -30,32 +36,17 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T11:05:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR实时语音识别", "docUrl": "https://help.aliyun.com/document_detail/2842554.html", - "predictConfig": [ - { - "name": "开启语义断句", - "key": "semantic_punctuation_enabled", - "default": false, - "tip": "开启语义断句后则将关闭VAD(语音活动检测)断句，具体见说明文档" - }, - { - "name": "VAD静音阈值", - "key": "max_sentence_silence", - "default": 1300, - "tip": "VAD(语音活动检测)断句的静音时长阈值(单位为ms)", - "range": [ - 200, - 6000 - ] - } - ], "samples": { "dashscope": { "default": { @@ -77,10 +68,16 @@ "description": "通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr-flash-8k-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -93,6 +90,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], @@ -100,37 +100,10 @@ "versionTag": "MAJOR", "equivalentSnapshot": "fun-asr-flash-8k-realtime-2026-01-28", "latestOnlineAt": "2026-02-12T11:34:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR-Flash-8k实时语音识别", "docUrl": "https://help.aliyun.com/document_detail/2842554.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json index eac84c5d..3cf5006c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json @@ -14,10 +14,16 @@ "description": "百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本等同于2025年11月7日的快照版本。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -30,15 +36,17 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-11-20T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR语音识别", "docUrl": "https://help.aliyun.com/document_detail/2880903.html", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -59,10 +67,16 @@ "description": "百聆多语言语音识别大模型，支持超过31种语言，支持语种自由切换，出海用户首推，尤其东南亚出海。fun-asr为该模型的升级版本，建议切换使用fun-asr。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr-mtl", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -75,15 +89,23 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-25T06:03:29.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR-MTL", "docUrl": "https://help.aliyun.com/document_detail/2978300.html", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json index 3751a275..90f25aac 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json @@ -14,10 +14,16 @@ "description": "百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "fun-music-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.002", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -30,46 +36,22 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-05-06T12:15:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "音乐生成", "docUrl": "https://help.aliyun.com/document_detail/3030448.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl -X POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-v1\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", + "curl": "curl -X POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-v1\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3030448.html" } } @@ -87,10 +69,16 @@ "description": "百聆音乐生成preview版大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。本次版本为预览快照版", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "fun-music-preview", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.005", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -103,45 +91,21 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-06-01T07:20:31.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "音乐生成 Preview", "docUrl": "https://help.aliyun.com/document_detail/3030448.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl -X POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-preview\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", + "curl": "curl -X POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-preview\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3030448.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json index ad33f246..9b7f2c9d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json @@ -19,10 +19,30 @@ "structured-outputs" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-5.2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -41,6 +61,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -51,71 +74,23 @@ "contextWindow": 1048576, "maxInputTokens": 1048576, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GLM-5.2", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -138,9 +113,6 @@ "structured-outputs" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-5.1", "qpmInfo": { "model-default-actual": { @@ -160,6 +132,95 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 204800 + } + ], "capabilities": [ "TG", "Reasoning" @@ -170,72 +231,24 @@ "latestOnlineAt": "2026-04-14T11:34:59.000+00:00", "contextWindow": 202745, "maxInputTokens": 202745, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GLM-5.1", "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'glm-5.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'glm-5.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -250,115 +263,120 @@ "Text" ] }, - "description": "智谱最新旗舰，具备更强的编程能力与更稳定的多步骤推理/执行能力。总参数355B，支持长程任务规划、编码、工具协同，问答自然、写作沉浸、创意角色扮演能力强。", + "description": "GLM-5是面向Coding与Agent场景的新一代大模型，在复杂系统工程与长程任务中达到开源 SOTA，真实编程体验逼近 Claude Opus 级别；基于 744B 新基座、异步强化学习与稀疏注意力，实现从“写代码”到“写工程”的全面升级。", "features": [ "model-experience", "function-calling", "cache" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-4.7", + "model": "glm-5", "qpmInfo": { "model-default-actual": { "count_limit_period": 6, - "usage_limit": 1000000, + "usage_limit": 100000, "usage_limit_field": "total_tokens", "count_limit": 50, - "usage_limit_period": 60, + "usage_limit_period": 6, "type": "model-default" }, "model-default": { "count_limit_period": 6, - "usage_limit": 1000000, + "usage_limit": 100000, "usage_limit_field": "total_tokens", "count_limit": 50, - "usage_limit_period": 60, + "usage_limit_period": 6, "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "22", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], "capabilities": [ "TG", "Reasoning" ], + "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 16384, - "latestOnlineAt": "2025-12-25T06:06:48.000+00:00", + "latestOnlineAt": "2026-02-18T04:44:16.000+00:00", "contextWindow": 202752, "maxInputTokens": 169984, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "GLM-4.7", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], + "inferenceProvider": "aliyun-bailian", + "name": "GLM-5", + "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.7\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.7\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } @@ -373,110 +391,125 @@ "Text" ] }, - "description": "GLM-5是面向Coding与Agent场景的新一代大模型，在复杂系统工程与长程任务中达到开源 SOTA，真实编程体验逼近 Claude Opus 级别；基于 744B 新基座、异步强化学习与稀疏注意力，实现从“写代码”到“写工程”的全面升级。", + "description": "智谱最新旗舰，具备更强的编程能力与更稳定的多步骤推理/执行能力。总参数355B，支持长程任务规划、编码、工具协同，问答自然、写作沉浸、创意角色扮演能力强。", "features": [ "model-experience", "function-calling", "cache" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-5", + "model": "glm-4.7", "qpmInfo": { "model-default-actual": { "count_limit_period": 6, - "usage_limit": 100000, + "usage_limit": 1000000, "usage_limit_field": "total_tokens", "count_limit": 50, - "usage_limit_period": 6, + "usage_limit_period": 60, "type": "model-default" }, "model-default": { "count_limit_period": 6, - "usage_limit": 100000, + "usage_limit": 1000000, "usage_limit_field": "total_tokens", "count_limit": 50, - "usage_limit_period": 6, + "usage_limit_period": 60, "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], "capabilities": [ "TG", "Reasoning" ], - "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 16384, - "latestOnlineAt": "2026-02-18T04:44:16.000+00:00", + "latestOnlineAt": "2025-12-25T06:06:48.000+00:00", "contextWindow": 202752, "maxInputTokens": 169984, - "inferenceProvider": "bailian", - "name": "GLM-5", - "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } - ], + }, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-4.7", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html", "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.7\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.7\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } @@ -497,9 +530,6 @@ "cache" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-4.6", "qpmInfo": { "model-default-actual": { @@ -519,6 +549,67 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], "capabilities": [ "Reasoning", "TG" @@ -530,75 +621,27 @@ "maxInputTokens": 169984, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GLM-4.6", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.6\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.6\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.6\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.6\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } @@ -618,9 +661,6 @@ "model-experience" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-4.5", "qpmInfo": { "model-default-actual": { @@ -640,6 +680,53 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + } + ], "capabilities": [ "TG" ], @@ -648,71 +735,29 @@ "latestOnlineAt": "2025-08-06T11:38:01.000+00:00", "contextWindow": 131072, "maxInputTokens": 98304, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "GLM-4.5", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } @@ -732,9 +777,6 @@ "model-experience" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-4.5-air", "qpmInfo": { "model-default-actual": { @@ -754,6 +796,53 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + } + ], "capabilities": [ "TG" ], @@ -762,71 +851,29 @@ "latestOnlineAt": "2025-08-06T12:02:34.000+00:00", "contextWindow": 131072, "maxInputTokens": 98304, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "GLM-4.5-Air", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5-air\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5-air\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5-air\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5-air\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json index 69f63105..b96913ed 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json +++ b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json @@ -21,6 +21,29 @@ ], "provider": "zhipu-ai", "model": "glm-5.2-fast-preview", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "56", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -39,6 +62,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -51,7 +77,7 @@ ], "contextWindow": 1048576, "maxInputTokens": 1048576, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GLM-5.2-Fast-Preview", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", "category": "Third-party", diff --git a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json index 3f5f3c4c..6bacfb0c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json @@ -15,10 +15,23 @@ "description": "GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "gui-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -37,6 +50,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -45,50 +61,23 @@ "latestOnlineAt": "2025-11-12T07:17:19.000+00:00", "contextWindow": 256000, "maxInputTokens": 254976, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GUI-Plus", "docUrl": "https://help.aliyun.com/document_detail/2997010.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "node": "import OpenAI from \"openai\";\n\nconst systemPrompt = `# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by \\`action=key\\`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by \\`action=wait\\`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by \\`action=terminate\\`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.`;\n\nconst messages = [\n {\n role: \"system\",\n content: systemPrompt\n },\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n type: \"text\",\n text: \"帮我打开浏览器。\"\n }\n ]\n }\n];\n\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"gui-plus\",\n messages: messages,\n });\n console.log(response.choices[0].message.content);\n}\nmain();\n", - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by \\`action=key\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by \\`action=wait\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by \\`action=terminate\\`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"帮我打开浏览器\"\n }\n ]\n }\n ]\n}\nEOF\n", - "python": "import os\nfrom openai import OpenAI\n\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": system_prompt,\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n },\n {\"type\": \"text\", \"text\": \"帮我打开浏览器。\"},\n ],\n },\n]\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(model=\"gui-plus\", messages=messages)\nprint(completion.choices[0].message.content)\n", + "node": "import OpenAI from \"openai\";\n\nconst systemPrompt = `# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by \\`action=key\\`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by \\`action=wait\\`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by \\`action=terminate\\`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.`;\n\nconst messages = [\n {\n role: \"system\",\n content: systemPrompt\n },\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n type: \"text\",\n text: \"帮我打开浏览器。\"\n }\n ]\n }\n];\n\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"gui-plus\",\n messages: messages,\n });\n console.log(response.choices[0].message.content);\n}\nmain();\n", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by \\`action=key\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by \\`action=wait\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by \\`action=terminate\\`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"帮我打开浏览器\"\n }\n ]\n }\n ]\n}\nEOF\n", + "python": "import os\nfrom openai import OpenAI\n\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": system_prompt,\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n },\n {\"type\": \"text\", \"text\": \"帮我打开浏览器。\"},\n ],\n },\n]\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(model=\"gui-plus\", messages=messages)\nprint(completion.choices[0].message.content)\n", "docUrl": "https://help.aliyun.com/document_detail/2997010.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* key: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* type: Type a string of text on the keyboard.\\\\n* mouse_move: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* left_click: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* left_click_drag: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* right_click: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* middle_click: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* double_click: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* triple_click: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* scroll: Performs a scroll of the mouse scroll wheel.\\\\n* hscroll: Performs a horizontal scroll (mapped to regular scroll).\\\\n* wait: Wait specified seconds for the change to happen.\\\\n* terminate: Terminate the current task and report its completion status.\\\\n* answer: Answer a question.\\\\n* interact: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by action=key.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by action=type, action=answer and action=interact.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by action=mouse_move and action=left_click_drag.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by action=scroll and action=hscroll.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by action=wait.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by action=terminate.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n {\n \"text\": \"帮我打开浏览器。\"\n }\n ]\n }\n ]\n }\n}\nEOF", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [{\n \"role\": \"system\",\n \"content\": system_prompt\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"},\n {\"text\": \"帮我打开浏览器。\"}]\n}]\n\nresponse = dashscope.MultiModalConversation.call(\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'gui-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n\n String systemPrompt = \"# Tools\\n\\n\" +\n \"You may call one or more functions to assist with the user query.\\n\\n\" +\n \"You are provided with function signatures within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* `type`: Type a string of text on the keyboard.\\\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* `wait`: Wait specified seconds for the change to happen.\\\\n* `terminate`: Terminate the current task and report its completion status.\\\\n* `answer`: Answer a question.\\\\n* `interact`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by `action=key`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by `action=type`, `action=answer` and `action=interact`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by `action=wait`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by `action=terminate`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\" +\n \"\\n\\n\" +\n \"For each function call, return a json object with function name and arguments within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"name\\\": , \\\"arguments\\\": }\\n\" +\n \"\\n\\n\" +\n \"# Response format\\n\\n\" +\n \"Response format for every step:\\n\" +\n \"1) Action: a short imperative describing what to do in the UI.\\n\" +\n \"2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\n\" +\n \"Rules:\\n\" +\n \"- Output exactly in the order: Action, .\\n\" +\n \"- Be brief: one for Action.\\n\" +\n \"- Do not output anything else outside those two parts.\\n\" +\n \"- If finishing, use action=terminate in the tool call.\";\n\n MultiModalMessage systemMsg = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", systemPrompt))).build();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"),\n Collections.singletonMap(\"text\", \"帮我打开浏览器。\"))).build();\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"gui-plus\")\n .messages(Arrays.asList(systemMsg, userMessage))\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}\n", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* key: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* type: Type a string of text on the keyboard.\\\\n* mouse_move: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* left_click: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* left_click_drag: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* right_click: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* middle_click: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* double_click: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* triple_click: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* scroll: Performs a scroll of the mouse scroll wheel.\\\\n* hscroll: Performs a horizontal scroll (mapped to regular scroll).\\\\n* wait: Wait specified seconds for the change to happen.\\\\n* terminate: Terminate the current task and report its completion status.\\\\n* answer: Answer a question.\\\\n* interact: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by action=key.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by action=type, action=answer and action=interact.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by action=mouse_move and action=left_click_drag.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by action=scroll and action=hscroll.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by action=wait.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by action=terminate.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n {\n \"text\": \"帮我打开浏览器。\"\n }\n ]\n }\n ]\n }\n}\nEOF", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [{\n \"role\": \"system\",\n \"content\": system_prompt\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"},\n {\"text\": \"帮我打开浏览器。\"}]\n}]\n\nresponse = dashscope.MultiModalConversation.call(\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'gui-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n\n String systemPrompt = \"# Tools\\n\\n\" +\n \"You may call one or more functions to assist with the user query.\\n\\n\" +\n \"You are provided with function signatures within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* `type`: Type a string of text on the keyboard.\\\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* `wait`: Wait specified seconds for the change to happen.\\\\n* `terminate`: Terminate the current task and report its completion status.\\\\n* `answer`: Answer a question.\\\\n* `interact`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by `action=key`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by `action=type`, `action=answer` and `action=interact`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by `action=wait`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by `action=terminate`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\" +\n \"\\n\\n\" +\n \"For each function call, return a json object with function name and arguments within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"name\\\": , \\\"arguments\\\": }\\n\" +\n \"\\n\\n\" +\n \"# Response format\\n\\n\" +\n \"Response format for every step:\\n\" +\n \"1) Action: a short imperative describing what to do in the UI.\\n\" +\n \"2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\n\" +\n \"Rules:\\n\" +\n \"- Output exactly in the order: Action, .\\n\" +\n \"- Be brief: one for Action.\\n\" +\n \"- Do not output anything else outside those two parts.\\n\" +\n \"- If finishing, use action=terminate in the tool call.\";\n\n MultiModalMessage systemMsg = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", systemPrompt))).build();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"),\n Collections.singletonMap(\"text\", \"帮我打开浏览器。\"))).build();\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"gui-plus\")\n .messages(Arrays.asList(systemMsg, userMessage))\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}\n", "docUrl": "https://help.aliyun.com/document_detail/2997010.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json index 69390930..25f4e375 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json @@ -16,10 +16,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "gummy-chat-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00015", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -32,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], @@ -40,13 +49,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "一句话识别及翻译V1.0", "docUrl": "https://help.aliyun.com/document_detail/2866122.html", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json index eefec95c..0b94192d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json @@ -16,10 +16,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "gummy-realtime-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00015", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -32,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Audio-Translate" ], @@ -39,14 +48,13 @@ "latestOnlineAt": "2025-03-04T04:07:10.000+00:00", "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "实时语音识别及翻译V1.0", "docUrl": "https://help.aliyun.com/document_detail/2865393.html", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json index 2e70e68e..a98be88f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json @@ -12,16 +12,43 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。", "features": [ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.1-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -40,45 +67,20 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2026-06-16T03:16:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.1-I2V", "docUrl": "https://help.aliyun.com/document_detail/3029821.html", "category": "Visual", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029821.html" } } @@ -99,15 +101,30 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.0-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "timeBand": "standard", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -115,51 +132,28 @@ "model-default": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-21T17:07:07.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-I2V", "docUrl": "https://help.aliyun.com/document_detail/3029821.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029821.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json index 8b69d7f3..c1473542 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json @@ -17,10 +17,23 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.1-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,50 +52,20 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2026-06-16T03:16:08.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.1-R2V", "docUrl": "https://help.aliyun.com/document_detail/3030778.html", "category": "Visual", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", "docUrl": "https://help.aliyun.com/document_detail/3030778.html" } } @@ -103,15 +86,30 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.0-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "timeBand": "standard", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -119,56 +117,28 @@ "model-default": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-26T12:42:22.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-R2V", "docUrl": "https://help.aliyun.com/document_detail/3030778.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", "docUrl": "https://help.aliyun.com/document_detail/3030778.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json index edb5bce7..2b6c624a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json @@ -16,10 +16,23 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.1-t2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,50 +51,20 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2026-06-16T06:14:47.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.1-T2V", "docUrl": "https://help.aliyun.com/document_detail/3029820.html", "category": "Visual", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029820.html" } } @@ -101,70 +84,57 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.0-t2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "timeBand": "standard", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "type": "model-default", "async_user_concurrency_limit": 5 }, "model-default": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "type": "model-default", "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-21T09:03:16.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-T2V", "docUrl": "https://help.aliyun.com/document_detail/3029820.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029820.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json index 01a4bf23..45abf738 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json @@ -8,6 +8,7 @@ "Video" ], "request_modality": [ + "Text", "Image", "Video" ] @@ -15,15 +16,30 @@ "description": "HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。", "features": [], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.0-video-edit", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "timeBand": "standard", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -31,54 +47,29 @@ "model-default": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-26T07:51:50.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-Video-Edit", "docUrl": "https://help.aliyun.com/document_detail/3030779.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration" - }, - { - "name": "声音设置", - "key": "audio_setting", - "tip": [ - "auto （默认）：模型根据 prompt 内容智能判断。若提示词涉及声音描述，可能重新生成音频；否则可能保留输入素材的原声。", - "origin：强制保留输入视频的原声，不重新生成。" - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], + "category": "Visual", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-video-edit\",\n \"input\": {\n \"prompt\": \"让视频中的马头人身角色穿上图片中的条纹毛衣\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260409/dozxak/Wan_Video_Edit_33_1.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260415/hynnff/wan-video-edit-clothes.webp\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-video-edit\",\n \"input\": {\n \"prompt\": \"让视频中的马头人身角色穿上图片中的条纹毛衣\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260409/dozxak/Wan_Video_Edit_33_1.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260415/hynnff/wan-video-edit-clothes.webp\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3030779.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json b/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json index 05868c52..bd5ffed3 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json +++ b/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json @@ -14,49 +14,19 @@ "description": "图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "image-erase-completion", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-08-19T01:17:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "图像擦除补全", "docUrl": "https://help.aliyun.com/document_detail/2840907.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'X-DashScope-DataInspection: enable' \\\n--data-raw '{\n \"model\": \"image-erase-completion\",\n \"input\": {\n \"image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E5%8E%9F%E5%9B%BE.png\",\n \"mask_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E6%93%A6%E9%99%A4.png\",\n \"foreground_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E4%BF%9D%E7%95%99.png\"\n },\n \"parameters\":{\n \"dilate_flag\":true\n }\n}' \n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'X-DashScope-DataInspection: enable' \\\n--data-raw '{\n \"model\": \"image-erase-completion\",\n \"input\": {\n \"image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E5%8E%9F%E5%9B%BE.png\",\n \"mask_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E6%93%A6%E9%99%A4.png\",\n \"foreground_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E4%BF%9D%E7%95%99.png\"\n },\n \"parameters\":{\n \"dilate_flag\":true\n }\n}' \n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json b/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json index e90a0e3b..6b98da94 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json +++ b/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json @@ -14,45 +14,15 @@ "description": "人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "image-instance-segmentation", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-08-19T01:17:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "人物实例分割", "docUrl": "https://help.aliyun.com/document_detail/2840906.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json b/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json index e6182d40..82349a0d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json +++ b/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json @@ -14,49 +14,25 @@ "description": "图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "image-out-painting", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-05-24T10:29:57.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "图像画面扩展", "docUrl": "https://help.aliyun.com/document_detail/2796845.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/out-painting' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"image-out-painting\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\"\n },\n \"parameters\":{\n \"x_scale\":2,\n \"y_scale\":2,\n \"best_quality\":false,\n \"limit_image_size\":true\n }\n}'" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/out-painting' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"image-out-painting\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\"\n },\n \"parameters\":{\n \"x_scale\":2,\n \"y_scale\":2,\n \"best_quality\":false,\n \"limit_image_size\":true\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json index 1646ab45..f5c19ff8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json @@ -2,6 +2,91 @@ "name": "Kimi", "description": "由月之暗面提供的Kimi系列模型的API服务。", "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Kimi K3 是 Kimi 迄今能力最强的旗舰模型，拥有 2.8 万亿参数，基于 KDA 混合线性注意力机制（Kimi Delta Attention）和注意力残差（Attention Residuals）技术构建，原生支持视觉理解，并拥有 100 万 token 上下文窗口。它是全球首个开源的 3 万亿级别模型，面向长程编程、知识工作和推理等前沿智能场景而设计。", + "features": [ + "function-calling", + "structured-outputs", + "cache", + "prefix-completion" + ], + "provider": "moonshot-ai", + "model": "kimi/kimi-k3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "100", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "priceTimeBands": [], + "capabilities": [ + "TG", + "VU", + "Reasoning" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 1048576, + "latestOnlineAt": "2026-07-17T08:23:12.000+00:00", + "contextWindow": 1048576, + "maxInputTokens": 1048576, + "inferenceProvider": "moonshot-ai", + "name": "kimi/kimi-k3", + "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", + "category": "Third-party", + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k3\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k3\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k3\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html" + } + } + } + }, { "inferenceMetadata": { "response_modality": [ @@ -21,10 +106,30 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi/kimi-k2.7-code-highspeed", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "13", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "54", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.6", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -43,6 +148,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning", @@ -57,33 +165,6 @@ "inferenceProvider": "moonshot-ai", "name": "kimi/kimi-k2.7-code-highspeed", "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -114,10 +195,30 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi/kimi-k2.7-code", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -136,6 +237,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "VU", @@ -150,33 +254,6 @@ "inferenceProvider": "moonshot-ai", "name": "kimi/kimi-k2.7-code", "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -207,10 +284,30 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi/kimi-k2.6", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.1", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -229,6 +326,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning", @@ -244,19 +344,6 @@ "inferenceProvider": "moonshot-ai", "name": "Kimi/Kimi K2.6", "docUrl": "https://help.aliyun.com/document_detail/3021620.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { @@ -287,11 +374,31 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi/kimi-k2.5", "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -310,6 +417,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning", @@ -324,19 +434,6 @@ "inferenceProvider": "moonshot-ai", "name": "Kimi/Kimi K2.5", "docUrl": "https://help.aliyun.com/document_detail/3021620.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json index 4a04f994..74f3a828 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json @@ -15,11 +15,38 @@ "description": "智能分镜可读懂剧本场景流转，自动调度机位和景别。原生多模态框架支持音画一致性。打破时长限制，多镜头故事创作更自由。", "features": [], "provider": "kling", - "limit": { - "message": "model not exist" - }, "model": "kling/kling-v3-video-generation", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.8", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,6 +65,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -46,31 +76,10 @@ "latestOnlineAt": "2026-03-26T13:45:28.000+00:00", "inferenceProvider": "kling", "name": "Kling Video 3.0", - "predictConfig": [ - { - "name": "mode", - "key": "mode", - "default": "pro" - }, - { - "name": "audio", - "key": "audio", - "default": false - }, - { - "name": "duration", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3026701.html" } } @@ -90,11 +99,52 @@ "description": "新增“全能参考”，支持3-8秒视频或多图锚定角色元素。可匹配原声及口型驱动，实现角色本色呈现。视频一致性更强，表现更灵动。支持音画同步、智能分镜。", "features": [], "provider": "kling", - "limit": { - "message": "model not exist" - }, "model": "kling/kling-v3-omni-video-generation", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "type": "720P_no_reference_video", + "priceName": "视频生成（720P 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "timeBand": "standard", + "type": "1080P_no_reference_video", + "priceName": "视频生成（1080P 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "type": "720P_no_audio_no_reference_video", + "priceName": "视频生成（720P 无声 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.8", + "timeBand": "standard", + "type": "1080P_no_audio_no_reference_video", + "priceName": "视频生成（1080P 无声 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "type": "720P_no_audio_reference_video", + "priceName": "视频生成（720P 无声 有参考视频）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "timeBand": "standard", + "type": "1080P_no_audio_reference_video", + "priceName": "视频生成（1080P 无声 有参考视频）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -113,6 +163,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -121,31 +174,10 @@ "latestOnlineAt": "2026-03-26T13:45:46.000+00:00", "inferenceProvider": "kling", "name": "Kling Video 3.0 Omni", - "predictConfig": [ - { - "name": "mode", - "key": "mode", - "default": "pro" - }, - { - "name": "audio", - "key": "audio", - "default": false - }, - { - "name": "duration", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-omni-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-omni-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3026701.html" } } @@ -164,11 +196,24 @@ "description": "支持最多10张参考图，可锁定主体、元素和色调，保证风格一致。融合风格转绘、人像/角色参考、多图融合及局部重绘，操作灵活。人像细节真实，整体画面细腻丰富，色彩氛围兼具影视感。", "features": [], "provider": "kling", - "limit": { - "message": "model not exist" - }, "model": "kling/kling-v3-image-generation", "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -187,6 +232,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -195,28 +243,10 @@ "latestOnlineAt": "2026-03-26T13:45:34.000+00:00", "inferenceProvider": "kling", "name": "Kling Image 3.0", - "predictConfig": [ - { - "name": "aspect_ratio", - "key": "aspect_ratio", - "default": "16:9" - }, - { - "name": "resolution", - "key": "resolution", - "default": "1k" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3026706.html" } } @@ -235,11 +265,31 @@ "description": "解锁影视级叙事画面，新增系列组图及2K/4K直出。深度解析提示词视听元素，精确响应创作指令。支持自由多参考图及全面效果升级，适合分镜、剧情概念图及场景设定。", "features": [], "provider": "kling", - "limit": { - "message": "model not exist" - }, "model": "kling/kling-v3-omni-image-generation", "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + }, + { + "priceUnit": "每秒", + "price": "0.4", + "timeBand": "standard", + "type": "image_type_4k", + "priceName": "图片生成（4K）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -258,6 +308,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -266,28 +319,10 @@ "latestOnlineAt": "2026-03-26T13:45:39.000+00:00", "inferenceProvider": "kling", "name": "Kling Image 3.0 Omni", - "predictConfig": [ - { - "name": "aspect_ratio", - "key": "aspect_ratio", - "default": "16:9" - }, - { - "name": "resolution", - "key": "resolution", - "default": "1k" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-omni-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-omni-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3026706.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json index a4387422..8caed608 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json +++ b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json @@ -12,49 +12,37 @@ "description": "LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "liveportrait-detect", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "timeBand": "standard", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-11-07T14:17:01.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "灵动人像LivePortrait-detect", "docUrl": "https://help.aliyun.com/document_detail/2856727.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"liveportrait-detect\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\"\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"liveportrait-detect\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\"\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json index 62fa8b89..cdb9e994 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json +++ b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json @@ -14,49 +14,37 @@ "description": "LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "liveportrait", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.02", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-11-07T14:17:00.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "灵动人像LivePortrait", "docUrl": "https://help.aliyun.com/document_detail/2856730.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"liveportrait\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/mbeygv/%E7%B4%A0%E6%8F%8F%E7%94%B7%E5%AD%A9.mp3\"\n },\n \"parameters\": {\n \"template_id\": \"normal\",\n \"eye_move_freq\": 0.5,\n \"video_fps\":30,\n \"mouth_move_strength\":1,\n \"paste_back\": true,\n \"head_move_strength\":0.7\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"liveportrait\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/mbeygv/%E7%B4%A0%E6%8F%8F%E7%94%B7%E5%AD%A9.mp3\"\n },\n \"parameters\": {\n \"template_id\": \"normal\",\n \"eye_move_freq\": 0.5,\n \"video_fps\":30,\n \"mouth_move_strength\":1,\n \"paste_back\": true,\n \"head_move_strength\":0.7\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json index 43ee7b5c..ce841311 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json @@ -19,10 +19,30 @@ "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/MiniMax-M3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16.8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.84", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -41,6 +61,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning", @@ -54,33 +77,6 @@ "inferenceProvider": "mini-max", "name": "MiniMax/MiniMax-M3", "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -107,10 +103,30 @@ "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/MiniMax-M2.7", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.42", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -129,6 +145,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -143,37 +162,10 @@ "inferenceProvider": "mini-max", "name": "MiniMax/MiniMax-M2.7", "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.7\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.7\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3021647.html" } } @@ -194,11 +186,31 @@ "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/MiniMax-M2.5", "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.21", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -217,6 +229,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -230,37 +245,10 @@ "inferenceProvider": "mini-max", "name": "MiniMax/MiniMax-M2.5", "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3021647.html" } } @@ -281,11 +269,31 @@ "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/MiniMax-M2.1", "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.21", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -304,6 +312,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -317,37 +328,10 @@ "inferenceProvider": "mini-max", "name": "MiniMax/MiniMax-M2.1", "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3021647.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json index e3116a64..f6a5cb21 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json @@ -16,10 +16,19 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-8k-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], @@ -28,19 +37,12 @@ "offlineInfo": { "inference": { "announceUrl": "mhttps://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-8k-v1", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json index 13d9ccb2..66eeb163 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json @@ -14,45 +14,27 @@ "description": "Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-8k-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-19T11:29:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-8k-v2", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json index cd175230..7e92a47b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json @@ -16,10 +16,19 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-mtl-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], @@ -28,19 +37,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-mtl-v1", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json index b99160b0..a973e02c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json @@ -14,10 +14,19 @@ "description": "Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-realtime-8k-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], @@ -26,19 +35,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer实时语音识别-8k-v1", "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json index a3ab4c5e..33683761 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json @@ -16,19 +16,27 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-realtime-8k-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-31T08:01:43.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer实时语音识别-8k-v2", "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json index 1d4292c4..7b803b38 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json @@ -16,10 +16,19 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-realtime-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], @@ -28,19 +37,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer实时语音识别-v1", "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json index 4b95da69..b30e2e67 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json @@ -16,25 +16,27 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-realtime-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-05T16:00:06.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer实时语音识别-v2", "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json index ea6b8ea1..095e5d0b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json @@ -16,10 +16,19 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], @@ -28,19 +37,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-v1", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json index d0e2040e..08d76e2c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json @@ -16,25 +16,27 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-05T16:00:06.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-v2", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json index f484629a..73c22ab7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json @@ -15,11 +15,66 @@ "description": "C1是PixVerse在26年3月底推出的影视行业大模型，r2v（多主体参考生成视频）输入2-7张图像，智能融合不同主体，同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力和想象力、更接近影视专业水准的打斗动作和术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合多主体群像、多人对话、多人交互等复杂剧情，适合中景、全景镜头。\n如果输入了1张多宫格分镜图片（最高支持九宫格），则可以一键生成连续分镜长视频。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-c1-r2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,6 +93,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -47,37 +105,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-C1-r2v", "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025612.html" } } @@ -95,11 +126,66 @@ "description": "C1是PixVerse在26年3月底推出的影视行业大模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-c1-t2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -118,6 +204,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -127,41 +216,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-C1-t2v", "docUrl": "https://help.aliyun.com/document_detail/3025608.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025608.html" } } @@ -180,11 +238,66 @@ "description": "C1是PixVerse在26年3月底推出的影视行业大模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-c1-kf2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -203,6 +316,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -212,37 +328,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-C1-kf2v", "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025611.html" } } @@ -261,11 +350,66 @@ "description": "C1是PixVerse在26年3月底推出的影视行业大模型，it2v（图片生成视频）模型除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征。相比V6可增强提示词，拥有更强的想象力、更接近影视专业水准的打斗动作、术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合单人特写、单人独白、定格/慢动作、空镜转场等短时长镜头。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-c1-it2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -282,6 +426,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -291,41 +438,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-C1-it2v", "docUrl": "https://help.aliyun.com/document_detail/3025609.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025609.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json index 9c8f461b..047a5a2d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json @@ -19,6 +19,7 @@ { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "video_generation", "priceName": "视频生成" } @@ -39,6 +40,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -48,33 +52,6 @@ "inferenceProvider": "pixverse", "name": "pixverse/pixverse-upscale", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -102,18 +79,21 @@ { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.36", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" } @@ -134,6 +114,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -143,33 +126,6 @@ "inferenceProvider": "pixverse", "name": "pixverse/pixverse-motioncontrol", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -199,6 +155,7 @@ { "priceUnit": "每秒", "price": "0.12", + "timeBand": "standard", "type": "video_generation", "priceName": "视频生成" } @@ -219,6 +176,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -228,33 +188,6 @@ "inferenceProvider": "pixverse", "name": "pixverse/pixverse-lipsync", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json index 02f27899..27c33ec6 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json @@ -14,11 +14,66 @@ "description": "输入文字描述，秒级生成与语义精准匹配的高质量视频，支持多种风格。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v5.6-t2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -35,6 +90,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -43,37 +101,10 @@ "latestOnlineAt": "2026-03-19T06:58:48.000+00:00", "inferenceProvider": "pixverse", "name": "PixVerse-V5.6-t2v", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025608.html" } } @@ -92,11 +123,66 @@ "description": "输入2–7张图像，智能融合不同主体，保持风格统一与动作协调，轻松构建丰富叙事场景，提升内容可控性与创意自由度。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v5.6-r2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -113,6 +199,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -121,37 +210,10 @@ "latestOnlineAt": "2026-03-19T06:43:04.000+00:00", "inferenceProvider": "pixverse", "name": "PixVerse-V5.6-r2v", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025612.html" } } @@ -170,11 +232,66 @@ "description": "在任意两张图片之间实现无缝转换，实现更流畅自然的场景过渡，打造视觉冲击力强的画面效果。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v5.6-kf2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -191,6 +308,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -199,37 +319,10 @@ "latestOnlineAt": "2026-03-19T06:56:40.000+00:00", "inferenceProvider": "pixverse", "name": "PixVerse-V5.6-kf2v", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025611.html" } } @@ -248,11 +341,66 @@ "description": "上传任意图片，自由定制剧情、节奏与风格，生成生动连贯的视频。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v5.6-it2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -269,6 +417,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -277,37 +428,10 @@ "latestOnlineAt": "2026-03-19T06:59:33.000+00:00", "inferenceProvider": "pixverse", "name": "PixVerse-V5.6-it2v", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025609.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json index 0740d773..300923d9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json @@ -14,11 +14,66 @@ "description": "V6是PixVerse在26年3月底推出的新模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v6-t2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -37,6 +92,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -46,46 +104,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-V6-t2v", "docUrl": "○ https://help.aliyun.com/document_detail/3025608.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - }, - { - "name": "shot_type", - "key": "shot_type", - "default": "single" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025608.html" } } @@ -104,10 +126,65 @@ "description": "V6是PixVerse在26年3月底推出的新模型，r2v（多主体参考生成视频）模型全球排名第二，输入2-7张图像，智能融合不同主体，适合复杂的中景、远景视频镜头。同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力、更强的情绪表现力、和更流畅的高速运动画面。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v6-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -126,6 +203,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -134,37 +214,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-V6-r2v", "docUrl": "https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025612.html" } } @@ -183,11 +236,66 @@ "description": "V6是PixVerse在26年3月底推出的新模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v6-kf2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -206,6 +314,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -215,37 +326,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-V6-kf2v", "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025611.html" } } @@ -264,11 +348,66 @@ "description": "V6是PixVerse在26年3月底推出的新模型，it2v（图片生成视频）模型全球排名第二，it2v除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征，拥有更强的人物情绪、高速运动表现力。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v6-it2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "timeBand": "standard", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "timeBand": "standard", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -287,6 +426,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -296,46 +438,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-V6-it2v", "docUrl": "https://help.aliyun.com/document_detail/3025609.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - }, - { - "name": "shot_type", - "key": "shot_type", - "default": "single" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025609.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json index 121259d4..6d822276 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json @@ -19,10 +19,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qvq-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -41,6 +54,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU" @@ -52,54 +68,27 @@ "maxInputTokens": 106496, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", + "announceUrl": "https://www.aliyun.com/notice/118177", "offlineTime": "2026-07-13 23:59:59" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "QVQ-Max", "docUrl": "https://help.aliyun.com/document_detail/2877996.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-max\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-max',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-max\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-max',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-max\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-max\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-max\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-max\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json index 1273bf44..ce0876d0 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json @@ -19,10 +19,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qvq-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -41,6 +54,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU" @@ -52,56 +68,27 @@ "maxInputTokens": 106496, "offlineInfo": { "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, + "announceUrl": "https://www.aliyun.com/notice/118177", "offlineTime": "2026-07-13 23:59:59" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "QVQ-Plus", "docUrl": "https://help.aliyun.com/document_detail/2877996.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-plus\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-plus',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-plus\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-plus',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-plus\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-plus\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-plus\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-plus\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json index 52c4ecf2..26689f0d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json @@ -23,24 +23,28 @@ { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "audio_text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "type": "audio_text_output_token", "priceName": "输出：文本" }, { "priceUnit": "每百万tokens", "price": "100", + "timeBand": "standard", "type": "audio_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -63,46 +67,22 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Chatting" ], "modelAlias": "", "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-07-14T06:59:47.013+00:00", - "contextWindow": 8192, - "maxInputTokens": 4096, + "maxOutputTokens": 8192, + "latestOnlineAt": "2026-07-14T06:59:47.000+00:00", + "contextWindow": 40960, + "maxInputTokens": 16384, "inferenceProvider": "aliyun-bailian", "name": "千问实时语音对话大模型3.0（极速版）", "docUrl": "https://help.aliyun.com/document_detail/3041584.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json index e02c2e41..0b6570bc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json @@ -23,24 +23,28 @@ { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "audio_text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "audio_text_output_token", "priceName": "输出：文本" }, { "priceUnit": "每百万tokens", "price": "150", + "timeBand": "standard", "type": "audio_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -63,46 +67,22 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Chatting" ], "modelAlias": "", "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-07-14T06:59:43.526+00:00", - "contextWindow": 8192, - "maxInputTokens": 4096, + "maxOutputTokens": 8192, + "latestOnlineAt": "2026-07-14T06:59:43.000+00:00", + "contextWindow": 40960, + "maxInputTokens": 16384, "inferenceProvider": "aliyun-bailian", "name": "千问实时语音大模型 （标准版）", "docUrl": "https://help.aliyun.com/document_detail/3041584.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json index f354c773..017170a2 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json @@ -19,6 +19,7 @@ { "priceUnit": "每万字符", "price": "1.4", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -35,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Text-to-Speech" ], @@ -45,37 +49,10 @@ "name": "qwen-audio-3.0-tts-plus", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-plus\"\nvoice = \"longanfengyue\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-plus\"\n\n#Please enter the correct voice below.\nvoice = \"\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", "docUrl": "https://help.aliyun.com/document_detail/2938790.html" } } @@ -98,6 +75,7 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -114,6 +92,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Text-to-Speech" ], @@ -124,37 +105,10 @@ "name": "qwen-audio-3.0-tts-flash", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-flash\"\nvoice = \"longanfengyue\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-flash\"\n\n#Please enter the correct voice below.\nvoice = \"\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", "docUrl": "https://help.aliyun.com/document_detail/2938790.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json index 11d030a0..f232cd2b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json @@ -17,10 +17,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-coder-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,6 +52,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -49,57 +65,28 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Coder-Plus", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json index 9d331b94..8515fe9f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json @@ -17,10 +17,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-coder-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,6 +52,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -49,57 +65,28 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Coder-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json index 419b305c..47c7c74c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json @@ -14,10 +14,23 @@ "description": "千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-deep-research", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "54", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "163", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -36,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -44,50 +60,23 @@ "latestOnlineAt": "2025-08-22T14:05:23.000+00:00", "contextWindow": 1000000, "maxInputTokens": 997952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "qwen-deep-research", "docUrl": "https://help.aliyun.com/document_detail/2975991.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-deep-research\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-deep-research\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-deep-research\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-deep-research\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-deep-research\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-deep-research\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-deep-research\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-deep-research\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-deep-research\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-deep-research\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-deep-research\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-deep-research\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json index a5b2499e..f298bbd1 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json @@ -16,10 +16,44 @@ "cache" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-doc-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -38,6 +72,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -46,50 +83,23 @@ "latestOnlineAt": "2025-07-23T13:21:02.000+00:00", "contextWindow": 262144, "maxInputTokens": 253952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Doc-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2948885.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-doc-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-doc-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-doc-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-doc-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-doc-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-doc-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-doc-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-doc-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-doc-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-doc-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-doc-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-doc-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json index f4d5508c..a1d9bd71 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json @@ -2,6 +2,71 @@ "name": "Qwen-Embedding", "description": "基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。", "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "是通义实验室基于Qwen3.7训练的多语言文本统一向量模型，相较text-embedding-v4版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升20%；支持256~2560维用户自定义向量维度。", + "collectionTag": "qwen3.7", + "features": [ + "model-experience" + ], + "provider": "qwen", + "model": "qwen3.7-text-embedding", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 2400, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 2400, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "priceTimeBands": [], + "capabilities": [ + "TR" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-15T02:24:33.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 131072, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.7-通用文本向量", + "docUrl": "https://help.aliyun.com/document_detail/2842587.html", + "category": "Embeddings", + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"qwen3.7-text-embedding\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"qwen3.7-text-embedding\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"qwen3.7-text-embedding\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + }, { "inferenceMetadata": { "response_modality": [ @@ -16,10 +81,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-v4", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,25 +116,23 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-06-05T03:07:20.000+00:00", "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-v4", "docUrl": "https://help.aliyun.com/document_detail/2842587.html", "category": "Embeddings", - "predictConfig": [ - { - "name": "topK" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v4\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v4\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v4\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v4\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -73,10 +149,23 @@ "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-v3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -95,45 +184,21 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-07-12T09:44:51.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-v3", "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v3\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v3\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v3\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v3\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -150,10 +215,23 @@ "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.35", + "timeBand": "standard", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -172,45 +250,21 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T09:03:19.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-v2", "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v2\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v2\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -227,10 +281,23 @@ "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.35", + "timeBand": "standard", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -249,45 +316,21 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T09:02:12.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-v1", "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v1\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -304,49 +347,31 @@ "description": "通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-async-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T09:05:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-async-v2", "docUrl": "https://help.aliyun.com/document_detail/2712516.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-async-v2\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-async-v2\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -363,49 +388,31 @@ "description": "通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-async-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T09:04:41.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-async-v1", "docUrl": "https://help.aliyun.com/document_detail/2712516.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-async-v1\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-async-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json index 6a62a6d2..a09a14d8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json @@ -18,10 +18,30 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-flash-character", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.05", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -40,6 +60,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -48,50 +71,23 @@ "latestOnlineAt": "2026-01-13T04:04:02.000+00:00", "contextWindow": 8192, "maxInputTokens": 8000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Flash-Character", "docUrl": "https://help.aliyun.com/document_detail/2874763.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-flash-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-flash-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-flash-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-flash-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-flash-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-flash-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json index b9ac0286..76b3815f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json @@ -23,9 +23,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-flash", "qpmInfo": { "model-default-actual": { @@ -45,6 +42,227 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.075", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.188", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.015", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "TG" @@ -54,82 +272,23 @@ "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 997952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Flash", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json index 1ee17628..d9fc3077 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json @@ -17,11 +17,17 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-2.0-pro", "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -34,6 +40,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -41,40 +50,15 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen-image-2.0-pro-2026-04-22", "latestOnlineAt": "2026-04-22T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-2.0-Pro", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "2048*2048", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0-pro\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0-pro\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0-pro\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0-pro\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json index 2b28d8f3..a11b024f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json @@ -17,11 +17,17 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-2.0", "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -34,6 +40,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -41,40 +50,15 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen-image-2.0-2026-03-03", "latestOnlineAt": "2026-03-03T11:31:36.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-2.0", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "2048*2048", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-3.0-pro.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-3.0-pro.json new file mode 100644 index 00000000..f90bf284 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-3.0-pro.json @@ -0,0 +1,53 @@ +{ + "name": "Qwen-Image-3.0-Pro", + "description": "内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。\n细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。\n知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。\nQwen-Image-3.0-Pro 不只是在追求\"好看\"，更在追求**“好用”**——让图像生成真正成为可落地的生产力工具。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。\n细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。\n知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。\nQwen-Image-3.0-Pro 不只是在追求\"好看\"，更在追求“好用”——让图像生成真正成为可落地的生产力工具。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "model": "qwen-image-3.0-pro", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "count_limit": 1, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "count_limit": 1, + "type": "model-default" + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-20T14:02:23.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image-3.0-Pro", + "docUrl": "https://help.aliyun.com/document_detail/3047054.html", + "category": "Visual", + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-3.0-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"画面是一张竖幅户外人像摄影，整体从上到下呈现温暖的午后街景氛围。顶部左侧到上方大面积被深绿色藤蔓和橙色小花覆盖，花叶从建筑檐口自然垂落，受阳光照射的叶片呈黄绿色高光，阴影处则偏深绿，形成浓密而柔和的背景层次。左上至中上区域是一块深蓝色横向招牌，招牌表面较暗、略带磨砂质感，上面以白色哥特体大字写着 Il Messaggero，文字位于画面左侧偏上，部分被前景花叶轻微遮挡，字体高对比、带装饰性尖角和粗细变化。招牌下方是报刊亭或书报摊的玻璃展示窗，黑色金属框架将橱窗分隔成多个矩形区域，内部陈列着许多报纸、杂志和书刊封面，但大多因景深虚化和光线反射而难以辨读，形成浅色纸张与深色边框交错的背景纹理。画面右上方是强烈的逆光区域，阳光从街道尽头照入，背景建筑被虚化成米灰色块面，边缘柔和，呈现明显的浅景深效果。画面中部偏右是一名年轻成年女性的半身至膝上人像，她回头面向镜头微笑，身体略向右转，肩背朝向观者，姿态自然放松。她有长而浓密的黑色波浪卷发，发丝被逆光勾勒出金色轮廓光，发梢在右侧向外散开，显得轻盈蓬松。她肤色白皙，脸型柔和偏鹅蛋形，眉形细致，眼睛明亮，眼妆清透，睫毛明显，面部带有自然高光，唇部为柔和珊瑚红色，笑容露齿，表情亲切明朗。她佩戴小巧耳饰，身穿黑色细肩带露背连衣裙，面料颜色深黑、轮廓简洁，细肩带从肩部向背部延伸，背部线条清晰。画面下部偏左到中部，她双手抱着一束玫瑰花，花束体积较大，主要由橙色、杏色、粉色和浅桃色玫瑰组成，花瓣层层卷曲，边缘被阳光照亮，绿色叶片和长花茎从花束下方垂出，花束与黑色裙装形成鲜明色彩对比。右侧背景是一条被阳光照亮的城市街道，地面呈暖灰与金黄色调，远处建筑、街边设施和一个模糊的红色圆形交通标志位于右下远景，均因焦外虚化而只保留色块和轮廓。整张照片采用暖色胶片感处理，带有细腻颗粒、柔和对比和明显逆光边缘光，人物位于视觉焦点，背景报刊亭、花藤、街道和阳光共同营造出浪漫、明亮、都市漫步式的氛围。\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"prompt_extend\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3047054.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json index dfdfec60..3632a0f1 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json @@ -17,10 +17,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-edit-max", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -33,27 +39,29 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2026-01-15T12:28:13.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Edit-Max", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-max\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-max\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-max\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-max\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-max\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-max\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json index ec102b64..9e108b88 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json @@ -16,10 +16,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-edit-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -32,27 +38,29 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2025-10-30T09:10:49.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Edit-Plus", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-plus\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-plus\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-plus\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-plus\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -72,10 +80,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-edit", + "prices": [ + { + "priceUnit": "每张", + "price": "0.3", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -88,27 +102,29 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2025-09-21T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Edit", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], "samples": { "dashscope": { "default": { "curl": "curl --location 'https://{Domain}/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json index 5f460769..e1afb6dd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json @@ -16,10 +16,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-max", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -38,38 +44,28 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "equivalentSnapshot": "qwen-image-max-2025-12-30", "latestOnlineAt": "2025-12-30T07:06:12.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Max", "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-max\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" + "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-max\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json index 55add199..631846a7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json @@ -16,10 +16,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -42,37 +48,27 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T10:49:31.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Plus", "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" + "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" } } } @@ -91,10 +87,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.25", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -117,37 +119,27 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-08-13T12:58:52.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image", "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" + "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json index 27b295ec..ed59eb91 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json @@ -16,10 +16,37 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-long-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,6 +65,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -46,50 +76,29 @@ "latestOnlineAt": "2025-03-19T02:45:13.000+00:00", "contextWindow": 10000000, "maxInputTokens": 10000000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Long-Latest", "docUrl": "https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -109,10 +118,37 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-long", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -131,6 +167,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -139,50 +178,23 @@ "latestOnlineAt": "2024-05-20T14:57:26.000+00:00", "contextWindow": 10000000, "maxInputTokens": 10000000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Long", "docUrl": "https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json index 89c3ed99..3b8f37fd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json @@ -16,10 +16,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -46,50 +62,29 @@ "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", "contextWindow": 4096, "maxInputTokens": 3072, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Plus", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -109,10 +104,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-plus-0919", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -131,6 +139,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -140,50 +151,29 @@ "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", "contextWindow": 4096, "maxInputTokens": 3072, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Plus-2024-09-19", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0919\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0919\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0919\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0919\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0919\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0919\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0919\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0919\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0919\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0919\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0919\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0919\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -203,10 +193,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-plus-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -225,6 +228,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -233,50 +239,29 @@ "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", "contextWindow": 4096, "maxInputTokens": 3072, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Plus-Latest", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -296,10 +281,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-plus-0816", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -318,6 +316,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -327,50 +328,29 @@ "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", "contextWindow": 4096, "maxInputTokens": 3072, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Plus-2024-08-16", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0816\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0816\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0816\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0816\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0816\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0816\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0816\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0816\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0816\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0816\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0816\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0816\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json index fc73582a..c692a9bb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json @@ -17,10 +17,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,6 +52,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -49,57 +65,28 @@ "maxInputTokens": 3072, "offlineInfo": { "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, + "announceUrl": "https://www.aliyun.com/notice/118177", "offlineTime": "2026-07-13 23:59:59" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json index b8bd3cb9..1877fc13 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json @@ -22,10 +22,44 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "9.6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 15, @@ -44,6 +78,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -52,62 +89,24 @@ "latestOnlineAt": "2024-10-15T05:39:20.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Max", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-max\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-max\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-max\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-max\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-max\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-max\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-max\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-max\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-max\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-max\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json index 6c36cbab..525b1c58 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json @@ -16,10 +16,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.95", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -38,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -47,30 +63,22 @@ "latestOnlineAt": "2025-11-06T08:07:15.000+00:00", "contextWindow": 16384, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Flash", "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-flash\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-flash\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-flash\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-flash\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-flash\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-flash\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-flash\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-flash\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json index d6a8ceda..bd1d8856 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json @@ -14,10 +14,16 @@ "description": "专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.003", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -34,45 +40,26 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-08-22T09:53:46.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Image", "docUrl": "https://help.aliyun.com/document_detail/2977163.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-mt-image\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i2/O1CN01XsvEqj1fNlMqLNBHR_!!6000000003995-0-tps-5933-2930.jpg\",\n \"source_lang\": \"en\",\n \"target_lang\": \"ja\"\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-mt-image\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i2/O1CN01XsvEqj1fNlMqLNBHR_!!6000000003995-0-tps-5933-2930.jpg\",\n \"source_lang\": \"en\",\n \"target_lang\": \"ja\"\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json index edd7ca5f..5494447d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json @@ -16,10 +16,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-lite", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -38,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -46,30 +62,22 @@ "latestOnlineAt": "2025-11-19T11:49:54.000+00:00", "contextWindow": 16384, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Lite", "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-lite\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-lite\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-lite\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-lite\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-lite\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-lite\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-lite\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-lite\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json index bc356ae0..030de0bc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json @@ -16,10 +16,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -38,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -46,30 +62,22 @@ "latestOnlineAt": "2025-07-22T06:16:44.000+00:00", "contextWindow": 16384, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Plus", "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-plus\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-plus\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-plus\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-plus\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-plus\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-plus\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-plus\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-plus\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json index a9872aef..b40ee375 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json @@ -16,10 +16,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.95", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -38,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -46,30 +62,28 @@ "latestOnlineAt": "2025-07-22T06:16:54.000+00:00", "contextWindow": 16384, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-turbo\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-turbo\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-turbo\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-turbo\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-turbo\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-turbo\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-turbo\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-turbo\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json index ddd80c4a..de420aed 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json @@ -18,10 +18,51 @@ "description": "千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-omni-turbo-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "timeBand": "standard", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "timeBand": "standard", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "timeBand": "standard", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "timeBand": "standard", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -40,6 +81,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], @@ -48,41 +92,20 @@ "latestOnlineAt": "2025-05-08T11:50:42.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Omni-Turbo-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } @@ -104,10 +127,51 @@ "description": "千问全新多模态理解生成大模型实时版，此版本为动态更新版本。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-omni-turbo-realtime-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "timeBand": "standard", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "timeBand": "standard", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "timeBand": "standard", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "timeBand": "standard", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -126,6 +190,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], @@ -134,41 +201,20 @@ "latestOnlineAt": "2025-05-08T12:29:07.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Omni-Turbo-Realtime-Latest", "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime-latest'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime-latest\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime-latest'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime-latest\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json index d8bba37b..08fd9c4d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json @@ -21,10 +21,114 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-omni-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "timeBand": "standard", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "timeBand": "standard", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "timeBand": "standard", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "vision_input_token_cache", + "priceName": "输入：图片/视频（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "timeBand": "standard", + "type": "text_input_token_cache", + "priceName": "输入：文本（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "audio_input_token_cache", + "priceName": "输入：音频（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "type": "text_input_token_batch", + "priceName": "输入：文本（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.5", + "timeBand": "standard", + "type": "audio_input_token_batch", + "priceName": "输入：音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "vision_input_token_batch", + "priceName": "输入：图片/视频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "timeBand": "standard", + "type": "multi_output_token_batch", + "priceName": "输出：文本+音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.25", + "timeBand": "standard", + "type": "multiin_text_output_token_batch", + "priceName": "输出：文本（Batch File，输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "purein_text_output_token_batch", + "priceName": "输出：文本（Batch File，输入仅包含文本时）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -43,6 +147,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], @@ -51,42 +158,21 @@ "latestOnlineAt": "2025-02-14T14:56:44.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Omni-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } @@ -110,10 +196,51 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-omni-turbo-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "timeBand": "standard", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "timeBand": "standard", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "timeBand": "standard", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -132,6 +259,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], @@ -140,42 +270,21 @@ "latestOnlineAt": "2025-02-14T14:55:08.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Omni-Turbo-Latest", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo-latest\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo-latest\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo-latest\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo-latest\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json index fff3c32a..6bf713f9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json @@ -19,10 +19,27 @@ "structured-outputs" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus-character", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -41,6 +58,7 @@ "type": "model-default" } }, + "priceTimeBands": [], "capabilities": [ "TG" ], @@ -49,50 +67,23 @@ "latestOnlineAt": "2025-03-20T09:16:40.000+00:00", "contextWindow": 32768, "maxInputTokens": 32768, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus-Character", "docUrl": "https://help.aliyun.com/document_detail/2874763.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json index ffe268ea..5e2a329f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json @@ -23,9 +23,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus", "qpmInfo": { "model-default-actual": { @@ -45,6 +42,422 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "timeBand": "standard", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "timeBand": "standard", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "timeBand": "standard", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "timeBand": "standard", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.96", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.96", + "timeBand": "standard", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "timeBand": "standard", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "timeBand": "standard", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "timeBand": "standard", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "TG" @@ -54,82 +467,23 @@ "latestOnlineAt": "2025-06-23T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 997952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -155,9 +509,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus-latest", "qpmInfo": { "model-default-actual": { @@ -177,6 +528,200 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "timeBand": "standard", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "TG" @@ -186,82 +731,23 @@ "latestOnlineAt": "2025-07-30T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 995904, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus-Latest", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-latest\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus-latest',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-latest\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus-latest',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-latest\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-latest\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -281,10 +767,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus-1220", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -303,6 +802,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -312,50 +814,29 @@ "latestOnlineAt": "2024-12-26T12:24:51.000+00:00", "contextWindow": 131072, "maxInputTokens": 129024, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus-2024-12-20", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-1220\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-1220\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-1220\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-1220\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-1220\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-1220\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-1220\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-1220\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-1220\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-1220\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-1220\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-1220\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -375,10 +856,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus-0112", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -397,6 +891,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -406,50 +903,29 @@ "latestOnlineAt": "2025-01-15T11:28:32.000+00:00", "contextWindow": 131072, "maxInputTokens": 129024, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus-2025-01-12", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-0112\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-0112\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-0112\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-0112\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-0112\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-0112\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-0112\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-0112\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-0112\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-0112\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-0112\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-0112\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json index 6fa115a4..56edeefc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json @@ -16,10 +16,23 @@ "description": "Qwen3-VL-Rerank重排模型，它能够深入理解文本、图片、视频的丰富多模态信息。在初步检索获得结果后，Qwen3-VL-Rerank 能够运用其先进的跨模态关联能力，对候选项目进行智能化的二次排序，将最相关的结果置于显要位置。通用用于提升跨模态搜索的准确率、优化图搜和视频检索的精准度、辅助图像聚类的分组质量、以及实现复杂多模态信息的高效检索和精确打标。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-rerank", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "timeBand": "standard", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,46 +51,23 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2026-01-29T10:23:42.000+00:00", "maxInputTokens": 120000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-Rerank", "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-rerank\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-rerank\",\n \"input\": {\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html" } } } @@ -94,10 +84,16 @@ "description": "基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐开源Qwen3-Rerank系列模型", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-rerank", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 10, @@ -116,6 +112,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -124,40 +123,14 @@ "latestOnlineAt": "2025-10-21T08:21:26.000+00:00", "contextWindow": 30000, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "千问3-Rerank", "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-rerank\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-rerank\",\n \"input\": {\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html" } } } @@ -174,10 +147,16 @@ "description": "gte-rerank-v2是通义实验室研发的多语言文本统一排序模型，面向全球多个主流语种，提供高水平的文本排序服务。通常用于语义检索、RAG等场景，可以简单、有效地提升文本检索的效果。给定查询 (Query) 和一系列候选文本 (documents)，模型会根据与查询的语义相关性从高到低对候选文本进行排序。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "gte-rerank-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -196,6 +175,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -204,40 +186,14 @@ "latestOnlineAt": "2025-03-20T08:32:00.000+00:00", "contextWindow": 30000, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "深度文本重排序", "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"gte-rerank-v2\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"gte-rerank-v2\",\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ],\n \"top_n\": 5,\n \"return_documents\": true\n}'", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json index aa990dde..e5bfe739 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json @@ -14,10 +14,23 @@ "description": "Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成模型。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-tts-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -36,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Text-to-Speech" ], @@ -46,40 +62,13 @@ "maxInputTokens": 512, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-TTS-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -101,10 +90,23 @@ "description": "Qwen-TTS实时模型是通义实验室千问模型中语音合成利器，始终与最新快照版能力相同。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。本模型是动态更新版本，模型更新不会提前通知。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-tts-realtime-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -123,6 +125,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Text-to-Speech" ], @@ -133,40 +138,13 @@ "maxInputTokens": 512, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1934", - "offlineTime": "2026-07-06 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118332", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-TTS-Realtime-Latest", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json index a0f026ff..7d9a5c8e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json @@ -14,10 +14,23 @@ "description": "千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持流式输出。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-tts", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "qwen_tts_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -36,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -46,14 +62,13 @@ "maxInputTokens": 512, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-TTS", "docUrl": "https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -75,10 +90,23 @@ "description": "模型是动态更新版本，等同于最新版本快照模型，模型更新时不会提前通知。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-tts-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "qwen_tts_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -97,6 +125,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -107,14 +138,13 @@ "maxInputTokens": 512, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1934", - "offlineTime": "2026-07-06 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118332", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-TTS-Latest", "docUrl": "https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json index af0c0b42..647da1d4 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json @@ -20,10 +20,86 @@ "structured-outputs" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "timeBand": "standard", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "timeBand": "standard", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 15, @@ -42,6 +118,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -53,87 +132,28 @@ "maxInputTokens": 98304, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-turbo\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-turbo',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-turbo\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-turbo',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-turbo\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-turbo\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json index 21d91205..5c455f4c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json @@ -10,72 +10,65 @@ "Image" ] }, - "description": "基于Qwen2.5-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", - "features": [], + "description": "基于Qwen3-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", + "features": [ + "model-experience" + ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen2.5-vl-embedding", + "model": "qwen3-vl-embedding", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "timeBand": "standard", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, - "usage_limit": 60000, + "usage_limit": 120000, "usage_limit_field": "total_usage", - "count_limit": 20, + "count_limit": 40, "usage_limit_period": 6, "type": "model-default" }, "model-default": { "count_limit_period": 1, - "usage_limit": 60000, + "usage_limit": 120000, "usage_limit_field": "total_usage", - "count_limit": 20, + "count_limit": 40, "usage_limit_period": 6, "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], "versionTag": "MAJOR", - "latestOnlineAt": "2025-10-21T08:21:32.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen2.5-VL-Embedding", + "latestOnlineAt": "2026-01-21T03:39:38.000+00:00", + "maxInputTokens": 32000, + "offlineInfo": {}, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-Embedding", "docUrl": "https://help.aliyun.com/document_detail/2842587.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen2.5-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", - "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen2.5-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", - "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen2.5-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" + "curl": "curl --silent --location --request POST '[workspace-id].cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen3-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", + "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen3-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" } } } @@ -88,54 +81,61 @@ "Image" ] }, - "description": "基于Qwen3-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", - "features": [ - "model-experience" - ], + "description": "基于Qwen2.5-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", + "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-embedding", + "model": "qwen2.5-vl-embedding", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "timeBand": "standard", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, - "usage_limit": 120000, + "usage_limit": 60000, "usage_limit_field": "total_usage", - "count_limit": 40, + "count_limit": 20, "usage_limit_period": 6, "type": "model-default" }, "model-default": { "count_limit_period": 1, - "usage_limit": 120000, + "usage_limit": 60000, "usage_limit_field": "total_usage", - "count_limit": 40, + "count_limit": 20, "usage_limit_period": 6, "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], "versionTag": "MAJOR", - "latestOnlineAt": "2026-01-21T03:39:38.000+00:00", - "maxInputTokens": 32000, - "offlineInfo": {}, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-Embedding", + "latestOnlineAt": "2025-10-21T08:21:32.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen2.5-VL-Embedding", "docUrl": "https://help.aliyun.com/document_detail/2842587.html", - "predictConfig": [ - { - "name": "topK" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", - "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen3-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", - "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen3-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" + "curl": "curl --silent --location --request POST '[workspace-id].cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen2.5-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen2.5-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", + "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen2.5-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json index e57df8de..2c8fb683 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json @@ -23,10 +23,51 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.32", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "150", + "timeBand": "standard", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -45,6 +86,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -55,60 +99,27 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-VL-Max", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-max\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-max\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-max\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-max\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-max',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-max\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-max',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-max\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json index 14c496eb..4e74e9fc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json @@ -18,28 +18,58 @@ "batch" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-ocr-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, - "usage_limit": 600000, + "usage_limit": 3000000, "usage_limit_field": "total_tokens", - "count_limit": 20, + "count_limit": 100, "usage_limit_period": 6, "type": "model-default" }, "model-default": { "count_limit_period": 1, - "usage_limit": 600000, + "usage_limit": 3000000, "usage_limit_field": "total_tokens", - "count_limit": 20, + "count_limit": 100, "usage_limit_period": 6, "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -48,49 +78,28 @@ "latestOnlineAt": "2025-09-22T16:00:00.000+00:00", "contextWindow": 38192, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "QwenVL-OCR-Latest", "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-latest\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-latest\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-latest\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-latest\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-latest',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-latest\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-latest',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-latest\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -110,10 +119,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-ocr-1028", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -132,6 +154,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -141,49 +166,28 @@ "latestOnlineAt": "2024-11-14T13:51:21.000+00:00", "contextWindow": 34096, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "QwenVL-OCR-2024-10-28", "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-1028\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-1028\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-1028\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-1028\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-1028',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-1028\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-1028',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-1028\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -204,10 +208,37 @@ "batch" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-ocr", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -226,6 +257,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -235,49 +269,28 @@ "latestOnlineAt": "2025-11-19T16:00:00.000+00:00", "contextWindow": 38192, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "QwenVL-OCR", "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json index 9e9b6ac1..e673bbbb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json @@ -24,10 +24,51 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "timeBand": "standard", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -46,6 +87,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -56,66 +100,27 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "QwenVL-Plus", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "vl_high_resolution_images", - "key": "vl_high_resolution_images", - "default": false, - "tip": "是否提高输入图片的默认Token上限" - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-plus\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-plus\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-plus\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-plus\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json index 526e2ad2..543cd8ec 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json @@ -14,11 +14,17 @@ "description": "千问voice-design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-voice-design", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "0.2", + "timeBand": "standard", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -31,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -39,26 +48,19 @@ "latestOnlineAt": "2025-12-12T07:41:45.000+00:00", "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Voice-Design", "docUrl": "https://help.aliyun.com/document_detail/3000986.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n}'", - "python": "import requests\nimport base64\nimport os\n\ndef create_voice_and_play():\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n \n if not api_key:\n print(\"错误: 未找到DASHSCOPE_API_KEY环境变量，请先设置API Key\")\n return None, None, None\n \n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n \n data = {\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n }\n \n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n \n try:\n response = requests.post(\n url,\n headers=headers,\n json=data,\n timeout=60\n )\n \n if response.status_code == 200:\n result = response.json()\n \n voice_name = result[\"output\"][\"voice\"]\n print(f\"音色名称: {voice_name}\")\n \n base64_audio = result[\"output\"][\"preview_audio\"][\"data\"]\n \n audio_bytes = base64.b64decode(base64_audio)\n \n filename = f\"{voice_name}_preview.wav\"\n \n with open(filename, 'wb') as f:\n f.write(audio_bytes)\n \n print(f\"音频已保存到本地文件: {filename}\")\n print(f\"文件路径: {os.path.abspath(filename)}\")\n \n return voice_name, audio_bytes, filename\n else:\n print(f\"请求失败，状态码: {response.status_code}\")\n print(f\"响应内容: {response.text}\")\n return None, None, None\n \n except requests.exceptions.RequestException as e:\n print(f\"网络请求发生错误: {e}\")\n return None, None, None\n except KeyError as e:\n print(f\"响应数据格式错误，缺少必要的字段: {e}\")\n print(f\"响应内容: {response.text if 'response' in locals() else 'No response'}\")\n return None, None, None\n except Exception as e:\n print(f\"发生未知错误: {e}\")\n return None, None, None\n\nif __name__ == \"__main__\":\n voice_name, audio_data, saved_filename = create_voice_and_play()\n \n if voice_name:\n print(f\"\\n成功创建音色 '{voice_name}'\")\n print(f\"音频文件已保存: '{saved_filename}'\")\n print(f\"文件大小: {os.path.getsize(saved_filename)} 字节\")\n else:\n print(\"\\n音色创建失败\")", - "java": "import com.google.gson.JsonObject;\nimport com.google.gson.JsonParser;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.util.Base64;\n\npublic class Main {\n public static void main(String[] args) {\n Main example = new Main();\n example.createVoice();\n }\n\n public void createVoice() {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonBody = \"{\\n\" +\n \" \\\"model\\\": \\\"qwen-voice-design\\\",\\n\" +\n \" \\\"input\\\": {\\n\" +\n \" \\\"action\\\": \\\"create\\\",\\n\" +\n \" \\\"target_model\\\": \\\"qwen3-tts-vd-realtime-2025-12-16\\\",\\n\" +\n \" \\\"voice_prompt\\\": \\\"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\\\",\\n\" +\n \" \\\"preview_text\\\": \\\"各位听众朋友，大家好，欢迎收听晚间新闻。\\\",\\n\" +\n \" \\\"preferred_name\\\": \\\"announcer\\\",\\n\" +\n \" \\\"language\\\": \\\"zh\\\"\\n\" +\n \" },\\n\" +\n \" \\\"parameters\\\": {\\n\" +\n \" \\\"sample_rate\\\": 24000,\\n\" +\n \" \\\"response_format\\\": \\\"wav\\\"\\n\" +\n \" }\\n\" +\n \"}\";\n\n HttpURLConnection connection = null;\n try {\n URL url = new URL(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\");\n connection = (HttpURLConnection) url.openConnection();\n\n connection.setRequestMethod(\"POST\");\n connection.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n connection.setRequestProperty(\"Content-Type\", \"application/json\");\n connection.setDoOutput(true);\n connection.setDoInput(true);\n\n \n try (OutputStream os = connection.getOutputStream()) {\n byte[] input = jsonBody.getBytes(\"UTF-8\");\n os.write(input, 0, input.length);\n os.flush();\n }\n\n \n int responseCode = connection.getResponseCode();\n if (responseCode == HttpURLConnection.HTTP_OK) {\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getInputStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n response.append(responseLine.trim());\n }\n }\n\n \n JsonObject jsonResponse = JsonParser.parseString(response.toString()).getAsJsonObject();\n JsonObject outputObj = jsonResponse.getAsJsonObject(\"output\");\n JsonObject previewAudioObj = outputObj.getAsJsonObject(\"preview_audio\");\n\n \n String voiceName = outputObj.get(\"voice\").getAsString();\n System.out.println(\"音色名称: \" + voiceName);\n\n \n String base64Audio = previewAudioObj.get(\"data\").getAsString();\n\n \n byte[] audioBytes = Base64.getDecoder().decode(base64Audio);\n\n \n String filename = voiceName + \"_preview.wav\";\n saveAudioToFile(audioBytes, filename);\n\n System.out.println(\"音频已保存到本地文件: \" + filename);\n\n } else {\n StringBuilder errorResponse = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getErrorStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n errorResponse.append(responseLine.trim());\n }\n }\n\n System.out.println(\"请求失败，状态码: \" + responseCode);\n System.out.println(\"错误响应: \" + errorResponse.toString());\n }\n\n } catch (Exception e) {\n System.err.println(\"请求发生错误: \" + e.getMessage());\n e.printStackTrace();\n } finally {\n if (connection != null) {\n connection.disconnect();\n }\n }\n }\n\n private void saveAudioToFile(byte[] audioBytes, String filename) {\n try {\n File file = new File(filename);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audioBytes);\n }\n System.out.println(\"音频已保存到: \" + file.getAbsolutePath());\n } catch (IOException e) {\n System.err.println(\"保存音频文件时发生错误: \" + e.getMessage());\n e.printStackTrace();\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n}'", + "python": "import requests\nimport base64\nimport os\n\ndef create_voice_and_play():\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n \n if not api_key:\n print(\"错误: 未找到DASHSCOPE_API_KEY环境变量，请先设置API Key\")\n return None, None, None\n \n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n \n data = {\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n }\n \n url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n \n try:\n response = requests.post(\n url,\n headers=headers,\n json=data,\n timeout=60\n )\n \n if response.status_code == 200:\n result = response.json()\n \n voice_name = result[\"output\"][\"voice\"]\n print(f\"音色名称: {voice_name}\")\n \n base64_audio = result[\"output\"][\"preview_audio\"][\"data\"]\n \n audio_bytes = base64.b64decode(base64_audio)\n \n filename = f\"{voice_name}_preview.wav\"\n \n with open(filename, 'wb') as f:\n f.write(audio_bytes)\n \n print(f\"音频已保存到本地文件: {filename}\")\n print(f\"文件路径: {os.path.abspath(filename)}\")\n \n return voice_name, audio_bytes, filename\n else:\n print(f\"请求失败，状态码: {response.status_code}\")\n print(f\"响应内容: {response.text}\")\n return None, None, None\n \n except requests.exceptions.RequestException as e:\n print(f\"网络请求发生错误: {e}\")\n return None, None, None\n except KeyError as e:\n print(f\"响应数据格式错误，缺少必要的字段: {e}\")\n print(f\"响应内容: {response.text if 'response' in locals() else 'No response'}\")\n return None, None, None\n except Exception as e:\n print(f\"发生未知错误: {e}\")\n return None, None, None\n\nif __name__ == \"__main__\":\n voice_name, audio_data, saved_filename = create_voice_and_play()\n \n if voice_name:\n print(f\"\\n成功创建音色 '{voice_name}'\")\n print(f\"音频文件已保存: '{saved_filename}'\")\n print(f\"文件大小: {os.path.getsize(saved_filename)} 字节\")\n else:\n print(\"\\n音色创建失败\")", + "java": "import com.google.gson.JsonObject;\nimport com.google.gson.JsonParser;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.util.Base64;\n\npublic class Main {\n public static void main(String[] args) {\n Main example = new Main();\n example.createVoice();\n }\n\n public void createVoice() {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonBody = \"{\\n\" +\n \" \\\"model\\\": \\\"qwen-voice-design\\\",\\n\" +\n \" \\\"input\\\": {\\n\" +\n \" \\\"action\\\": \\\"create\\\",\\n\" +\n \" \\\"target_model\\\": \\\"qwen3-tts-vd-realtime-2025-12-16\\\",\\n\" +\n \" \\\"voice_prompt\\\": \\\"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\\\",\\n\" +\n \" \\\"preview_text\\\": \\\"各位听众朋友，大家好，欢迎收听晚间新闻。\\\",\\n\" +\n \" \\\"preferred_name\\\": \\\"announcer\\\",\\n\" +\n \" \\\"language\\\": \\\"zh\\\"\\n\" +\n \" },\\n\" +\n \" \\\"parameters\\\": {\\n\" +\n \" \\\"sample_rate\\\": 24000,\\n\" +\n \" \\\"response_format\\\": \\\"wav\\\"\\n\" +\n \" }\\n\" +\n \"}\";\n\n HttpURLConnection connection = null;\n try {\n URL url = new URL(\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\");\n connection = (HttpURLConnection) url.openConnection();\n\n connection.setRequestMethod(\"POST\");\n connection.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n connection.setRequestProperty(\"Content-Type\", \"application/json\");\n connection.setDoOutput(true);\n connection.setDoInput(true);\n\n \n try (OutputStream os = connection.getOutputStream()) {\n byte[] input = jsonBody.getBytes(\"UTF-8\");\n os.write(input, 0, input.length);\n os.flush();\n }\n\n \n int responseCode = connection.getResponseCode();\n if (responseCode == HttpURLConnection.HTTP_OK) {\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getInputStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n response.append(responseLine.trim());\n }\n }\n\n \n JsonObject jsonResponse = JsonParser.parseString(response.toString()).getAsJsonObject();\n JsonObject outputObj = jsonResponse.getAsJsonObject(\"output\");\n JsonObject previewAudioObj = outputObj.getAsJsonObject(\"preview_audio\");\n\n \n String voiceName = outputObj.get(\"voice\").getAsString();\n System.out.println(\"音色名称: \" + voiceName);\n\n \n String base64Audio = previewAudioObj.get(\"data\").getAsString();\n\n \n byte[] audioBytes = Base64.getDecoder().decode(base64Audio);\n\n \n String filename = voiceName + \"_preview.wav\";\n saveAudioToFile(audioBytes, filename);\n\n System.out.println(\"音频已保存到本地文件: \" + filename);\n\n } else {\n StringBuilder errorResponse = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getErrorStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n errorResponse.append(responseLine.trim());\n }\n }\n\n System.out.println(\"请求失败，状态码: \" + responseCode);\n System.out.println(\"错误响应: \" + errorResponse.toString());\n }\n\n } catch (Exception e) {\n System.err.println(\"请求发生错误: \" + e.getMessage());\n e.printStackTrace();\n } finally {\n if (connection != null) {\n connection.disconnect();\n }\n }\n }\n\n private void saveAudioToFile(byte[] audioBytes, String filename) {\n try {\n File file = new File(filename);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audioBytes);\n }\n System.out.println(\"音频已保存到: \" + file.getAbsolutePath());\n } catch (IOException e) {\n System.err.println(\"保存音频文件时发生错误: \" + e.getMessage());\n e.printStackTrace();\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json index b1ef359c..9dd791fe 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json @@ -14,10 +14,16 @@ "description": "千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-voice-enrollment", + "prices": [ + { + "priceUnit": "次", + "price": "0.01", + "timeBand": "standard", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -30,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -38,20 +47,19 @@ "latestOnlineAt": "2025-11-27T05:44:15.000+00:00", "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Voice-Enrollment", "docUrl": "https://help.aliyun.com/document_detail/2975034.html", - "predictConfig": [], "samples": { "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vc-realtime-2025-11-27\",\n \"preferred_name\": \"guanyu\",\n \"audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n}'", - "python": "import os\nimport requests\nimport base64, pathlib\n\ntarget_model = \"qwen3-tts-vc-realtime-2025-11-27\"\npreferred_name = \"guanyu\"\naudio_mime_type = \"audio/mpeg\"\n\nfile_path = pathlib.Path(\"input.mp3\")\nbase64_str = base64.b64encode(file_path.read_bytes()).decode()\ndata_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\nurl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n\npayload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\n \"data\": data_uri\n }\n }\n}\n\nheaders = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n}\n\nresp = requests.post(url, json=payload, headers=headers)\n\nif resp.status_code == 200:\n data = resp.json()\n voice = data[\"output\"][\"voice\"]\n print(f\"voice name is: {voice}\")\nelse:\n print(\"Failed: \", resp.status_code, resp.text)", - "java": "import com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.util.Base64;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2025-11-27\";\n private static final String PREFERRED_NAME = \"guanyu\";\n private static final String AUDIO_FILE = \"input.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n\n public static String toDataUrl(String filePath) throws Exception {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static void main(String[] args) {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n String apiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\";\n\n try {\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(apiUrl).openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(\"UTF-8\"));\n }\n\n int status = con.getResponseCode();\n InputStream is = (status >= 200 && status < 300)\n ? con.getInputStream()\n : con.getErrorStream();\n\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(new InputStreamReader(is, \"UTF-8\"))) {\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n }\n\n System.out.println(\"HTTP status: \" + status);\n System.out.println(\"Response is: \" + response.toString());\n\n if (status == 200) {\n Gson gson = new Gson();\n JsonObject jsonObj = gson.fromJson(response.toString(), JsonObject.class);\n String voice = jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n System.out.println(\"voice name is: \" + voice);\n }\n\n } catch (Exception e) {\n e.printStackTrace();\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vc-realtime-2025-11-27\",\n \"preferred_name\": \"guanyu\",\n \"audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n}'", + "python": "import os\nimport requests\nimport base64, pathlib\n\ntarget_model = \"qwen3-tts-vc-realtime-2025-11-27\"\npreferred_name = \"guanyu\"\naudio_mime_type = \"audio/mpeg\"\n\nfile_path = pathlib.Path(\"input.mp3\")\nbase64_str = base64.b64encode(file_path.read_bytes()).decode()\ndata_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\nurl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n\npayload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\n \"data\": data_uri\n }\n }\n}\n\nheaders = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n}\n\nresp = requests.post(url, json=payload, headers=headers)\n\nif resp.status_code == 200:\n data = resp.json()\n voice = data[\"output\"][\"voice\"]\n print(f\"voice name is: {voice}\")\nelse:\n print(\"Failed: \", resp.status_code, resp.text)", + "java": "import com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.util.Base64;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2025-11-27\";\n private static final String PREFERRED_NAME = \"guanyu\";\n private static final String AUDIO_FILE = \"input.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n\n public static String toDataUrl(String filePath) throws Exception {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static void main(String[] args) {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n String apiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\";\n\n try {\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(apiUrl).openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(\"UTF-8\"));\n }\n\n int status = con.getResponseCode();\n InputStream is = (status >= 200 && status < 300)\n ? con.getInputStream()\n : con.getErrorStream();\n\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(new InputStreamReader(is, \"UTF-8\"))) {\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n }\n\n System.out.println(\"HTTP status: \" + status);\n System.out.println(\"Response is: \" + response.toString());\n\n if (status == 200) {\n Gson gson = new Gson();\n JsonObject jsonObj = gson.fromJson(response.toString(), JsonObject.class);\n String voice = jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n System.out.println(\"voice name is: \" + voice);\n }\n\n } catch (Exception e) {\n e.printStackTrace();\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json index cc1c7767..79028dbb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json @@ -20,10 +20,51 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen2.5-omni-7b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "38", + "timeBand": "standard", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "76", + "timeBand": "standard", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -42,6 +83,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], @@ -50,42 +94,21 @@ "latestOnlineAt": "2025-03-26T12:01:58.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen2.5-Omni-7B", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen2.5-omni-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen2.5-omni-7b\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen2.5-omni-7b\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen2.5-omni-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen2.5-omni-7b\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen2.5-omni-7b\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json index d6052ad9..0f99e445 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json @@ -15,10 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-asr-flash-filetrans", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -31,47 +37,23 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-11-17T13:12:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-ASR-Flash-Filetrans", "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\":[\n 0\n ],\n \"language\": \"zh\", \n \"enable_itn\": false, \n \"corpus\": {\n \"text\": \"张三，李四，王五\"\n }\n }\n}'\n\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json'", - "python": "import os\nimport time\nimport requests\nimport json\n\n\nAPI_URL_SUBMIT = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\"\nAPI_URL_QUERY_BASE = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/\"\n\n\ndef main():\n # If no environment variable is configured, please replace the downlink with the Bailian API Key: api_key = \"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\",\n \"X-DashScope-Async\": \"enable\"\n }\n\n\n payload = {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n # \"language\": \"zh\",\n \"enable_itn\": False\n # \"corpus\": {\n # \"text\": \"\"\n # }\n }\n }\n\n\n try:\n submit_resp = requests.post(API_URL_SUBMIT, headers=headers, data=json.dumps(payload))\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if submit_resp.status_code != 200:\n print(f\"Failed! HTTP code: {submit_resp.status_code}\")\n print(submit_resp.text)\n return\n\n resp_data = submit_resp.json()\n output = resp_data.get(\"output\")\n if not output or \"task_id\" not in output:\n print(\"resp_data:\", resp_data)\n return\n\n task_id = output[\"task_id\"]\n print(f\"任务已提交，task_id: {task_id}\")\n\n\n finished = False\n while not finished:\n time.sleep(2)\n\n query_url = API_URL_QUERY_BASE + task_id\n try:\n query_resp = requests.get(query_url, headers=headers)\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if query_resp.status_code != 200:\n print(f\"Failed! HTTP code: {query_resp.status_code}\")\n print(query_resp.text)\n return\n\n query_data = query_resp.json()\n output = query_data.get(\"output\")\n if output and \"task_status\" in output:\n status = output[\"task_status\"]\n print(f\"status: {status}\")\n\n if status.upper() in (\"SUCCEEDED\", \"FAILED\", \"UNKNOWN\"):\n finished = True\n print(\"task finished：\")\n print(json.dumps(query_data, indent=2, ensure_ascii=False))\n else:\n print(\"query data:\", query_data)\n\n\nif __name__ == \"__main__\":\n main()", - "java": "import com.google.gson.Gson;\nimport com.google.gson.annotations.SerializedName;\nimport okhttp3.*;\n\nimport java.io.IOException;\nimport java.util.concurrent.TimeUnit;\n\npublic class Main {\n private static final String API_URL_SUBMIT = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\";\n private static final String API_URL_QUERY = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/\";\n private static final Gson gson = new Gson();\n\n public static void main(String[] args) {\n // If no environment variable is configured, please replace the downlink with the Bailian API Key: String apiKey = \"sk-xxx\"\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n OkHttpClient client = new OkHttpClient();\n\n String payloadJson = \"\"\"\n {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n \"enable_itn\": false\n }\n }\n \"\"\";\n\n RequestBody body = RequestBody.create(payloadJson, MediaType.get(\"application/json; charset=utf-8\"));\n Request submitRequest = new Request.Builder()\n .url(API_URL_SUBMIT)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"Content-Type\", \"application/json\")\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .post(body)\n .build();\n\n String taskId = null;\n\n try (Response response = client.newCall(submitRequest).execute()) {\n if (response.isSuccessful() && response.body() != null) {\n String respBody = response.body().string();\n ApiResponse apiResp = gson.fromJson(respBody, ApiResponse.class);\n if (apiResp.output != null) {\n taskId = apiResp.output.taskId;\n System.out.println(\"task_id: \" + taskId);\n } else {\n System.out.println(\"respBody: \" + respBody);\n return;\n }\n } else {\n System.out.println(\"Failed! HTTP code: \" + response.code());\n if (response.body() != null) {\n System.out.println(response.body().string());\n }\n return;\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n\n boolean finished = false;\n while (!finished) {\n try {\n TimeUnit.SECONDS.sleep(2);\n } catch (InterruptedException e) {\n Thread.currentThread().interrupt();\n return;\n }\n\n String queryUrl = API_URL_QUERY + taskId;\n Request queryRequest = new Request.Builder()\n .url(queryUrl)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .addHeader(\"Content-Type\", \"application/json\")\n .get()\n .build();\n\n try (Response response = client.newCall(queryRequest).execute()) {\n if (response.body() != null) {\n String queryResponse = response.body().string();\n ApiResponse apiResp = gson.fromJson(queryResponse, ApiResponse.class);\n\n if (apiResp.output != null && apiResp.output.taskStatus != null) {\n String status = apiResp.output.taskStatus;\n System.out.println(\"task status: \" + status);\n if (\"SUCCEEDED\".equalsIgnoreCase(status)\n || \"FAILED\".equalsIgnoreCase(status)\n || \"UNKNOWN\".equalsIgnoreCase(status)) {\n finished = true;\n System.out.println(\"task finished: \");\n System.out.println(queryResponse);\n }\n } else {\n System.out.println(\"query response: \" + queryResponse);\n }\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n }\n }\n\n static class ApiResponse {\n @SerializedName(\"request_id\")\n String requestId;\n Output output;\n }\n\n static class Output {\n @SerializedName(\"task_id\")\n String taskId;\n @SerializedName(\"task_status\")\n String taskStatus;\n }\n}" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\":[\n 0\n ],\n \"language\": \"zh\", \n \"enable_itn\": false, \n \"corpus\": {\n \"text\": \"张三，李四，王五\"\n }\n }\n}'\n\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json'", + "python": "import os\nimport time\nimport requests\nimport json\n\n\nAPI_URL_SUBMIT = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\"\nAPI_URL_QUERY_BASE = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/\"\n\n\ndef main():\n # If no environment variable is configured, please replace the downlink with the Bailian API Key: api_key = \"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\",\n \"X-DashScope-Async\": \"enable\"\n }\n\n\n payload = {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n # \"language\": \"zh\",\n \"enable_itn\": False\n # \"corpus\": {\n # \"text\": \"\"\n # }\n }\n }\n\n\n try:\n submit_resp = requests.post(API_URL_SUBMIT, headers=headers, data=json.dumps(payload))\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if submit_resp.status_code != 200:\n print(f\"Failed! HTTP code: {submit_resp.status_code}\")\n print(submit_resp.text)\n return\n\n resp_data = submit_resp.json()\n output = resp_data.get(\"output\")\n if not output or \"task_id\" not in output:\n print(\"resp_data:\", resp_data)\n return\n\n task_id = output[\"task_id\"]\n print(f\"任务已提交，task_id: {task_id}\")\n\n\n finished = False\n while not finished:\n time.sleep(2)\n\n query_url = API_URL_QUERY_BASE + task_id\n try:\n query_resp = requests.get(query_url, headers=headers)\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if query_resp.status_code != 200:\n print(f\"Failed! HTTP code: {query_resp.status_code}\")\n print(query_resp.text)\n return\n\n query_data = query_resp.json()\n output = query_data.get(\"output\")\n if output and \"task_status\" in output:\n status = output[\"task_status\"]\n print(f\"status: {status}\")\n\n if status.upper() in (\"SUCCEEDED\", \"FAILED\", \"UNKNOWN\"):\n finished = True\n print(\"task finished：\")\n print(json.dumps(query_data, indent=2, ensure_ascii=False))\n else:\n print(\"query data:\", query_data)\n\n\nif __name__ == \"__main__\":\n main()", + "java": "import com.google.gson.Gson;\nimport com.google.gson.annotations.SerializedName;\nimport okhttp3.*;\n\nimport java.io.IOException;\nimport java.util.concurrent.TimeUnit;\n\npublic class Main {\n private static final String API_URL_SUBMIT = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\";\n private static final String API_URL_QUERY = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/\";\n private static final Gson gson = new Gson();\n\n public static void main(String[] args) {\n // If no environment variable is configured, please replace the downlink with the Bailian API Key: String apiKey = \"sk-xxx\"\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n OkHttpClient client = new OkHttpClient();\n\n String payloadJson = \"\"\"\n {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n \"enable_itn\": false\n }\n }\n \"\"\";\n\n RequestBody body = RequestBody.create(payloadJson, MediaType.get(\"application/json; charset=utf-8\"));\n Request submitRequest = new Request.Builder()\n .url(API_URL_SUBMIT)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"Content-Type\", \"application/json\")\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .post(body)\n .build();\n\n String taskId = null;\n\n try (Response response = client.newCall(submitRequest).execute()) {\n if (response.isSuccessful() && response.body() != null) {\n String respBody = response.body().string();\n ApiResponse apiResp = gson.fromJson(respBody, ApiResponse.class);\n if (apiResp.output != null) {\n taskId = apiResp.output.taskId;\n System.out.println(\"task_id: \" + taskId);\n } else {\n System.out.println(\"respBody: \" + respBody);\n return;\n }\n } else {\n System.out.println(\"Failed! HTTP code: \" + response.code());\n if (response.body() != null) {\n System.out.println(response.body().string());\n }\n return;\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n\n boolean finished = false;\n while (!finished) {\n try {\n TimeUnit.SECONDS.sleep(2);\n } catch (InterruptedException e) {\n Thread.currentThread().interrupt();\n return;\n }\n\n String queryUrl = API_URL_QUERY + taskId;\n Request queryRequest = new Request.Builder()\n .url(queryUrl)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .addHeader(\"Content-Type\", \"application/json\")\n .get()\n .build();\n\n try (Response response = client.newCall(queryRequest).execute()) {\n if (response.body() != null) {\n String queryResponse = response.body().string();\n ApiResponse apiResp = gson.fromJson(queryResponse, ApiResponse.class);\n\n if (apiResp.output != null && apiResp.output.taskStatus != null) {\n String status = apiResp.output.taskStatus;\n System.out.println(\"task status: \" + status);\n if (\"SUCCEEDED\".equalsIgnoreCase(status)\n || \"FAILED\".equalsIgnoreCase(status)\n || \"UNKNOWN\".equalsIgnoreCase(status)) {\n finished = true;\n System.out.println(\"task finished: \");\n System.out.println(queryResponse);\n }\n } else {\n System.out.println(\"query response: \" + queryResponse);\n }\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n }\n }\n\n static class ApiResponse {\n @SerializedName(\"request_id\")\n String requestId;\n Output output;\n }\n\n static class Output {\n @SerializedName(\"task_id\")\n String taskId;\n @SerializedName(\"task_status\")\n String taskStatus;\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json index 9afac99e..6dc7164d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json @@ -15,10 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-asr-flash-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00033", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -31,41 +37,17 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-10-27T10:00:46.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-ASR-Flash-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2989727.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json index 0d5c8ec0..708242d0 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json @@ -15,10 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-asr-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "timeBand": "standard", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -31,21 +37,24 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], "modelAlias": "qwen3-asr-flash", "versionTag": "MAJOR", "latestOnlineAt": "2025-09-08T05:39:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-ASR-Flash", "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [], "samples": { "dashscope": { "default": { - "python": "import os\nimport dashscope\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": [\n # 此处用于配置定制化识别的Context\n {\"text\": \"\"},\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"audio\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"},\n ]\n }\n]\nresponse = dashscope.MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-asr-flash\",\n messages=messages,\n result_format=\"message\",\n asr_options={\n # \"language\": \"zh\", # 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n \"enable_lid\":True,\n \"enable_itn\":False\n }\n)\nprint(response)", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"audio\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\")))\n .build();\n\n MultiModalMessage sysMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n // 此处用于配置定制化识别的Context\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"\")))\n .build();\n\n Map asrOptions = new HashMap<>();\n asrOptions.put(\"enable_lid\", true);\n asrOptions.put(\"enable_itn\", false);\n // asrOptions.put(\"language\", \"zh\"); // 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-asr-flash\")\n .message(userMessage)\n .message(sysMessage)\n .parameter(\"asr_options\", asrOptions)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": [\n # 此处用于配置定制化识别的Context\n {\"text\": \"\"},\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"audio\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"},\n ]\n }\n]\nresponse = dashscope.MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-asr-flash\",\n messages=messages,\n result_format=\"message\",\n asr_options={\n # \"language\": \"zh\", # 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n \"enable_lid\":True,\n \"enable_itn\":False\n }\n)\nprint(response)", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"audio\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\")))\n .build();\n\n MultiModalMessage sysMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n // 此处用于配置定制化识别的Context\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"\")))\n .build();\n\n Map asrOptions = new HashMap<>();\n asrOptions.put(\"enable_lid\", true);\n asrOptions.put(\"enable_itn\", false);\n // asrOptions.put(\"language\", \"zh\"); // 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-asr-flash\")\n .message(userMessage)\n .message(sysMessage)\n .parameter(\"asr_options\", asrOptions)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2986952.html" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json index 53b49766..cf19c7ef 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json @@ -17,9 +17,6 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-30b-a3b-instruct", "qpmInfo": { "model-default-actual": { @@ -39,6 +36,95 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.25", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "9", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3.75", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "37.5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG" ], @@ -49,54 +135,28 @@ "maxInputTokens": 204800, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Coder-30B-A3B-Instruct", "docUrl": "https://help.aliyun.com/zh/model-studio/qwen-coder#272bcaccea8ls", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], + "category": "Cost-optimized", "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-30b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-30b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-30b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-30b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json index d9010d47..fb3827be 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json @@ -17,9 +17,6 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-480b-a35b-instruct", "qpmInfo": { "model-default-actual": { @@ -39,6 +36,95 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "9", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG" ], @@ -49,54 +135,27 @@ "maxInputTokens": 204800, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Coder-480B-A35B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-480b-a35b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-480b-a35b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-480b-a35b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-480b-a35b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json index 71e5e762..f4aeedcb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json @@ -19,9 +19,6 @@ "cache" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-flash", "qpmInfo": { "model-default-actual": { @@ -41,64 +38,204 @@ "type": "model-default" } }, - "capabilities": [ - "TG" + "priceTimeBands": [ + "standard" ], - "versionTag": "MAJOR", - "maxOutputTokens": 65536, - "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", - "contextWindow": 1000000, - "maxInputTokens": 997952, - "inferenceProvider": "bailian", - "name": "Qwen3-Coder-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ + "multiPrices": [ { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 }, { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.875", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.125", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 }, { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "6.25", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 1000000 } ], + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", + "contextWindow": 1000000, + "maxInputTokens": 997952, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Coder-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2850166.html", "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-flash\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-flash\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-flash\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-flash\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-flash\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-flash\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-flash\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-flash\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-flash\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-flash\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json index 1d369815..92e082b9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json @@ -19,9 +19,6 @@ "cache" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-plus", "qpmInfo": { "model-default-actual": { @@ -41,6 +38,179 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "200", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG" ], @@ -51,60 +221,27 @@ "maxInputTokens": 997952, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Coder-Plus", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json index 3b4b1748..161c9a99 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json @@ -17,10 +17,37 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-livetranslate-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "64", + "timeBand": "standard", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "timeBand": "standard", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "240", + "timeBand": "standard", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -39,6 +66,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Audio-Translate" ], @@ -47,40 +77,19 @@ "latestOnlineAt": "2025-09-23T11:11:30.000+00:00", "contextWindow": 53248, "maxInputTokens": 49152, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-LiveTranslate-Flash-Realtime", "docUrl": "https://www.alibabacloud.com/help/en/document_detail/2983281.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", + "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", "docUrl": "https://help.aliyun.com/document_detail/2983281.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json index 73d436d0..ac5f1cad 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json @@ -16,10 +16,37 @@ "description": "Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-livetranslate-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "timeBand": "standard", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -38,6 +65,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], @@ -46,36 +76,14 @@ "latestOnlineAt": "2025-12-04T12:21:57.000+00:00", "contextWindow": 53248, "maxInputTokens": 49152, - "inferenceProvider": "bailian", - "name": "Qwen3-LiveTranslate-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2999748.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "offlineInfo": { + "inference": { + "announceUrl": "" } - ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-LiveTranslate-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2999748.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json index 95e6b638..53a805fd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json @@ -11,7 +11,74 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Completions API", + "name": "search_strategy:agent_max", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent_max", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY", + "tag": "限时优惠" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + } + ], "description": "千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。", "collectionTag": "qwen3", "features": [ @@ -24,9 +91,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-max", "qpmInfo": { "model-default-actual": { @@ -46,6 +110,227 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.125", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.4", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.5", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.75", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "TG", "Reasoning" @@ -58,93 +343,34 @@ "maxInputTokens": 258048, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Max", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3-max\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3-max\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3-max\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3-max\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -159,7 +385,21 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3系列Max模型Preview版本，实现思考模式和非思考模式的有效融合。思考模式下在智能体编程能力、常识知识推理能力、数学/科学/通用类推理等能力上均有显著增强。", "collectionTag": "qwen3", "features": [ @@ -167,9 +407,6 @@ "cache" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-max-preview", "qpmInfo": { "model-default-actual": { @@ -189,6 +426,95 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "TG", "Reasoning" @@ -201,86 +527,27 @@ "maxInputTokens": 258048, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max-preview\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-max-preview\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-max-preview\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max-preview\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-max-preview\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-max-preview\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-max-preview\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max-preview\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-max-preview\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max-preview\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json index 98eec78f..28679897 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json @@ -15,10 +15,23 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-omni-30b-a3b-captioner", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15.8", + "timeBand": "standard", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "timeBand": "standard", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -37,6 +50,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], @@ -45,36 +61,15 @@ "latestOnlineAt": "2025-09-16T16:38:17.000+00:00", "contextWindow": 65536, "maxInputTokens": 32768, - "inferenceProvider": "bailian", - "name": "Qwen3-Omni-30b-a3b-Captioner", - "docUrl": "https://help.aliyun.com/document_detail/2980468.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" } - ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Omni-30b-a3b-Captioner", + "docUrl": "https://help.aliyun.com/document_detail/2980468.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json index 50dfdaf6..1cdbea0b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json @@ -19,10 +19,51 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-omni-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.2", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "18.9", + "timeBand": "standard", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.9", + "timeBand": "standard", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "8.3", + "timeBand": "standard", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "15.2", + "timeBand": "standard", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "75.1", + "timeBand": "standard", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -41,6 +82,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], @@ -50,49 +94,20 @@ "latestOnlineAt": "2025-12-03T16:00:00.000+00:00", "contextWindow": 65536, "maxInputTokens": 49152, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Omni-Flash-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "开启深度思考，开启后将不支持音频输出" - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json index 494aeafe..ad450cbc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json @@ -22,10 +22,86 @@ "function-calling" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-omni-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "timeBand": "standard", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "15.8", + "timeBand": "standard", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "timeBand": "standard", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.9", + "timeBand": "standard", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "timeBand": "standard", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "62.6", + "timeBand": "standard", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.8", + "timeBand": "standard", + "type": "thinking_text_input_token", + "priceName": "输入：文本（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "15.8", + "timeBand": "standard", + "type": "thinking_audio_input_token", + "priceName": "输入：音频（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "timeBand": "standard", + "type": "thinking_vision_input_token", + "priceName": "输入：图片/视频（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "6.9", + "timeBand": "standard", + "type": "thinking_purein_text_output_token", + "priceName": "输出：文本（思考模式下，输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "timeBand": "standard", + "type": "thinking_multiin_text_output_token", + "priceName": "输出：文本（思考模式下，输入包含图片/音频/视频时）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -44,6 +120,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], @@ -54,50 +133,15 @@ "latestOnlineAt": "2025-12-03T16:00:00.000+00:00", "contextWindow": 65536, "maxInputTokens": 49152, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Omni-Flash", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "开启深度思考，开启后将不支持音频输出" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json index fd92af23..dd611505 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json @@ -15,10 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-flash-realtime", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -31,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -38,16 +47,9 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen3-tts-flash-realtime-2025-11-27", "latestOnlineAt": "2025-11-26T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-Flash-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json index e1cf65b5..cbdeb533 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json @@ -17,10 +17,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -33,6 +39,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -41,16 +50,9 @@ "shortDescription": "韵律拟人，低延迟，支持十种语言和国内多种方言输出", "equivalentSnapshot": "qwen3-tts-flash-2025-11-27", "latestOnlineAt": "2025-11-26T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-Flash", "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json index dd146c2e..8548bb42 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json @@ -15,10 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-instruct-flash-realtime", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -31,42 +37,24 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "modelAlias": "qwen3-tts-instruct-flash-realtime", "versionTag": "MAJOR", "latestOnlineAt": "2026-01-21T07:33:55.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "qwen3-tts-instruct-flash-realtime", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json index 875e188b..c44a82a8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json @@ -15,11 +15,17 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-instruct-flash", "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -32,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -39,46 +48,20 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen3-tts-instruct-flash-2026-01-26", "latestOnlineAt": "2026-02-10T02:56:41.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-Instruct-Flash", "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageHint", - "default": "" - }, - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - }, - { - "name": "指令控制", - "key": "instructions", - "tip": "仅支持中英文，通过自然语言合成语音的语气、语速、情感及人物性格，需要具体客观的描述文字，如：\n· 请用非常激昂且高亢的语气说话，表现出获得重大成功后的狂喜与激动。\n· 语速请保持中等偏慢，语气要显得优雅、知性，给人以从容不迫的安心感。" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ntext = \"Dear listeners, hello everyone. Welcome to the evening news.\"\n\nresponse = dashscope.MultiModalConversation.call(\n model=\"qwen3-tts-instruct-flash\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n instructions='The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.',\n optimize_instructions=True,\n stream=False\n)\nprint(response)", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.io.FileOutputStream;\nimport java.io.InputStream;\nimport java.net.URL;\n\npublic class Main {\n private static final String MODEL = \"qwen3-tts-instruct-flash\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(MODEL)\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .parameter(\"instructions\",\"The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.\")\n .parameter(\"optimize_instructions\",true)\n .build();\n MultiModalConversationResult result = conv.call(param);\n String audioUrl = result.getOutput().getAudio().getUrl();\n System.out.print(audioUrl);\n\n // 下载音频文件到本地\n try (InputStream in = new URL(audioUrl).openStream();\n FileOutputStream out = new FileOutputStream(\"downloaded_audio.wav\")) {\n byte[] buffer = new byte[1024];\n int bytesRead;\n while ((bytesRead = in.read(buffer)) != -1) {\n out.write(buffer, 0, bytesRead);\n }\n } catch (Exception e) {\n System.out.println(\"\\nError message: \" + e.getMessage());\n }\n }\n public static void main(String[] args) {\n try {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\ntext = \"Dear listeners, hello everyone. Welcome to the evening news.\"\n\nresponse = dashscope.MultiModalConversation.call(\n model=\"qwen3-tts-instruct-flash\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n instructions='The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.',\n optimize_instructions=True,\n stream=False\n)\nprint(response)", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.io.FileOutputStream;\nimport java.io.InputStream;\nimport java.net.URL;\n\npublic class Main {\n private static final String MODEL = \"qwen3-tts-instruct-flash\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(MODEL)\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .parameter(\"instructions\",\"The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.\")\n .parameter(\"optimize_instructions\",true)\n .build();\n MultiModalConversationResult result = conv.call(param);\n String audioUrl = result.getOutput().getAudio().getUrl();\n System.out.print(audioUrl);\n\n // 下载音频文件到本地\n try (InputStream in = new URL(audioUrl).openStream();\n FileOutputStream out = new FileOutputStream(\"downloaded_audio.wav\")) {\n byte[] buffer = new byte[1024];\n int bytesRead;\n while ((bytesRead = in.read(buffer)) != -1) {\n out.write(buffer, 0, bytesRead);\n }\n } catch (Exception e) {\n System.out.println(\"\\nError message: \" + e.getMessage());\n }\n }\n public static void main(String[] args) {\n try {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json index 0c9eee16..4f320d79 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json @@ -15,10 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-vc-realtime-2026-01-15", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -31,28 +37,30 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "modelAlias": "qwen3-tts-vc-realtime-0115", "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-01-14T11:23:59.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "qwen3-tts-vc-realtime-2026-01-15", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", "category": "Audio", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { - "python": "# DashScope SDK Version>=1.23.9，Python Version >=3.10\n# coding=utf-8\n# Installation instructions for pyaudio:\n# APPLE Mac OS X\n# brew install portaudio\n# pip install pyaudio\n# Debian/Ubuntu\n# sudo apt-get install python-pyaudio python3-pyaudio\n# or\n# pip install pyaudio\n# CentOS\n# sudo yum install -y portaudio portaudio-devel && pip install pyaudio\n# Microsoft Windows\n# python -m pip install pyaudio\n\nimport pyaudio\nimport os\nimport requests\nimport base64\nimport pathlib\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import QwenTtsRealtime, QwenTtsRealtimeCallback, AudioFormat\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\nTEXT_TO_SYNTHESIZE = [\n 'Today is a wonderful day to build something people love!'\n]\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"The audio file does not exist {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"Failed to create voice: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"The voice response failed to be resolved: {e}\")\n\ndef init_dashscope_api_key():\n dashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self._player = pyaudio.PyAudio()\n self._stream = self._player.open(\n format=pyaudio.paInt16, channels=1, rate=24000, output=True\n )\n\n def on_open(self) -> None:\n print('[TTS] has been established')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self._stream.stop_stream()\n self._stream.close()\n self._player.terminate()\n print(f'[TTS] close, code={close_status_code}, msg={close_msg}')\n\n def on_event(self, response: dict) -> None:\n try:\n event_type = response.get('type', '')\n if event_type == 'session.created':\n print(f'[TTS] session begin: {response[\"session\"][\"id\"]}')\n elif event_type == 'response.audio.delta':\n audio_data = base64.b64decode(response['delta'])\n self._stream.write(audio_data)\n elif event_type == 'response.done':\n print(f'[TTS] response complete, Response ID: {qwen_tts_realtime.get_last_response_id()}')\n elif event_type == 'session.finished':\n print('[TTS] session end')\n self.complete_event.set()\n except Exception as e:\n print(f'[Error] callback error: {e}')\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n print('Qwen TTS Realtime ...')\n\n callback = MyCallback()\n qwen_tts_realtime = QwenTtsRealtime(\n model=DEFAULT_TARGET_MODEL,\n callback=callback,\n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n qwen_tts_realtime.connect()\n \n qwen_tts_realtime.update_session(\n voice=create_voice(VOICE_FILE_PATH),\n response_format=AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode='server_commit'\n )\n\n for text_chunk in TEXT_TO_SYNTHESIZE:\n print(f'[send text]: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n\n print(f'[Metric] session_id={qwen_tts_realtime.get_session_id()}, '\n f'first_audio_delay={qwen_tts_realtime.get_first_audio_delay()}s')", - "java": "// Java DashScope SDK Version >= 2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport javax.sound.sampled.*;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.nio.charset.StandardCharsets;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\";\n private static final String PREFERRED_NAME = \"guanyu\";\n \n private static final String AUDIO_FILE = \"voice.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n private static String[] textToSynthesize = {\n \"Today is a wonderful day to build something people love!\"\n };\n\n public static String toDataUrl(String filePath) throws IOException {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static String createVoice() throws Exception {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\").openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(StandardCharsets.UTF_8));\n }\n\n int status = con.getResponseCode();\n System.out.println(\"HTTP status: \" + status);\n\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(status >= 200 && status < 300 ? con.getInputStream() : con.getErrorStream(),\n StandardCharsets.UTF_8))) {\n StringBuilder response = new StringBuilder();\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n System.out.println(\"response: \" + response);\n\n if (status == 200) {\n JsonObject jsonObj = new Gson().fromJson(response.toString(), JsonObject.class);\n return jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n }\n throw new IOException(\"failed: \" + status + \" - \" + response);\n }\n }\n\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws Exception {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(TARGET_MODEL)\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n\n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // Processing when the connection is established\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // Processing at the time of session creation\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n \n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n break;\n case \"session.finished\":\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // Handling when the connection is closed\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(createVoice())\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n\n\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + "python": "# DashScope SDK Version>=1.23.9，Python Version >=3.10\n# coding=utf-8\n# Installation instructions for pyaudio:\n# APPLE Mac OS X\n# brew install portaudio\n# pip install pyaudio\n# Debian/Ubuntu\n# sudo apt-get install python-pyaudio python3-pyaudio\n# or\n# pip install pyaudio\n# CentOS\n# sudo yum install -y portaudio portaudio-devel && pip install pyaudio\n# Microsoft Windows\n# python -m pip install pyaudio\n\nimport pyaudio\nimport os\nimport requests\nimport base64\nimport pathlib\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import QwenTtsRealtime, QwenTtsRealtimeCallback, AudioFormat\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\nTEXT_TO_SYNTHESIZE = [\n 'Today is a wonderful day to build something people love!'\n]\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"The audio file does not exist {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\n url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"Failed to create voice: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"The voice response failed to be resolved: {e}\")\n\ndef init_dashscope_api_key():\n dashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self._player = pyaudio.PyAudio()\n self._stream = self._player.open(\n format=pyaudio.paInt16, channels=1, rate=24000, output=True\n )\n\n def on_open(self) -> None:\n print('[TTS] has been established')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self._stream.stop_stream()\n self._stream.close()\n self._player.terminate()\n print(f'[TTS] close, code={close_status_code}, msg={close_msg}')\n\n def on_event(self, response: dict) -> None:\n try:\n event_type = response.get('type', '')\n if event_type == 'session.created':\n print(f'[TTS] session begin: {response[\"session\"][\"id\"]}')\n elif event_type == 'response.audio.delta':\n audio_data = base64.b64decode(response['delta'])\n self._stream.write(audio_data)\n elif event_type == 'response.done':\n print(f'[TTS] response complete, Response ID: {qwen_tts_realtime.get_last_response_id()}')\n elif event_type == 'session.finished':\n print('[TTS] session end')\n self.complete_event.set()\n except Exception as e:\n print(f'[Error] callback error: {e}')\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n print('Qwen TTS Realtime ...')\n\n callback = MyCallback()\n qwen_tts_realtime = QwenTtsRealtime(\n model=DEFAULT_TARGET_MODEL,\n callback=callback,\n url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n qwen_tts_realtime.connect()\n \n qwen_tts_realtime.update_session(\n voice=create_voice(VOICE_FILE_PATH),\n response_format=AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode='server_commit'\n )\n\n for text_chunk in TEXT_TO_SYNTHESIZE:\n print(f'[send text]: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n\n print(f'[Metric] session_id={qwen_tts_realtime.get_session_id()}, '\n f'first_audio_delay={qwen_tts_realtime.get_first_audio_delay()}s')", + "java": "// Java DashScope SDK Version >= 2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport javax.sound.sampled.*;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.nio.charset.StandardCharsets;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\";\n private static final String PREFERRED_NAME = \"guanyu\";\n \n private static final String AUDIO_FILE = \"voice.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n private static String[] textToSynthesize = {\n \"Today is a wonderful day to build something people love!\"\n };\n\n public static String toDataUrl(String filePath) throws IOException {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static String createVoice() throws Exception {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\").openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(StandardCharsets.UTF_8));\n }\n\n int status = con.getResponseCode();\n System.out.println(\"HTTP status: \" + status);\n\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(status >= 200 && status < 300 ? con.getInputStream() : con.getErrorStream(),\n StandardCharsets.UTF_8))) {\n StringBuilder response = new StringBuilder();\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n System.out.println(\"response: \" + response);\n\n if (status == 200) {\n JsonObject jsonObj = new Gson().fromJson(response.toString(), JsonObject.class);\n return jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n }\n throw new IOException(\"failed: \" + status + \" - \" + response);\n }\n }\n\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws Exception {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(TARGET_MODEL)\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n\n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // Processing when the connection is established\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // Processing at the time of session creation\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n \n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n break;\n case \"session.finished\":\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // Handling when the connection is closed\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(createVoice())\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n\n\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json index 0d15a100..62c48e27 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json @@ -15,11 +15,17 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-vc-2026-01-22", "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -32,26 +38,28 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "modelAlias": "qwen3-tts-vc-0122", "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-02-10T03:00:44.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-VC-2026-01-22", "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport requests\nimport base64\nimport pathlib\nimport dashscope\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-2026-01-22\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"音频文件不存在: {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"create voice failed: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"failed: {e}\")\n\n\nif __name__ == '__main__':\n dashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n text = \"今天天气怎么样？\"\n \n response = dashscope.MultiModalConversation.call(\n model=DEFAULT_TARGET_MODEL,\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=create_voice(VOICE_FILE_PATH),\n stream=False\n )\n print(response)" + "python": "import os\nimport requests\nimport base64\nimport pathlib\nimport dashscope\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-2026-01-22\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"音频文件不存在: {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"create voice failed: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"failed: {e}\")\n\n\nif __name__ == '__main__':\n dashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n text = \"今天天气怎么样？\"\n \n response = dashscope.MultiModalConversation.call(\n model=DEFAULT_TARGET_MODEL,\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=create_voice(VOICE_FILE_PATH),\n stream=False\n )\n print(response)" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json index 518a47eb..dfe9a084 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json @@ -15,10 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-vd-realtime-2026-01-15", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -31,48 +37,30 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "modelAlias": "qwen3-tts-vd-realtime-0115", "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-01-14T11:14:10.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "qwen3-tts-vd-realtime-2026-01-15", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-realtime-2026-01-15',\n callback=callback, \n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-realtime-2026-01-15\")\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-realtime-2026-01-15',\n callback=callback, \n url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-realtime-2026-01-15\")\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json index f95a31a7..fd45e733 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json @@ -15,11 +15,17 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-vd-2026-01-26", "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "timeBand": "standard", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -32,27 +38,29 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "modelAlias": "qwen3-tts-vd-0126", "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-02-10T02:59:43.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-VD-2026-01-26", "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-2026-01-26',\n callback=callback, \n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-2026-01-26\")\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-2026-01-26',\n callback=callback, \n url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-2026-01-26\")\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json index 8f009242..1a0d0c07 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json @@ -23,9 +23,6 @@ "cache" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-flash", "qpmInfo": { "model-default-actual": { @@ -45,6 +42,227 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.075", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1875", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.015", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.375", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "VU", "Reasoning" @@ -58,85 +276,26 @@ "maxInputTokens": 260096, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-Flash", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-flash\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-flash',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-flash\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-flash',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-flash\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-flash\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-flash\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-flash\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json index f662f875..d62b1543 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json @@ -24,9 +24,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-plus", "qpmInfo": { "model-default-actual": { @@ -46,6 +43,227 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.875", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.75", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "VU", "Reasoning" @@ -57,81 +275,22 @@ "latestOnlineAt": "2026-01-25T16:00:00.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-Plus", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-plus',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-plus',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-plus\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-plus\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-plus\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-plus\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json index 5d0abd08..32e7eb61 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5原生视觉语言系列Flash模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步；响应速度快，兼具推理速度和性能。", "collectionTag": "qwen3.5", "features": [ @@ -26,9 +92,6 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-flash", "iconUrl": "", "qpmInfo": { @@ -49,10 +112,210 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.02", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", - "VU", - "TG" + "TG", + "VU" ], "modelAlias": "", "versionTag": "MAJOR", @@ -60,88 +323,30 @@ "latestOnlineAt": "2026-02-23T03:28:01.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Flash", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], + "category": "Cost-optimized", "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json index 63a65f52..fefde783 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json @@ -17,10 +17,37 @@ "collectionTag": "qwen3.5", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-livetranslate-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "40", + "timeBand": "standard", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "timeBand": "standard", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "100", + "timeBand": "standard", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "160", + "timeBand": "standard", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -39,6 +66,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Audio-Translate" ], @@ -48,41 +78,14 @@ "latestOnlineAt": "2026-05-19T08:25:22.000+00:00", "contextWindow": 53248, "maxInputTokens": 49152, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-LiveTranslate-Flash-Realtime", "docUrl": "https://www.alibabacloud.com/help/en/document_detail/2983281.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", + "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", "docUrl": "https://help.aliyun.com/document_detail/2983281.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json index 91c827b8..456eef48 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json @@ -17,10 +17,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-ocr", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,6 +52,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -48,49 +64,22 @@ "contextWindow": 65536, "maxInputTokens": 49152, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-OCR", "docUrl": "https://help.aliyun.com/document_detail/2860683.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3.5-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3.5-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3.5-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3.5-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3.5-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3.5-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3.5-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3.5-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json index 9954766f..d5f395ce 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json @@ -15,7 +15,21 @@ "Audio" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", "collectionTag": "Qwen3.5", "features": [ @@ -23,11 +37,38 @@ "function-calling" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-omni-flash-realtime", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "27", + "timeBand": "standard", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "107", + "timeBand": "standard", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "timeBand": "standard", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,6 +87,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], @@ -56,42 +100,15 @@ "latestOnlineAt": "2026-03-30T03:54:16.000+00:00", "contextWindow": 262144, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Omni-Flash-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2880812.html", "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json index 7a872831..67ce60dc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json @@ -15,18 +15,59 @@ "Audio" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", "collectionTag": "Qwen3.5", "features": [ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-omni-flash", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "18", + "timeBand": "standard", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "72", + "timeBand": "standard", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.2", + "timeBand": "standard", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "13.3", + "timeBand": "standard", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -45,6 +86,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], @@ -55,45 +99,16 @@ "latestOnlineAt": "2026-03-30T03:58:51.000+00:00", "contextWindow": 262144, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Omni-Flash", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json index e20db8f9..13196f6e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json @@ -15,7 +15,21 @@ "Audio" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", "collectionTag": "Qwen3.5", "features": [ @@ -23,11 +37,38 @@ "function-calling" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-omni-plus-realtime", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "80", + "timeBand": "standard", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "300", + "timeBand": "standard", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "timeBand": "standard", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,6 +87,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], @@ -56,42 +100,15 @@ "latestOnlineAt": "2026-03-30T03:54:11.000+00:00", "contextWindow": 262144, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Omni-Plus-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-plus-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-plus-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-plus-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-plus-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json index 6adf2613..1796efaa 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json @@ -15,18 +15,106 @@ "Audio" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", "collectionTag": "Qwen3.5", "features": [ - "web-search" + "web-search", + "function-calling", + "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-omni-plus", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "53", + "timeBand": "standard", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "213", + "timeBand": "standard", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "timeBand": "standard", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "timeBand": "standard", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "26.5", + "timeBand": "standard", + "type": "omni_audio_input_token_batch", + "priceName": "输入：音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.5", + "timeBand": "standard", + "type": "omni_no_audio_input_token_batch", + "priceName": "输入：文本/图片/视频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "omni_no_audio_output_token_batch", + "priceName": "输出：文本（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "53", + "timeBand": "standard", + "discount": 0.5, + "type": "omni_audio_input_token_batch_chat", + "priceName": "输入：音频（Batch Chat）" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "timeBand": "standard", + "discount": 0.5, + "type": "omni_no_audio_input_token_batch_chat", + "priceName": "输入：文本/图片/视频（Batch Chat）" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "timeBand": "standard", + "discount": 0.5, + "type": "omni_no_audio_output_token_batch_chat", + "priceName": "输出：文本（Batch Chat）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -45,6 +133,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], @@ -55,45 +146,16 @@ "latestOnlineAt": "2026-03-30T03:58:25.000+00:00", "contextWindow": 262144, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Omni-Plus", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-plus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-plus\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-plus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-plus\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json index fc319d5d..024154ec 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5原生视觉语言系列Plus模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在多项任务评测中，3.5系列均展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。", "collectionTag": "qwen3.5", "features": [ @@ -26,9 +92,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-plus", "iconUrl": "", "qpmInfo": { @@ -49,6 +112,206 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG", "Reasoning", @@ -61,88 +324,29 @@ "latestOnlineAt": "2026-02-15T09:15:31.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json index b83e2fe2..7e4c9192 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json @@ -13,40 +13,150 @@ "Video" ] }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。", + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。", "collectionTag": "qwen3.5", "features": [ "model-experience", "function-calling", "structured-outputs", "web-search", - "prefix-completion", - "fine-tuning" + "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-27b", + "model": "qwen3.5-397b-a17b", + "iconUrl": "", "qpmInfo": { "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, + "count_limit_period": 1, + "usage_limit": 500000, "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, + "count_limit": 10, + "usage_limit_period": 30, "type": "model-default" }, "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, + "count_limit_period": 1, + "usage_limit": 500000, "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, + "count_limit": 10, + "usage_limit_period": 30, "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "VU", @@ -55,97 +165,32 @@ "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-23T03:42:27.000+00:00", + "latestOnlineAt": "2026-02-15T09:18:22.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "trainingTypes": { - "sft": [ - "lora", - "full" - ] - }, - "inferenceProvider": "bailian", - "name": "Qwen3.5-27B", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-397B-A17B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-397b-a17b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-397b-a17b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-27b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-27b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-397b-a17b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-397b-a17b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-397b-a17b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-397b-a17b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } @@ -162,8 +207,74 @@ "Video" ] }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。", + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。", "collectionTag": "qwen3.5", "features": [ "model-experience", @@ -173,29 +284,72 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-397b-a17b", - "iconUrl": "", + "model": "qwen3.5-35b-a3b", "qpmInfo": { "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 500000, + "count_limit_period": 6, + "usage_limit": 1000000, "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 30, + "count_limit": 60, + "usage_limit_period": 60, "type": "model-default" }, "model-default": { - "count_limit_period": 1, - "usage_limit": 500000, + "count_limit_period": 6, + "usage_limit": 1000000, "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 30, + "count_limit": 60, + "usage_limit_period": 60, "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3.2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12.8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "VU", @@ -204,91 +358,32 @@ "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-15T09:18:22.000+00:00", + "latestOnlineAt": "2026-02-23T03:27:40.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", - "name": "Qwen3.5-397B-A17B", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-35B-A3B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-397b-a17b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-397b-a17b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-397b-a17b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-397b-a17b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-397b-a17b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-397b-a17b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } @@ -305,21 +400,85 @@ "Video" ] }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。", + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。", "collectionTag": "qwen3.5", "features": [ "model-experience", "function-calling", "structured-outputs", "web-search", - "prefix-completion" + "prefix-completion", + "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-35b-a3b", + "model": "qwen3.5-27b", "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -338,6 +497,53 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "VU", @@ -346,91 +552,39 @@ "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-23T03:27:40.000+00:00", + "latestOnlineAt": "2026-02-23T03:42:27.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", - "name": "Qwen3.5-35B-A3B", + "trainingTypes": { + "sft": [ + "lora", + "full" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-27B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], + "category": "Cost-optimized", "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-27b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-27b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } @@ -447,7 +601,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5系列122B-A10B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现仅次于Qwen3.5-397B-A17B，文本能力显著优于Qwen3-235B-2507，视觉能力优于Qwen3-VL-235B。", "collectionTag": "qwen3.5", "features": [ @@ -458,9 +678,6 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-122b-a10b", "qpmInfo": { "model-default-actual": { @@ -480,6 +697,53 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "VU", @@ -491,88 +755,29 @@ "latestOnlineAt": "2026-02-23T03:28:11.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-122B-A10B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-122b-a10b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-122b-a10b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-122b-a10b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-122b-a10b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-122b-a10b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-122b-a10b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-122b-a10b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-122b-a10b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-122b-a10b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-122b-a10b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-122b-a10b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-122b-a10b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json index 33ca24e5..d1f2011e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。", "collectionTag": "qwen3.6", "features": [ @@ -26,9 +92,6 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-flash", "qpmInfo": { "model-default-actual": { @@ -48,6 +111,137 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.6", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "timeBand": "standard", + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28.8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "14.4", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "timeBand": "standard", + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "28.8", + "timeBand": "standard", + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "VU", @@ -59,88 +253,29 @@ "latestOnlineAt": "2026-04-16T14:02:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Flash", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json index 79bc5b4f..2cbc8a4a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json @@ -22,9 +22,6 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-max-preview", "qpmInfo": { "model-default-actual": { @@ -44,6 +41,81 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "9", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "54", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "11.25", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.9", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "90", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "18.75", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "TG" @@ -56,87 +128,28 @@ "maxInputTokens": 245760, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-max-preview\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-max-preview',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-max-preview\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-max-preview',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-max-preview\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-max-preview\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json index 102483c2..26a02835 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。", "collectionTag": "qwen3.6", "features": [ @@ -26,9 +92,6 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-plus", "qpmInfo": { "model-default-actual": { @@ -48,6 +111,141 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "VU", @@ -60,89 +258,30 @@ "latestOnlineAt": "2026-04-01T11:55:34.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json index 20e285d4..b846d044 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果相较3.5-35B-A3B显著提升了agentic coding能力、数学推理和代码推理能力、空间智能能力、物体定位与目标检测能力。", "collectionTag": "qwen3.6", "features": [ @@ -24,10 +90,23 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-35b-a3b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10.8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -46,6 +125,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "TG", @@ -57,89 +139,30 @@ "latestOnlineAt": "2026-04-16T14:01:43.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-35B-A3B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-35b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-35b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } } @@ -165,10 +188,23 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-27b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -187,6 +223,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU", @@ -198,82 +237,23 @@ "latestOnlineAt": "2026-04-22T12:50:52.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-27B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-flash.json new file mode 100644 index 00000000..8ebeca67 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-flash.json @@ -0,0 +1,347 @@ +{ + "name": "Qwen3.7-Flash", + "description": "Qwen3.7原生视觉语言系列Flash模型，相较3.6-Flash全面提升多模态理解与Agent执行能力。重点强化多模态基础能力、万物识别能力更强，真实世界感知与空间智能进一步提升，Search Agent、CI Agent等多模态Agent场景能力显著升级、端到端任务执行更稳定，多模态Coding能力优化、vibe coding 体验更加流畅。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Image", + "Text", + "Video" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.7原生视觉语言系列Flash模型，相较3.6-Flash全面提升多模态理解与Agent执行能力。重点强化多模态基础能力、万物识别能力更强，真实世界感知与空间智能进一步提升，Search Agent、CI Agent等多模态Agent场景能力显著升级、端到端任务执行更稳定，多模态Coding能力优化、vibe coding 体验更加流畅。", + "collectionTag": "qwen3.7", + "features": [ + "batch", + "cache", + "function-calling", + "model-experience", + "prefix-completion", + "structured-outputs", + "web-search" + ], + "provider": "qwen", + "model": "qwen3.7-flash", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 5000000, + "usage_limit_field": "total_tokens", + "count_limit": 3000, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 5000000, + "usage_limit_field": "total_tokens", + "count_limit": 3000, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "priceTimeBands": [], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.04", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.02", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], + "capabilities": [ + "TG", + "VU", + "Reasoning" + ], + "modelAlias": "qwen3.7-flash", + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-07-20T17:02:37.000+00:00", + "contextWindow": 1000000, + "maxInputTokens": 991808, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.7-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "category": "Cost-optimized", + "samples": { + "openai": { + "completionsAPI": { + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + }, + "responsesAPI": { + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "docUrl": "https://help.aliyun.com/document_detail/3016808.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.7-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.7-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json index e56dd891..069d1a07 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json @@ -11,7 +11,47 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。", "collectionTag": "qwen3.7", "features": [ @@ -23,10 +63,79 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.7-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -45,6 +154,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -56,77 +168,24 @@ "latestOnlineAt": "2026-05-21T06:37:11.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Max", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -141,7 +200,47 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中规模最大、综合能力最强的Max模型预览版，仅支持思考模式，开放纯文本模型能力供体验。主要优化面向用户的通用对话场景，例如知识问答、指令跟随、创意写作等。", "collectionTag": "qwen3.7", "features": [ @@ -151,10 +250,23 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.7-max-preview", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -173,6 +285,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -183,76 +298,29 @@ "latestOnlineAt": "2026-05-19T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" }, "responsesAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\": \"9.9和9.11哪个大？\"\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-max-preview\",\n input=\"9.9和9.11哪个大？\"\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-max-preview\",\n input: \"9.9和9.11哪个大？\"\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\": \"9.9和9.11哪个大？\"\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-max-preview\",\n input=\"9.9和9.11哪个大？\"\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-max-preview\",\n input: \"9.9和9.11哪个大？\"\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json index c1fce0bc..b9aa9afa 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。", "collectionTag": "qwen3.7", "features": [ @@ -26,9 +92,6 @@ "structured-outputs" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.7-plus", "qpmInfo": { "model-default-actual": { @@ -48,6 +111,165 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "discount": 0.8, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "discount": 0.8, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "discount": 0.8, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "timeBand": "standard", + "discount": 0.8, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "discount": 0.8, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "discount": 0.8, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "discount": 0.8, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "timeBand": "standard", + "discount": 0.8, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "timeBand": "standard", + "discount": 0.8, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "timeBand": "standard", + "discount": 0.8, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "timeBand": "standard", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG", "Reasoning", @@ -61,83 +283,30 @@ "contextWindow": 1000000, "maxInputTokens": 991808, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.7-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.7-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json index b3a42bb4..d54be612 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json @@ -21,10 +21,23 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-32b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -43,6 +56,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "Reasoning" @@ -52,73 +68,26 @@ "contextWindow": 131072, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-32B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-32b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-32b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-32b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-32b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-32b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-32b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } @@ -142,10 +111,23 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-32b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -164,6 +146,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -174,58 +159,26 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-32B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-32b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-32b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-32b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-32b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-32b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-32b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -250,10 +203,23 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-30b-a3b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -272,6 +238,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "Reasoning" @@ -283,73 +252,26 @@ "maxInputTokens": 126976, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-30B-A3B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-30b-a3b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-30b-a3b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-30b-a3b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-30b-a3b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } @@ -374,10 +296,23 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-30b-a3b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -396,6 +331,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -406,58 +344,26 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-30B-A3B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-30b-a3b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-30b-a3b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-30b-a3b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-30b-a3b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-30b-a3b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-30b-a3b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -483,10 +389,30 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-8b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -505,6 +431,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "Reasoning" @@ -523,72 +452,25 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-8B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-8b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-8b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-8b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-8b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-8b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-8b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } @@ -614,10 +496,30 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-8b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -636,6 +538,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -653,57 +558,25 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-8B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-8b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-8b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-8b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-8b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-8b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-8b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -728,10 +601,23 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-235b-a22b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -750,6 +636,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "Reasoning" @@ -761,73 +650,26 @@ "maxInputTokens": 126976, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-235B-A22B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-235b-a22b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-235b-a22b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-235b-a22b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-235b-a22b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } @@ -852,10 +694,23 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-235b-a22b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -874,6 +729,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -884,58 +742,26 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-235B-A22B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-235b-a22b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-235b-a22b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-235b-a22b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-235b-a22b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-235b-a22b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-235b-a22b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -953,10 +779,23 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-next-80b-a3b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -975,6 +814,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -986,58 +828,26 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Next-80B-A3B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-next-80b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-next-80b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1056,10 +866,23 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-next-80b-a3b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -1078,6 +901,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1090,72 +916,25 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Next-80B-A3B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-next-80b-a3b-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-next-80b-a3b-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1179,10 +958,23 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-30b-a3b-instruct-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -1201,6 +993,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -1221,58 +1016,26 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-30B-A3B-Instruct-2507", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-30b-a3b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-30b-a3b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1293,10 +1056,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-30b-a3b-thinking-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -1315,6 +1091,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1326,73 +1105,26 @@ "maxInputTokens": 126976, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-30B-A3B-Thinking-2507", "docUrl": "https://help.aliyun.com/document_detail/2870973.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1412,10 +1144,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-235b-a22b-thinking-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -1434,6 +1179,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1443,73 +1191,26 @@ "contextWindow": 131072, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-235B-A22B-Thinking-2507", "docUrl": "https://help.aliyun.com/document_detail/2870973.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1532,10 +1233,23 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-235b-a22b-instruct-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -1554,6 +1268,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -1564,59 +1281,27 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-235B-A22B-Instruct-2507", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-235b-a22b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-235b-a22b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1640,10 +1325,37 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-235b-a22b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -1662,6 +1374,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1673,75 +1388,27 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-235B-A22B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1765,10 +1432,37 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-30b-a3b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 30, @@ -1787,6 +1481,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1799,74 +1496,26 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-30B-A3B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1891,10 +1540,44 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-32b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "timeBand": "standard", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -1913,6 +1596,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1937,75 +1623,27 @@ }, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-32B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-32b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-32b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-32b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-32b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-32b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-32b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-32b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-32b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -2030,10 +1668,44 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-14b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "timeBand": "standard", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -2052,6 +1724,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -2073,75 +1748,27 @@ }, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-14B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-14b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-14b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-14b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-14b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-14b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-14b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-14b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-14b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -2166,10 +1793,44 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-8b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "timeBand": "standard", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "timeBand": "standard", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -2188,6 +1849,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -2210,74 +1874,26 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-8B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-8b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-8b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-8b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-8b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-8b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-8b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-8b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-8b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -2298,9 +1914,6 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-next", "qpmInfo": { "model-default-actual": { @@ -2320,6 +1933,74 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "TG" ], @@ -2331,54 +2012,27 @@ "maxInputTokens": 204800, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通义千问3-Coder-Next", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-next\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-next\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-next\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-next\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-next\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-next\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-next\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-next\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-next\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-next\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-next\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-next\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json index 302d2793..a055face 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json @@ -20,10 +20,37 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwq-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "timeBand": "standard", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -42,6 +69,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning" ], @@ -52,54 +82,27 @@ "maxInputTokens": 98304, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "QwQ-Plus", "docUrl": "https://help.aliyun.com/document_detail/2870973.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwq-plus\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwq-plus',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwq-plus\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwq-plus',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwq-plus\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwq-plus\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwq-plus\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwq-plus\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/sambert.json b/skills/bailian-docs-llm-wiki/models/groups/sambert.json index 86b2dd73..c4001266 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/sambert.json +++ b/skills/bailian-docs-llm-wiki/models/groups/sambert.json @@ -14,45 +14,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiyue-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:17:01.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知悦", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -74,45 +56,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiyuan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-20T09:14:08.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知媛", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -134,45 +98,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiying-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:17:43.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知颖", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -194,45 +140,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiye-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:17:07.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知晔", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -254,45 +182,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiya-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:17:55.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知雅", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -314,45 +224,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhixiao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知笑", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -374,45 +266,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhixiang-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:13.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知祥", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -434,45 +308,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiwei-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:04.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知薇", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -494,45 +350,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiting-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:16.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知婷", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -554,45 +392,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhistella-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:19:54.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知莎", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -614,45 +434,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhishuo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:18.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知硕", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -674,45 +476,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhishu-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:05.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知树", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -734,45 +518,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiru-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:22.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知茹", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -794,45 +560,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiqian-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:08.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知倩", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -854,45 +602,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiqi-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:25.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知琪", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -914,45 +644,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhinan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T08:10:47.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知楠", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -974,45 +686,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhina-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:46.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知娜", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1034,45 +728,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhimo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知墨", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1094,45 +770,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiming-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:49.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知茗", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1154,45 +812,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhimiao-emo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:38.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知妙（多情感）", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1214,45 +854,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhimao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:34.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知猫", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1274,45 +896,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhilun-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:12.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知伦", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1334,45 +938,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhijing-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:41.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知婧", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1394,45 +980,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhijia-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知佳", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1454,45 +1022,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhihao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:45.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知浩", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1514,45 +1064,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhigui-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:15.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知柜", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1574,45 +1106,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhifei-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:49.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知飞", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1634,45 +1148,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhide-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:17.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知德", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1694,45 +1190,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhida-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:52.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知达", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1754,45 +1232,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhichu-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:12.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知厨", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1814,45 +1274,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-waan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:56.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Waan", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1874,45 +1316,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-perla-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:20.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Perla", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1934,45 +1358,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-indah-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:59.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Indah", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1994,45 +1400,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-hanna-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:09.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Hanna", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2054,45 +1442,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-eva-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Eva", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2114,45 +1484,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-donna-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:23.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Donna", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2174,45 +1526,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-clara-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:06.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Clara", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2234,45 +1568,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-cindy-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:26.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Cindy", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2294,45 +1610,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-camila-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:27.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Camila", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2354,45 +1652,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-cally-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:58.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Cally", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2414,45 +1694,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-brian-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:54.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Brian", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2474,45 +1736,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-betty-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:16:48.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Betty", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2534,45 +1778,27 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-beth-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "timeBand": "standard", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:30.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Beth", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json b/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json index e4546f6c..ccc4abba 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json @@ -14,49 +14,19 @@ "description": "鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "shoemodel-v1", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-21T02:47:24.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "鞋靴模特", "docUrl": "https://help.aliyun.com/document_detail/2804662.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"shoemodel-v1\",\n \"input\": {\n \"template_image_url\": \"https://img.alicdn.com/imgextra/i1/O1CN01EyPuz31d79mKv75CI_!!6000000003688-49-tps-1120-1680.webp\",\n \"shoe_image_url\": [\"https://img.alicdn.com/imgextra/i2/O1CN01zTIls120gdcrI7dX2_!!6000000006879-49-tps-1120-1493.webp\"]\n },\n \"parameters\": \n {\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https:/ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"shoemodel-v1\",\n \"input\": {\n \"template_image_url\": \"https://img.alicdn.com/imgextra/i1/O1CN01EyPuz31d79mKv75CI_!!6000000003688-49-tps-1120-1680.webp\",\n \"shoe_image_url\": [\"https://img.alicdn.com/imgextra/i2/O1CN01zTIls120gdcrI7dX2_!!6000000006879-49-tps-1120-1493.webp\"]\n },\n \"parameters\": \n {\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https:/[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json index 848f0974..abc4406d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json +++ b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json @@ -17,11 +17,24 @@ "prefix-completion" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "siliconflow/deepseek-v3.2", "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -40,6 +53,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -52,53 +68,10 @@ "inferenceProvider": "siliconflow", "name": "SiliconFlow DeepSeek-V3.2", "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3014912.html" } } @@ -119,11 +92,24 @@ "prefix-completion" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "siliconflow/deepseek-v3.1-terminus", "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -142,6 +128,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -154,47 +143,10 @@ "inferenceProvider": "siliconflow", "name": "SiliconFlow DeepSeek-V3.1-Terminus", "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3014912.html" } } @@ -215,11 +167,24 @@ "prefix-completion" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "siliconflow/deepseek-v3-0324", "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -238,6 +203,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -249,47 +217,10 @@ "inferenceProvider": "siliconflow", "name": "SiliconFlow DeepSeek-V3-0324", "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3-0324\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3-0324\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3014912.html" } } @@ -310,11 +241,24 @@ "prefix-completion" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "siliconflow/deepseek-r1-0528", "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -333,6 +277,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -346,27 +293,10 @@ "inferenceProvider": "siliconflow", "name": "SiliconFlow DeepSeek-R1-0528", "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-r1-0528\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-r1-0528\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3014912.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json b/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json index 3896d924..290d0f82 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json +++ b/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json @@ -14,9 +14,6 @@ "description": "热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "speech-biasing", "qpmInfo": { "model-default-actual": { @@ -35,36 +32,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音识别热词", - "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] + "docUrl": "https://help.aliyun.com/document_detail/2712535.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json index 58eca4da..104849fb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json @@ -21,10 +21,30 @@ "batch" ], "provider": "stepfun", - "limit": { - "message": "model not exist" - }, "model": "stepfun/step-3.7-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.35", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.1", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.27", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -43,8 +63,12 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ - "TG" + "TG", + "VU" ], "versionTag": "MAJOR", "maxOutputTokens": 262144, @@ -55,39 +79,6 @@ "inferenceProvider": "stepfun", "name": "stepfun/step-3.7-flash", "docUrl": "https://help.aliyun.com/document_detail/3036697.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 1, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json index 115a1609..bbdb79ad 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json @@ -14,10 +14,23 @@ "description": "意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "tongyi-intent-detect-v3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -36,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -44,50 +60,23 @@ "latestOnlineAt": "2024-12-12T11:33:25.000+00:00", "contextWindow": 8192, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "意图分类模型", "docUrl": "https://help.aliyun.com/document_detail/2861138.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"tongyi-intent-detect-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"tongyi-intent-detect-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"tongyi-intent-detect-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"tongyi-intent-detect-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"tongyi-intent-detect-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-intent-detect-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"tongyi-intent-detect-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-intent-detect-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json index 1510e8ce..98ed097a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json @@ -14,10 +14,23 @@ "description": "通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "tongyi-xiaomi-analysis-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.2", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -36,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -44,49 +60,22 @@ "latestOnlineAt": "2026-01-09T03:23:03.000+00:00", "contextWindow": 32768, "maxInputTokens": 28672, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通义晓蜜-对话分析-flash", "docUrl": "https://help.aliyun.com/document_detail/3015075.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", - "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", - "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-flash\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", + "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", + "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-flash\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" } }, "dashscope": { "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-flash\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-flash\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json index 706099a7..15aeedae 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json @@ -14,10 +14,23 @@ "description": "通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "tongyi-xiaomi-analysis-pro", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2.7", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -36,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -44,49 +60,22 @@ "latestOnlineAt": "2026-01-09T03:31:49.000+00:00", "contextWindow": 32768, "maxInputTokens": 28672, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通义晓蜜-对话分析-pro", "docUrl": "https://help.aliyun.com/document_detail/3015075.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", - "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", - "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-pro\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", + "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", + "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-pro\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" } }, "dashscope": { "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-pro\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-pro\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json index a467552c..431ec936 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json @@ -18,10 +18,72 @@ "batch" ], "provider": "tripo", - "limit": { - "message": "model not exist" - }, "model": "Tripo/Tripo-P1.0", + "prices": [ + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_no_texture", + "priceName": "文生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_no_texture", + "priceName": "单图生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_no_texture", + "priceName": "多图生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_sd_texture", + "priceName": "文生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_sd_texture", + "priceName": "单图生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_sd_texture", + "priceName": "多图生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_hd_texture", + "priceName": "文生3D（带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_hd_texture", + "priceName": "单图生3D（带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_hd_texture", + "priceName": "多图生3D（带高清贴图）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -40,6 +102,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "3D-generation" ], @@ -49,37 +114,10 @@ "inferenceProvider": "tripo", "name": "Tripo-P1.0", "docUrl": "https://help.aliyun.com/document_detail/3030679.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-P1.0\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-P1.0\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3030679.html" } } @@ -101,10 +139,135 @@ "structured-outputs" ], "provider": "tripo", - "limit": { - "message": "model not exist" - }, "model": "Tripo/Tripo-H3.1", + "prices": [ + { + "priceUnit": "每次", + "price": "0.7", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_standard_no_texture", + "priceName": "文生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_standard_no_texture", + "priceName": "单图生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_standard_no_texture", + "priceName": "多图生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_standard_sd_texture", + "priceName": "文生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_standard_sd_texture", + "priceName": "单图生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_standard_sd_texture", + "priceName": "多图生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_standard_hd_texture", + "priceName": "文生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_standard_hd_texture", + "priceName": "单图生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_standard_hd_texture", + "priceName": "多图生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_ultra_no_texture", + "priceName": "文生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_ultra_no_texture", + "priceName": "单图生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_ultra_no_texture", + "priceName": "多图生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_ultra_sd_texture", + "priceName": "文生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_ultra_sd_texture", + "priceName": "单图生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_ultra_sd_texture", + "priceName": "多图生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_ultra_hd_texture", + "priceName": "文生3D（超清版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_ultra_hd_texture", + "priceName": "单图生3D（超清版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_ultra_hd_texture", + "priceName": "多图生3D（超清版+带高清贴图）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -123,6 +286,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "3D-generation" ], @@ -132,37 +298,10 @@ "inferenceProvider": "tripo", "name": "Tripo-H3.1", "docUrl": "https://help.aliyun.com/document_detail/3030679.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-H3.1\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-H3.1\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3030679.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json index a7fcb0ab..8e6333ca 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json @@ -23,18 +23,21 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -57,6 +60,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -69,33 +75,6 @@ "inferenceProvider": "vanchin", "name": "vanchin/deepseek-v4-pro", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -128,18 +107,21 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -162,6 +144,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -175,49 +160,6 @@ "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-V3.2-think", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -248,18 +190,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -282,6 +227,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -295,49 +243,6 @@ "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-V3.1-Terminus", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -370,18 +275,21 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -404,6 +312,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -416,33 +327,6 @@ "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-V3", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -474,18 +358,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -508,6 +395,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -521,33 +411,6 @@ "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-R1", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -578,12 +441,14 @@ { "priceUnit": "每百万tokens", "price": "0.216", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "0.216", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -606,6 +471,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "TG" @@ -618,33 +486,6 @@ "maxInputTokens": 8192, "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-OCR", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json index d9aa0e65..9af6a681 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json +++ b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json @@ -14,49 +14,38 @@ "description": "视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "video-style-transform", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.2", + "timeBand": "standard", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.5", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "视频风格重绘", "docUrl": "https://help.aliyun.com/document_detail/2846319.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"video-style-transform\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250704/viwndw/%E5%8E%9F%E8%A7%86%E9%A2%91.mp4\"\n },\n \"parameters\": {\n \"style\": 0,\n \"video_fps\": 15,\n \"min_len\": 540\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"video-style-transform\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250704/viwndw/%E5%8E%9F%E8%A7%86%E9%A2%91.mp4\"\n },\n \"parameters\": {\n \"style\": 0,\n \"video_fps\": 15,\n \"min_len\": 540\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json index b12ee860..fda316eb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json +++ b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json @@ -15,49 +15,31 @@ "description": "VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "videoretalk", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-10T06:23:15.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "声动人像VideoRetalk", "docUrl": "https://help.aliyun.com/document_detail/2860466.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"videoretalk\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/pvegot/input_video_01.mp4\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/aumwir/stella2-%E6%9C%89%E5%A3%B0%E4%B9%A67.wav\",\n \"ref_image_url\": \"\"\n },\n \"parameters\": {\n \"video_extension\": false\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"videoretalk\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/pvegot/input_video_01.mp4\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/aumwir/stella2-%E6%9C%89%E5%A3%B0%E4%B9%A67.wav\",\n \"ref_image_url\": \"\"\n },\n \"parameters\": {\n \"video_extension\": false\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json index deec3f50..0f35b352 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json @@ -20,18 +20,21 @@ { "priceUnit": "每张", "price": "0.625", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" }, { "priceUnit": "每张", "price": "1", + "timeBand": "standard", "type": "image_type_2k", "priceName": "图片生成（2K）" }, { "priceUnit": "每张", "price": "1.46875", + "timeBand": "standard", "type": "image_type_4k", "priceName": "图片生成（4K）" } @@ -52,6 +55,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -62,33 +68,6 @@ "name": "Vidu-image_reference2image", "docUrl": "https://help.aliyun.com/document_detail/3045893.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -116,18 +95,21 @@ { "priceUnit": "每张", "price": "0.46875", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" }, { "priceUnit": "每张", "price": "0.78125", + "timeBand": "standard", "type": "image_type_2k", "priceName": "图片生成（2K）" }, { "priceUnit": "每张", "price": "1.09375", + "timeBand": "standard", "type": "image_type_4k", "priceName": "图片生成（4K）" } @@ -148,6 +130,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -158,33 +143,6 @@ "name": "ViduQ3-fast_reference2image", "docUrl": "https://help.aliyun.com/document_detail/3045893.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -212,18 +170,21 @@ { "priceUnit": "每张", "price": "0.9375", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" }, { "priceUnit": "每张", "price": "0.9375", + "timeBand": "standard", "type": "image_type_2k", "priceName": "图片生成（2K）" }, { "priceUnit": "每张", "price": "1.71875", + "timeBand": "standard", "type": "image_type_4k", "priceName": "图片生成（4K）" } @@ -244,6 +205,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -254,33 +218,6 @@ "name": "ViduQ2-Pro_reference2image", "docUrl": "https://help.aliyun.com/document_detail/3045893.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -308,6 +245,7 @@ { "priceUnit": "每张", "price": "0.28125", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" } @@ -328,6 +266,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -338,33 +279,6 @@ "name": "ViduQ2-fast_reference2image", "docUrl": "https://help.aliyun.com/document_detail/3045893.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json index c01401c5..9c6aeb50 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json @@ -20,12 +20,14 @@ { "priceUnit": "每秒", "price": "0.75", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.90625", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -46,6 +48,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -56,33 +61,6 @@ "name": "ViduQ3-Ad_reference2video", "docUrl": "https://help.aliyun.com/document_detail/3045891.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -110,6 +88,7 @@ { "priceUnit": "每秒", "price": "0.875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -130,6 +109,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -140,33 +122,6 @@ "name": "ViduQ3-Drama_reference2video", "docUrl": "https://help.aliyun.com/document_detail/3045892.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -194,12 +149,14 @@ { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.46875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -220,6 +177,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -230,33 +190,6 @@ "name": "ViduQ3-Pro-fast_img2video", "docUrl": "https://help.aliyun.com/document_detail/3026198.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -285,18 +218,21 @@ { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.9375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -319,6 +255,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -327,37 +266,6 @@ "latestOnlineAt": "2026-03-26T13:47:00.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Pro_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -385,18 +293,21 @@ { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.9375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -419,6 +330,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -427,37 +341,6 @@ "latestOnlineAt": "2026-03-26T13:45:57.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Pro_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -488,18 +371,21 @@ { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.9375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -522,6 +408,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -530,33 +419,6 @@ "latestOnlineAt": "2026-03-26T13:46:33.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Pro_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -585,18 +447,21 @@ { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.4375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -619,6 +484,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -627,37 +495,6 @@ "latestOnlineAt": "2026-03-26T13:46:46.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Turbo_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -685,18 +522,21 @@ { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.4375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -719,6 +559,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -727,37 +570,6 @@ "latestOnlineAt": "2026-03-26T13:46:51.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Turbo_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -788,18 +600,21 @@ { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.4375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -822,6 +637,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -830,33 +648,6 @@ "latestOnlineAt": "2026-03-26T13:46:28.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Turbo_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -885,18 +676,21 @@ { "priceUnit": "每秒", "price": "0.15625", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.34375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -919,6 +713,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -927,32 +724,6 @@ "latestOnlineAt": "2026-03-26T13:46:18.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Pro_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -981,18 +752,21 @@ { "priceUnit": "每秒", "price": "0.15625", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.34375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1015,6 +789,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1023,33 +800,6 @@ "latestOnlineAt": "2026-03-26T13:46:02.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Pro_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1079,18 +829,21 @@ { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1113,6 +866,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1121,32 +877,6 @@ "latestOnlineAt": "2026-03-26T13:46:23.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Pro_reference2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1175,18 +905,21 @@ { "priceUnit": "每秒", "price": "0.0875", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.46875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1209,6 +942,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1217,32 +953,6 @@ "latestOnlineAt": "2026-03-26T13:46:08.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Turbo_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1270,18 +980,21 @@ { "priceUnit": "每秒", "price": "0.1125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.21875", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1304,6 +1017,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1312,32 +1028,6 @@ "latestOnlineAt": "2026-03-26T13:45:21.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1368,18 +1058,21 @@ { "priceUnit": "每秒", "price": "0.0875", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.46875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1402,6 +1095,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1410,33 +1106,6 @@ "latestOnlineAt": "2026-03-26T13:46:14.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Turbo_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1465,18 +1134,21 @@ { "priceUnit": "每秒", "price": "0.21875", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.28125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1499,6 +1171,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1507,32 +1182,6 @@ "latestOnlineAt": "2026-03-26T13:45:51.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2_reference2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1560,18 +1209,21 @@ { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.625", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1594,6 +1246,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1602,33 +1257,6 @@ "latestOnlineAt": "2026-04-27T08:35:14.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1656,18 +1284,21 @@ { "priceUnit": "每秒", "price": "0.15625", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.40625", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1690,6 +1321,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1698,33 +1332,6 @@ "latestOnlineAt": "2026-04-27T08:33:07.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Turbo_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1752,12 +1359,14 @@ { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.9375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1780,6 +1389,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1788,33 +1400,6 @@ "latestOnlineAt": "2026-04-28T02:40:03.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-mix_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1842,12 +1427,14 @@ { "priceUnit": "每秒", "price": "0.1", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.2", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1870,6 +1457,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1878,32 +1468,6 @@ "latestOnlineAt": "2026-04-27T12:09:25.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Pro-fast_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json b/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json index b603e699..b0dd8943 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json @@ -14,49 +14,19 @@ "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "virtualmodel-v2", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-25T15:18:50.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "虚拟模特V2", "docUrl": "https://help.aliyun.com/document_detail/2796985.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"virtualmodel-v2\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"virtualmodel-v2\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json index 943af02a..128b08cc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json +++ b/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json @@ -14,9 +14,6 @@ "description": "大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "voice-enrollment", "qpmInfo": { "model-default-actual": { @@ -35,10 +32,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "大模型声音复刻及声音设计", - "docUrl": "https://help.aliyun.com/document_detail/2861519.html", - "predictConfig": [] + "docUrl": "https://help.aliyun.com/document_detail/2861519.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json index 582f6265..142946c5 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json @@ -18,10 +18,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.7-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,49 +44,25 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-01T03:48:10.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-Image", "docUrl": "https://help.aliyun.com/document_detail/3026980.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "2048*2048" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "生成数量", - "range": [ - 1, - 4 - ] - }, - { - "name": "组图生成", - "key": "enable_sequential", - "default": false - }, - { - "name": "智能改写", - "key": "thinking_mode", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", - "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", - "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", + "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", + "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3026980.html" } } @@ -102,10 +84,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.7-image-pro", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -122,49 +110,30 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-01T05:32:16.000+00:00", - "inferenceProvider": "bailian", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-Image-Pro", "docUrl": "https://help.aliyun.com/document_detail/3026980.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "2048*2048" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "生成数量", - "range": [ - 1, - 4 - ] - }, - { - "name": "组图生成", - "key": "enable_sequential", - "default": false - }, - { - "name": "智能改写", - "key": "thinking_mode", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", - "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image-pro',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", - "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image-pro\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", + "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image-pro',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", + "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image-pro\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3026980.html" } } @@ -185,10 +154,16 @@ "collectionTag": "wan2.6", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.6-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -207,45 +182,21 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-12-15T11:55:32.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-Image", "docUrl": "https://help.aliyun.com/document_detail/3001143.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--data '{\n \"model\": \"wan2.6-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"给我一个3张图辣椒炒肉教程\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"size\": \"1280*1280\",\n \"enable_interleave\":true\n }\n}'\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--data '{\n \"model\": \"wan2.6-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"给我一个3张图辣椒炒肉教程\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"size\": \"1280*1280\",\n \"enable_interleave\":true\n }\n}'\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } @@ -266,10 +217,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.5-i2i-preview", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -288,45 +245,21 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T13:38:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.5-I2I-Preview", "docUrl": "https://help.aliyun.com/document_detail/2982258.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n-H 'X-DashScope-Async: enable' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n\"model\": \"wan2.5-i2i-preview\",\n\"input\": {\n\"prompt\": \"将花卉连衣裙换成一件复古风格的蕾丝长裙，领口和袖口有精致的刺绣细节。\",\n\"images\": [\n\"https://img.alicdn.com/imgextra/i3/O1CN01Z1BLz61dMGqxmijRd_!!6000000003721-2-tps-1080-1620.png\"\n]\n},\n\"parameters\": {\n\"size\": \"1280*1280\",\n\"n\": 1\n}\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n-H 'X-DashScope-Async: enable' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n\"model\": \"wan2.5-i2i-preview\",\n\"input\": {\n\"prompt\": \"将花卉连衣裙换成一件复古风格的蕾丝长裙，领口和袖口有精致的刺绣细节。\",\n\"images\": [\n\"https://img.alicdn.com/imgextra/i3/O1CN01Z1BLz61dMGqxmijRd_!!6000000003721-2-tps-1080-1620.png\"\n]\n},\n\"parameters\": {\n\"size\": \"1280*1280\",\n\"n\": 1\n}\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } @@ -343,51 +276,33 @@ "description": "万相-通义图像编辑，支持预设编辑任务与指令式编辑，包含多种局部/全图编辑能力，如图像风格化、线稿生图、局部重绘、参考图生成、图像外扩、图像超分等。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-imageedit", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-25T07:22:03.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-ImageEdit", "docUrl": "https://help.aliyun.com/document_detail/2868981.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-imageedit\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-imageedit\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-imageedit\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n syncCall();\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-imageedit\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-imageedit\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-imageedit\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n syncCall();\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json index f06cfe9b..dc5a125c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json @@ -19,10 +19,23 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.7-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,51 +52,28 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-03T03:21:23.000+00:00", - "inferenceProvider": "bailian", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-I2V", "docUrl": "https://help.aliyun.com/document_detail/3025059.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由rap构成，没有其他对话或杂音。\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\"\n },\n {\n \"type\": \"driving_audio\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n \n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 10,\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由rap构成，没有其他对话或杂音。\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\"\n },\n {\n \"type\": \"driving_audio\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n \n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 10,\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025059.html" } } @@ -107,10 +97,37 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.6-i2v-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.5", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "timeBand": "standard", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.25", + "timeBand": "standard", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -127,58 +144,22 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2026-01-15T07:18:54.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "wan2.6-I2V-flash", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", "category": "Wan", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" } } } @@ -201,10 +182,39 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.6-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "discount": 0.5, + "type": "720P_batch", + "priceName": "视频生成（720P Batch Chat）" + }, + { + "priceUnit": "每秒", + "price": "1", + "timeBand": "standard", + "discount": 0.5, + "type": "1080P_batch", + "priceName": "视频生成（1080P Batch Chat）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -221,57 +231,21 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-12-03T13:03:01.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-I2V", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" } } } @@ -295,10 +269,30 @@ "fine-tuning" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.5-i2v-preview", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.3", + "timeBand": "standard", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -317,6 +311,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -327,46 +324,13 @@ "lora" ] }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.5-I2V-Preview", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.5-i2v-preview',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.5-i2v-preview',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } @@ -384,10 +348,23 @@ "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-i2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.14", + "timeBand": "standard", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -404,46 +381,21 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-I2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } @@ -462,10 +414,30 @@ "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-kf2v-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.1", + "timeBand": "standard", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.2", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.48", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -484,6 +456,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -494,40 +469,13 @@ "lora" ] }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-KF2V-Flash", "docUrl": "https://help.aliyun.com/document_detail/2880649.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wan2.2-kf2v-flash\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wan2.2-kf2v-flash\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" } } } @@ -546,10 +494,23 @@ "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-animate-move", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.4", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "type": "video_ratio_pro", + "priceName": "视频生成（pro）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -568,46 +529,22 @@ "async_user_concurrency_limit": 1 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "modelAlias": "Wan2.2-Animate-Move", "versionTag": "MAJOR", "latestOnlineAt": "2025-09-19T04:02:30.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-Animate-Move", "docUrl": "https://help.aliyun.com/document_detail/2981852.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-move\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/adsyrp/move_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/kaakcn/move_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-move\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/adsyrp/move_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/kaakcn/move_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } @@ -626,10 +563,23 @@ "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-animate-mix", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + }, + { + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "type": "video_ratio_pro", + "priceName": "视频生成（pro）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -648,187 +598,110 @@ "async_user_concurrency_limit": 1 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-19T04:02:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-Animate-Mix", "docUrl": "https://help.aliyun.com/document_detail/2982219.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-mix\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/bhkfor/mix_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/wqefue/mix_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-mix\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/bhkfor/mix_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/wqefue/mix_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } }, { "inferenceMetadata": { - "response_modality": [ - "Video" - ], + "response_modality": [], "request_modality": [ "Image" ] }, - "description": "全新升级的万相2.2图生视频，生成速度更快。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。", + "description": "wan2.2-s2v-detect 是 wan2.2-s2v 的辅助模型，用于确认输入的人物肖像图片是否符合 wan2.2-s2v 模型所需的人物肖像图片规范。wan2.2-s2v 模型基于 wan2.2-s2v-detect 检测通过的图片和人声音频文件进行视频生成。", "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-i2v-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 2 + "model": "wan2.2-s2v-detect", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "timeBand": "standard", + "type": "image_detect_number", + "priceName": "图片检测" } - }, + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-11T03:53:00.000+00:00", - "trainingTypes": { - "sft": [ - "lora" - ] - }, - "inferenceProvider": "bailian", - "name": "Wan2.2-I2V-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "latestOnlineAt": "2025-08-25T11:54:09.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-S2V-Detect", + "docUrl": "https://help.aliyun.com/zh/document_detail/2978214.html", "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-flash',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "curl": "curl 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"wan2.2-s2v-detect\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\"\n }\n }'" } } } }, { "inferenceMetadata": { - "response_modality": [], + "response_modality": [ + "Video" + ], "request_modality": [ "Image" ] }, - "description": "wan2.2-s2v-detect 是 wan2.2-s2v 的辅助模型，用于确认输入的人物肖像图片是否符合 wan2.2-s2v 模型所需的人物肖像图片规范。wan2.2-s2v 模型基于 wan2.2-s2v-detect 检测通过的图片和人声音频文件进行视频生成。", + "description": "wan2.2-s2v 是一款视频生成模型，可基于人物图片和人声音频文件，生成高质量的人物说话/唱歌/表演动态视频。", "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-s2v-detect", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-25T11:54:09.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.2-S2V-Detect", - "docUrl": "https://help.aliyun.com/zh/document_detail/2978214.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, + "model": "wan2.2-s2v", + "prices": [ { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "priceUnit": "每秒", + "price": "0.5", + "timeBand": "standard", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "priceUnit": "每秒", + "price": "0.9", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" } ], + "priceTimeBands": [ + "standard" + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-25T11:54:37.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "通义万相2.2-数字人-S2V", + "docUrl": "https://help.aliyun.com/zh/document_detail/2978215.html", "samples": { "dashscope": { "default": { - "curl": "curl 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"wan2.2-s2v-detect\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\"\n }\n }'" + "curl": "curl 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n --header 'X-DashScope-Async: enable' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header 'Content-Type: application/json' \\\n --data '{\n \"model\": \"wan2.2-s2v\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/iaqpio/input_audio.MP3\"\n },\n \"parameters\": {\n \"resolution\": \"480P\"\n }\n }'" } } } @@ -842,53 +715,72 @@ "Image" ] }, - "description": "wan2.2-s2v 是一款视频生成模型，可基于人物图片和人声音频文件，生成高质量的人物说话/唱歌/表演动态视频。", + "description": "全新升级的万相2.2图生视频，生成速度更快。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。", "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-s2v", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-25T11:54:37.000+00:00", - "inferenceProvider": "bailian", - "name": "通义万相2.2-数字人-S2V", - "docUrl": "https://help.aliyun.com/zh/document_detail/2978215.html", - "predictConfig": [ + "model": "wan2.2-i2v-flash", + "prices": [ { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" + "priceUnit": "每秒", + "price": "0.1", + "timeBand": "standard", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" }, { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "priceUnit": "每秒", + "price": "0.2", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "priceUnit": "每秒", + "price": "0.48", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" } ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "priceTimeBands": [ + "standard" + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-11T03:53:00.000+00:00", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-I2V-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", "samples": { "dashscope": { "default": { - "curl": "curl 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n --header 'X-DashScope-Async: enable' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header 'Content-Type: application/json' \\\n --data '{\n \"model\": \"wan2.2-s2v\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/iaqpio/input_audio.MP3\"\n },\n \"parameters\": {\n \"resolution\": \"480P\"\n }\n }'" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-flash',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } @@ -905,10 +797,16 @@ "description": "万相2.1-首尾帧-Plus，两张图片生成丝滑过度视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成画面细节更丰富。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-kf2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -925,45 +823,21 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-04-20T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-KF2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2880649.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wanx2.1-kf2v-plus\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wanx2.1-kf2v-plus\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" } } } @@ -980,10 +854,16 @@ "description": "万相2.1-图生视频-Plus，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，视频质量更高。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-i2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -1000,46 +880,21 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-20T03:30:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-I2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } @@ -1056,50 +911,31 @@ "description": "万相2.1-图生视频-Turbo，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成速度更快。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-i2v-turbo", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-02-27T02:24:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-I2V-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-turbo',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-turbo',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json index bbadbc33..50aa4cec 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json @@ -23,12 +23,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -49,6 +51,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -58,43 +63,7 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-R2V", "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { @@ -124,24 +93,28 @@ { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.5", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -162,6 +135,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -170,39 +146,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-R2V-Flash", "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -232,12 +175,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -258,6 +203,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -266,34 +214,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-R2V", "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json index 71e37806..7a31917e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json @@ -17,10 +17,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.6-t2i", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,44 +45,17 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-12-15T08:05:15.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-T2I", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*1280", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { @@ -100,10 +79,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.5-t2i-preview", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -122,48 +107,21 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-19T08:44:37.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.5-T2I-Preview", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*1280", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.5-t2i-preview\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.5-t2i-preview\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -183,10 +141,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-t2i-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -203,48 +167,21 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-T2I-Plus", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -264,10 +201,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-t2i-flash", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -284,48 +227,21 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-T2I-Flash", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-flash\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-flash\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -344,10 +260,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-t2i-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -364,48 +286,21 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-08T16:09:10.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-T2I-Plus", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -424,10 +319,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-t2i-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -444,48 +345,21 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-08T16:12:34.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-T2I-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -504,116 +378,32 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-v1", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-01-05T08:26:20.000+00:00", - "inferenceProvider": "bailian", - "name": "wanx-t2i", - "docUrl": "https://help.aliyun.com/document_detail/2712483.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "style", - "key": "style", - "default": "", - "tip": "输出风格" - }, - { - "name": "size", - "key": "size", - "default": "1024*1024", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, + "prices": [ { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "种子值", - "range": [ - 1, - 4294967289 - ] + "priceUnit": "每张", + "price": "0.16", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" } ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通，本模型为通义万相的2024年5月21号的历史快照。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-v1-0521", + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2024-05-22T13:57:26.000+00:00", - "inferenceProvider": "bailian", - "name": "万相-文本生成图像-2024-05-21", + "versionTag": "MAJOR", + "latestOnlineAt": "2024-01-05T08:26:20.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "wanx-t2i", "docUrl": "https://help.aliyun.com/document_detail/2712483.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1-0521\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wanx-v1\",\n \"input\": {\n \"prompt\": \"近景镜头，18岁的中国女孩，古代服饰，圆脸，正面看着镜头，民族优雅的服装，商业摄影，室外，电影级光照，半身特写，精致的淡妆，锐利的边缘。\"\n },\n \"parameters\": {\n \"style\": \"\",\n \"size\": \"1024*1024\",\n \"n\": 1\n }\n}'\n ", + "docUrl": "https://help.aliyun.com/document_detail/2712483.html" } } } @@ -632,10 +422,16 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.0-t2i-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.04", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -652,48 +448,50 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-20T03:29:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.0-T2I-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1024*1024", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.0-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通，本模型为通义万相的2024年5月21号的历史快照。", + "features": [], + "provider": "wan", + "model": "wanx-v1-0521", + "capabilities": [ + "IG" ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2024-05-22T13:57:26.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "万相-文本生成图像-2024-05-21", + "docUrl": "https://help.aliyun.com/document_detail/2712483.html", "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.0-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1-0521\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json index bdaeaa25..3700e359 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json @@ -23,12 +23,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -49,6 +51,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -58,43 +63,7 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-T2V", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { @@ -126,18 +95,21 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "discount": 0.5, "type": "720P_batch", "priceName": "视频生成（720P Batch Chat）" @@ -145,6 +117,7 @@ { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "discount": 0.5, "type": "1080P_batch", "priceName": "视频生成（1080P Batch Chat）" @@ -166,6 +139,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -174,45 +150,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-T2V", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -243,18 +180,21 @@ { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -277,6 +217,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -285,39 +228,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.5-T2V-Preview", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -344,12 +254,14 @@ { "priceUnit": "每秒", "price": "0.14", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -370,6 +282,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -378,34 +293,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-T2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { @@ -431,6 +318,7 @@ { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" } @@ -451,6 +339,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -459,34 +350,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-T2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1280*720" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { @@ -512,12 +375,14 @@ { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" } @@ -538,6 +403,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -546,34 +414,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-T2V-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1280*720" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json index 0d3e9cb0..f1126d6c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json @@ -19,10 +19,23 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.7-videoedit", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "timeBand": "standard", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "timeBand": "standard", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,54 +52,23 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-03T03:19:53.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-VideoEdit", "docUrl": "https://help.aliyun.com/document_detail/3021842.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长", - "key": "duration", - "tip": "可以选择与输入视频时长相同，或者指定不大于输入视频的时长" - }, - { - "name": "声音设置", - "key": "audio_setting", - "tip": [ - "auto （默认）：模型根据 prompt 内容智能判断。若提示词涉及声音描述，可能重新生成音频；否则可能保留输入素材的原声。", - "origin：强制保留输入视频的原声，不重新生成。" - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-videoedit\",\n \"input\": {\n \"prompt\": \"将视频中女孩的衣服替换为图片中的衣服\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260403/nlspwm/T2VA_22.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260402/fwjpqf/wan2.7-videoedit-change-clothes.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-videoedit\",\n \"input\": {\n \"prompt\": \"将视频中女孩的衣服替换为图片中的衣服\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260403/nlspwm/T2VA_22.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260402/fwjpqf/wan2.7-videoedit-change-clothes.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3021842.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json index a4634e68..9e5d1660 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json @@ -14,49 +14,37 @@ "description": "图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-background-generation-v2", + "prices": [ + { + "priceUnit": "每张", + "price": "0.08", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-03-22T03:33:37.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "图像背景生成", "docUrl": "https://help.aliyun.com/document_detail/2712497.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-background-generation-v2\",\n \"input\": {\n \"base_image_url\": \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/main_images/new_main_img/a.png\",\n \"ref_image_url\": \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_images/c5e50d27be534709817b2ab080b0162f_0.jpg\",\n \"ref_prompt\": \"山脉和晚霞\",\n \"reference_edge\": {\n \"foreground_edge\": [\n \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/huaban_soft_edge/6cdd13941cef1b11d885aea1717b983ae566b8efc9094-vcsvxa_fw658webp.png\",\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/2c36cc4b7da027279e87311dac48fc2d5d784b1e72c0e-x4f1wC_fw658webp.png\"\n ],\n \"background_edge\": [\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/0718a9741e07c52ca5506e75c4f2b99e22fff68a4c7d3-P9WGLr_fw658webp.png\"\n ],\n \"foreground_edge_prompt\": [\n \"粉色桃花\",\n \"可爱小狗\"\n ],\n \"background_edge_prompt\": [\n \"树叶\"\n ]\n }\n },\n \"parameters\": {\n \"n\": 4,\n \"ref_prompt_weight\": 0.5,\n \"model_version\": \"v3\"\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-background-generation-v2\",\n \"input\": {\n \"base_image_url\": \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/main_images/new_main_img/a.png\",\n \"ref_image_url\": \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_images/c5e50d27be534709817b2ab080b0162f_0.jpg\",\n \"ref_prompt\": \"山脉和晚霞\",\n \"reference_edge\": {\n \"foreground_edge\": [\n \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/huaban_soft_edge/6cdd13941cef1b11d885aea1717b983ae566b8efc9094-vcsvxa_fw658webp.png\",\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/2c36cc4b7da027279e87311dac48fc2d5d784b1e72c0e-x4f1wC_fw658webp.png\"\n ],\n \"background_edge\": [\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/0718a9741e07c52ca5506e75c4f2b99e22fff68a4c7d3-P9WGLr_fw658webp.png\"\n ],\n \"foreground_edge_prompt\": [\n \"粉色桃花\",\n \"可爱小狗\"\n ],\n \"background_edge_prompt\": [\n \"树叶\"\n ]\n }\n },\n \"parameters\": {\n \"n\": 4,\n \"ref_prompt_weight\": 0.5,\n \"model_version\": \"v3\"\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json index e6b46d90..b741bf9a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json @@ -14,49 +14,19 @@ "description": "创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-poster-generation-v1", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-21T02:49:20.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "创意海报生成", "docUrl": "https://help.aliyun.com/document_detail/2807172.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\":\"wanx-poster-generation-v1\",\n \"input\": {\n \"title\":\"春节快乐\",\n \"sub_title\":\"家庭团聚，共享天伦之乐\",\n \"body_text\":\"春节是中国最重要的传统节日之一，它象征着新的开始和希望\",\n \"prompt_text_zh\":\"灯笼，小猫，梅花\",\n \"wh_ratios\":\"竖版\",\n \"lora_name\":\"童话油画\",\n \"lora_weight\":0.8,\n \"ctrl_ratio\":0.7,\n \"ctrl_step\":0.7,\n \"generate_mode\":\"generate\",\n \"generate_num\":1\n },\n \"parameters\":{}\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\":\"wanx-poster-generation-v1\",\n \"input\": {\n \"title\":\"春节快乐\",\n \"sub_title\":\"家庭团聚，共享天伦之乐\",\n \"body_text\":\"春节是中国最重要的传统节日之一，它象征着新的开始和希望\",\n \"prompt_text_zh\":\"灯笼，小猫，梅花\",\n \"wh_ratios\":\"竖版\",\n \"lora_name\":\"童话油画\",\n \"lora_weight\":0.8,\n \"ctrl_ratio\":0.7,\n \"ctrl_step\":0.7,\n \"generate_mode\":\"generate\",\n \"generate_num\":1\n },\n \"parameters\":{}\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json index 96a6474d..10bfcb49 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json @@ -14,51 +14,33 @@ "description": "万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-sketch-to-image-lite", + "prices": [ + { + "priceUnit": "每张", + "price": "0.06", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-11T11:21:09.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "万相-涂鸦作画", "docUrl": "https://help.aliyun.com/document_detail/2712498.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-sketch-to-image-lite\",\n \"input\": {\n \"sketch_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\",\n \"prompt\": \"一棵参天大树\"\n },\n \"parameters\": {\n \"size\": \"768*768\",\n \"n\": 2,\n \"sketch_weight\": 3,\n \"style\": \"\"\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一棵参天大树\"\nsketch_image_url = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\"\nmodel = \"wanx-sketch-to-image-lite\"\ntask = \"image2image\"\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=model,\n prompt=prompt,\n n=1,\n style='',\n size='768*768',\n sketch_image_url=sketch_image_url,\n task=task)\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp.output)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String prompt = \"一棵参天大树\";\n String sketchImageUrl = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\";\n String model = \"wanx-sketch-to-image-lite\";\n ImageSynthesisParam param = ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"768*768\")\n .sketchImageUrl(sketchImageUrl)\n .style(\"\")\n .build();\n\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-sketch-to-image-lite\",\n \"input\": {\n \"sketch_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\",\n \"prompt\": \"一棵参天大树\"\n },\n \"parameters\": {\n \"size\": \"768*768\",\n \"n\": 2,\n \"sketch_weight\": 3,\n \"style\": \"\"\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一棵参天大树\"\nsketch_image_url = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\"\nmodel = \"wanx-sketch-to-image-lite\"\ntask = \"image2image\"\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=model,\n prompt=prompt,\n n=1,\n style='',\n size='768*768',\n sketch_image_url=sketch_image_url,\n task=task)\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp.output)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String prompt = \"一棵参天大树\";\n String sketchImageUrl = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\";\n String model = \"wanx-sketch-to-image-lite\";\n ImageSynthesisParam param = ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"768*768\")\n .sketchImageUrl(sketchImageUrl)\n .style(\"\")\n .build();\n\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json index a1c224b1..da9ab951 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json @@ -14,49 +14,31 @@ "description": "人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-style-repaint-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.12", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-03-22T03:32:59.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "人像风格重绘", "docUrl": "https://help.aliyun.com/document_detail/2712493.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-style-repaint-v1\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\",\n \"style_index\": 3\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-style-repaint-v1\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\",\n \"style_index\": 3\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json index c8356a15..b1b06280 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json @@ -14,49 +14,19 @@ "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-virtualmodel", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-25T15:19:20.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "虚拟模特", "docUrl": "https://help.aliyun.com/document_detail/2796985.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-virtualmodel\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-virtualmodel\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json index 564e557a..8e9f4d14 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json @@ -14,51 +14,21 @@ "description": "万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-x-painting", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-05-28T10:45:39.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "万相-图像局部重绘", "docUrl": "https://help.aliyun.com/document_detail/2797051.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-x-painting\",\n \"input\": {\n \"prompt\": \"一只狗戴着红色眼镜\",\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n },\n \"parameters\": {\n \"size\": \"1024*1024\",\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nprompt = \"一只狗戴着红色眼镜\"\nmodel = \"wanx-x-painting\"\ntask = \"image2image\"\nextra_input = {\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n}\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(model=model,\n prompt=prompt,\n n=1,\n size='1024*1024',\n task=task,\n extra_input=extra_input)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisParam param = genImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n private ImageSynthesisParam genImageSynthesis(){\n HashMap extraInputMap = new HashMap<>();\n extraInputMap.put(\"base_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\");\n extraInputMap.put(\"mask_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\");\n String prompt = \"一只狗戴着红色眼镜\";\n String model = \"wanx-x-painting\";\n return ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"1024*1024\")\n .extraInputs(extraInputMap)\n .build();\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-x-painting\",\n \"input\": {\n \"prompt\": \"一只狗戴着红色眼镜\",\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n },\n \"parameters\": {\n \"size\": \"1024*1024\",\n \"n\": 1\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nprompt = \"一只狗戴着红色眼镜\"\nmodel = \"wanx-x-painting\"\ntask = \"image2image\"\nextra_input = {\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n}\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(model=model,\n prompt=prompt,\n n=1,\n size='1024*1024',\n task=task,\n extra_input=extra_input)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisParam param = genImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n private ImageSynthesisParam genImageSynthesis(){\n HashMap extraInputMap = new HashMap<>();\n extraInputMap.put(\"base_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\");\n extraInputMap.put(\"mask_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\");\n String prompt = \"一只狗戴着红色眼镜\";\n String model = \"wanx-x-painting\";\n return ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"1024*1024\")\n .extraInputs(extraInputMap)\n .build();\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json index 5e744220..e0026151 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json @@ -16,10 +16,16 @@ "description": "万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-vace-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "timeBand": "standard", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -36,47 +42,23 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-05-13T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-VACE-Plus", "docUrl": "https://help.aliyun.com/document_detail/2922183.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-vace-plus\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-vace-plus\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n\n\n\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-vace-plus\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n \n\n public static void main(String[] args) {\n syncCall();\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-vace-plus\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-vace-plus\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n\n\n\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-vace-plus\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n \n\n public static void main(String[] args) {\n syncCall();\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json index af883d86..bde791f5 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json @@ -14,49 +14,37 @@ "description": "WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "wordart-semantic", + "prices": [ + { + "priceUnit": "每张", + "price": "0.24", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T08:30:26.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "WordArt锦书-文字变形", "docUrl": "https://help.aliyun.com/document_detail/2712513.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/semantic' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\": \"wordart-semantic\",\n \"input\": {\n \"text\": \"文字创意\",\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\"\n },\n \"parameters\": {\n \"steps\": 80,\n \"n\": 2,\n \"output_image_ratio\": \"1024x1024\",\n \"font_name\": \"dongfangdakai\"\n }\n}'" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/semantic' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\": \"wordart-semantic\",\n \"input\": {\n \"text\": \"文字创意\",\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\"\n },\n \"parameters\": {\n \"steps\": 80,\n \"n\": 2,\n \"output_image_ratio\": \"1024x1024\",\n \"font_name\": \"dongfangdakai\"\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json index 540893be..a9a14eb7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json @@ -15,49 +15,37 @@ "description": "WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "wordart-texture", + "prices": [ + { + "priceUnit": "每张", + "price": "0.08", + "timeBand": "standard", + "type": "image_number", + "priceName": "图片生成" + } + ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T08:29:05.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "WordArt锦书-文字纹理生成", "docUrl": "https://help.aliyun.com/document_detail/2712510.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/texture' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--data '{\n \"model\": \"wordart-texture\",\n \"input\": {\n \"image\": \n {\n \"image_url\": \"https://dmshared-new.oss-cn-hangzhou.aliyuncs.com/junyan.hjy/wordart/lcy/example.png\"\n },\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\",\n \"texture_style\": \"material\"\n },\n \"parameters\": \n {\n \"image_short_size\": 704,\n \"n\": 2,\n \"alpha_channel\": false\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/texture' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--data '{\n \"model\": \"wordart-texture\",\n \"input\": {\n \"image\": \n {\n \"image_url\": \"https://dmshared-new.oss-cn-hangzhou.aliyuncs.com/junyan.hjy/wordart/lcy/example.png\"\n },\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\",\n \"texture_style\": \"material\"\n },\n \"parameters\": \n {\n \"image_short_size\": 704,\n \"n\": 2,\n \"alpha_channel\": false\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json index f2a2f453..74d5362f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json @@ -18,10 +18,30 @@ "cache" ], "provider": "xiaomi", - "limit": { - "message": "model not exist" - }, "model": "xiaomi/mimo-v2.5-pro", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.4", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -40,6 +60,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -53,34 +76,13 @@ "inferenceProvider": "xiaomi", "name": "xiaomi/mimo-v2.5-pro", "docUrl": "https://help.aliyun.com/document_detail/3033942.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], "samples": { "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"xiaomi/mimo-v2.5-pro\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html" + "default": { + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"xiaomi/mimo-v2.5-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"xiaomi/mimo-v2.5-pro\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// Initialize the OpenAI client\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // Read from the environment variable\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'xiaomi/mimo-v2.5-pro',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + 'Full response' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3033942.html" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json index 5ad27363..d22f5a8e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json @@ -17,10 +17,23 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "z-image-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.1", + "timeBand": "standard", + "type": "image_standard", + "priceName": "图片生成（标准）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "timeBand": "standard", + "type": "image_thinking", + "priceName": "图片生成（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -33,37 +46,17 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2025-12-18T06:43:39.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Z-Image-Turbo", "docUrl": "https://help.aliyun.com/document_detail/3002354.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1024*1024", - "tip": "请先选择输出分辨率，再选择输出宽高比" - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json index fe07e72c..3c10f860 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json @@ -19,10 +19,30 @@ "cache" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "ZHIPU/GLM-5.2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -41,6 +61,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -54,37 +77,10 @@ "inferenceProvider": "zhipu-ai", "name": "ZHIPU/GLM-5.2", "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3026315.html" } } @@ -107,10 +103,30 @@ "prefix-completion" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "ZHIPU/GLM-5.1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -129,6 +145,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -142,33 +161,10 @@ "inferenceProvider": "zhipu-ai", "name": "ZHIPU/GLM-5.1", "docUrl": "https://help.aliyun.com/zh/model-studio/glm-zhipu", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3026315.html" } } @@ -191,11 +187,31 @@ "prefix-completion" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "ZHIPU/GLM-5", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "timeBand": "standard", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "22", + "timeBand": "standard", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "timeBand": "standard", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -214,6 +230,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -228,33 +247,10 @@ "inferenceProvider": "zhipu-ai", "name": "ZHIPU/GLM-5", "docUrl": "https://help.aliyun.com/document_detail/3026315.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3026315.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/index.json b/skills/bailian-docs-llm-wiki/models/index.json index c5024785..e2794732 100644 --- a/skills/bailian-docs-llm-wiki/models/index.json +++ b/skills/bailian-docs-llm-wiki/models/index.json @@ -1,10 +1,10 @@ { - "updatedAt": "2026-07-16", - "totalFamilies": 170, - "totalModels": 385, + "updatedAt": "2026-07-29", + "totalFamilies": 172, + "totalModels": 389, "capabilityDistribution": { - "TG": 35, - "IG": 30, + "TG": 36, + "IG": 31, "VG": 26, "TTS": 16, "Reasoning": 14, @@ -21,7 +21,7 @@ "3D-generation": 1 }, "providerDistribution": { - "qwen": 100, + "qwen": 103, "qwen-domain-model": 34, "wan": 13, "happyhorse": 4, @@ -635,20 +635,21 @@ "primaryCapability": "TG", "capabilities": [ "TG", - "Reasoning", - "VU" + "VU", + "Reasoning" ], "providers": [ "moonshot-ai" ], - "itemCount": 4, + "itemCount": 5, "items": [ "kimi/kimi-k2.5", "kimi/kimi-k2.6", "kimi/kimi-k2.7-code", - "kimi/kimi-k2.7-code-highspeed" + "kimi/kimi-k2.7-code-highspeed", + "kimi/kimi-k3" ], - "maxContextWindow": 262144 + "maxContextWindow": 1048576 }, { "slug": "kling-models-market-place", @@ -974,7 +975,7 @@ "items": [ "qwen-audio-3.0-realtime-flash" ], - "maxContextWindow": 8192 + "maxContextWindow": 40960 }, { "slug": "qwen-audio-realtime-plus", @@ -990,7 +991,7 @@ "items": [ "qwen-audio-3.0-realtime-plus" ], - "maxContextWindow": 8192 + "maxContextWindow": 40960 }, { "slug": "qwen-audio-tts", @@ -1080,17 +1081,20 @@ "TR" ], "providers": [ + "qwen", "qwen-domain-model" ], - "itemCount": 6, + "itemCount": 7, "items": [ + "qwen3.7-text-embedding", "text-embedding-async-v1", "text-embedding-async-v2", "text-embedding-v1", "text-embedding-v2", "text-embedding-v3", "text-embedding-v4" - ] + ], + "maxContextWindow": 131072 }, { "slug": "qwen-flash-character", @@ -1155,6 +1159,21 @@ "qwen-image-2.0" ] }, + { + "slug": "qwen-image-3.0-pro", + "name": "Qwen-Image-3.0-Pro", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-image-3.0-pro" + ] + }, { "slug": "qwen-image-edit-max", "name": "Qwen-Image-Edit-Max", @@ -1983,8 +2002,8 @@ "primaryCapability": "Reasoning", "capabilities": [ "Reasoning", - "VU", - "TG" + "TG", + "VU" ], "providers": [ "qwen" @@ -2202,6 +2221,24 @@ ], "maxContextWindow": 262144 }, + { + "slug": "qwen3.7-flash", + "name": "Qwen3.7-Flash", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "VU", + "Reasoning" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.7-flash" + ], + "maxContextWindow": 1000000 + }, { "slug": "qwen3.7-max", "name": "Qwen3.7-Max", @@ -2403,7 +2440,8 @@ "name": "StepFun推理模型", "primaryCapability": "TG", "capabilities": [ - "TG" + "TG", + "VU" ], "providers": [ "stepfun" diff --git a/skills/bailian-docs-llm-wiki/models/index.md b/skills/bailian-docs-llm-wiki/models/index.md index 16b4073f..1961349d 100644 --- a/skills/bailian-docs-llm-wiki/models/index.md +++ b/skills/bailian-docs-llm-wiki/models/index.md @@ -1,6 +1,6 @@ # 百炼模型市场索引 -> 自动生成 · 共 170 个模型家族 · 385 个主干模型 · 更新于 2026-07-16 +> 自动生成 · 共 172 个模型家族 · 389 个主干模型 · 更新于 2026-07-29 **机器查询走结构化文件**： @@ -11,7 +11,7 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[].slug`。 -## 文本生成 `TG` — 35 个家族 +## 文本生成 `TG` — 36 个家族 - [GLM](groups/glm-4.5.json) — GLM是由智谱提供的开源模型。 - 模型：`glm-4.5`, `glm-4.5-air`, `glm-4.6`, `glm-4.7`, `glm-5`, `glm-5.1`, `glm-5.2` @@ -20,7 +20,7 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - [Kimi](groups/Kimi-K2.json) — Kimi是由月之暗面提供的开源模型，包含k2.7-code、k2.6、k2.5、k2-thinking、k2-instruct等多模态和大语言模型。 - 模型：`kimi-k2-thinking`, `kimi-k2.5`, `kimi-k2.6`, `kimi-k2.7-code`, `Moonshot-Kimi-K2-Instruct` - [Kimi](groups/kimi-models-market-place.json) — 由月之暗面提供的Kimi系列模型的API服务。 - - 模型：`kimi/kimi-k2.5`, `kimi/kimi-k2.6`, `kimi/kimi-k2.7-code`, `kimi/kimi-k2.7-code-highspeed` + - 模型：`kimi/kimi-k2.5`, `kimi/kimi-k2.6`, `kimi/kimi-k2.7-code`, `kimi/kimi-k2.7-code-highspeed`, `kimi/kimi-k3` - [MiMo文本模型](groups/xiaomi-models-market-place.json) — 由小米MiMo提供的MiMo文本模型API服务 - 模型：`xiaomi/mimo-v2.5-pro` - [MiniMax文本模型](groups/minimax-models-market-place.json) — 由MiniMax提供的MiniMax-M系列文本模型API服务。 @@ -65,6 +65,8 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - 模型：`qwen3-max`, `qwen3-max-preview` - [Qwen3.5-Plus](groups/qwen3.5-plus.json) — Qwen3.5原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 - 模型：`qwen3.5-plus` +- [Qwen3.7-Flash](groups/qwen3.7-flash.json) — Qwen3.7原生视觉语言系列Flash模型，相较3.6-Flash全面提升多模态理解与Agent执行能力。重点强化多模态基础能力、万物识别能力更强，真实世界感知与空间智能进一步提升，Search A… + - 模型：`qwen3.7-flash` - [Qwen3.7-Plus](groups/qwen3.7-plus.json) — Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真… - 模型：`qwen3.7-plus` - [SiliconFlow DeepSeek](groups/siliconflow-models.json) — 由硅基流动提供的DeepSeek系列模型API服务。 @@ -84,7 +86,7 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - [通义法睿-Plus-32K](groups/farui-plus.json) — 通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分… - 模型：`farui-plus` -## 图像生成 `IG` — 30 个家族 +## 图像生成 `IG` — 31 个家族 - [AI试衣-Plus版](groups/aitryon-plus.json) — aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服… - 模型：`aitryon-plus` @@ -102,6 +104,8 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - 模型：`qwen-image-2.0` - [Qwen-Image-2.0-Pro](groups/qwen-image-2.0-pro.json) — Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系… - 模型：`qwen-image-2.0-pro` +- [Qwen-Image-3.0-Pro](groups/qwen-image-3.0-pro.json) — 内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。 细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实… + - 模型：`qwen-image-3.0-pro` - [Qwen-Image-Edit-Max](groups/qwen-image-edit-max.json) — 千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。 - 模型：`qwen-image-edit-max` - [Qwen-Image-Edit-Plus](groups/qwen-image-edit.json) — 千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。 @@ -390,7 +394,7 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. ## 翻译 `TR` — 2 个家族 - [Qwen-Embedding](groups/qwen-embedding.json) — 基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度… - - 模型：`text-embedding-async-v1`, `text-embedding-async-v2`, `text-embedding-v1`, `text-embedding-v2`, `text-embedding-v3`, `text-embedding-v4` + - 模型：`qwen3.7-text-embedding`, `text-embedding-async-v1`, `text-embedding-async-v2`, `text-embedding-v1`, `text-embedding-v2`, `text-embedding-v3`, `text-embedding-v4` - [Qwen-Rerank](groups/qwen-rerank.json) — 基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。 - 模型：`gte-rerank-v2`, `qwen3-rerank`, `qwen3-vl-rerank` diff --git a/skills/bailian-docs-llm-wiki/models/models.jsonl b/skills/bailian-docs-llm-wiki/models/models.jsonl index 5e7d3d5e..1729d270 100644 --- a/skills/bailian-docs-llm-wiki/models/models.jsonl +++ b/skills/bailian-docs-llm-wiki/models/models.jsonl @@ -1,28 +1,28 @@ -{"model":"kimi-k2-thinking","name":"Kimi-K2-Thinking","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning"],"features":["model-experience","cache","function-calling"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.5","name":"Kimi-K2.5","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.6","name":"Kimi-K2.6","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["cache","function-calling","model-experience"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.7-code","name":"kimi-k2.7-code","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"Moonshot-Kimi-K2-Instruct","name":"Moonshot-Kimi-K2-Instruct","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG"],"features":["model-experience","cache","function-calling"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"MiniMax-M2.1","name":"MiniMax-M2.1","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":172032,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} -{"model":"MiniMax-M2.5","name":"MiniMax-M2.5","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":196608,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} -{"model":"MiniMax/speech-02-hd","name":"speech-02-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-02-turbo","name":"speech-02-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-2.8-hd","name":"speech-2.8-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-2.8-turbo","name":"speech-2.8-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","family":"aitryon-parsing-v1","familyName":"AI试衣OutfitAnyone-图片分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865249.html","detailPath":"groups/aitryon-parsing-v1.json"} -{"model":"aitryon-plus","name":"AI试衣-Plus版","family":"aitryon-plus","familyName":"AI试衣-Plus版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881846.html","detailPath":"groups/aitryon-plus.json"} +{"model":"kimi-k2-thinking","name":"Kimi-K2-Thinking","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning"],"features":["model-experience","cache","function-calling"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.5","name":"Kimi-K2.5","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"5"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.6","name":"Kimi-K2.6","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["cache","function-calling","model-experience"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"8.125"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.65"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.7-code","name":"kimi-k2.7-code","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"8.125"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.65"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"Moonshot-Kimi-K2-Instruct","name":"Moonshot-Kimi-K2-Instruct","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG"],"features":["model-experience","cache","function-calling"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"MiniMax-M2.1","name":"MiniMax-M2.1","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":172032,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} +{"model":"MiniMax-M2.5","name":"MiniMax-M2.5","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":196608,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} +{"model":"MiniMax/speech-02-hd","name":"speech-02-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"3.5"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-02-turbo","name":"speech-02-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-2.8-hd","name":"speech-2.8-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"3.5"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-2.8-turbo","name":"speech-2.8-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","family":"aitryon-parsing-v1","familyName":"AI试衣OutfitAnyone-图片分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865249.html","detailPath":"groups/aitryon-parsing-v1.json"} +{"model":"aitryon-plus","name":"AI试衣-Plus版","family":"aitryon-plus","familyName":"AI试衣-Plus版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881846.html","detailPath":"groups/aitryon-plus.json"} {"model":"aitryon-refiner","name":"AI试衣OutfitAnyone-图片精修","family":"aitryon-refiner","familyName":"AI试衣OutfitAnyone-图片精修","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796663.html","detailPath":"groups/aitryon-refiner.json"} -{"model":"aitryon","name":"AI试衣-基础版","family":"aitryon","familyName":"AI试衣-基础版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796626.html","detailPath":"groups/aitryon.json"} -{"model":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","family":"animate-anyone-detect-gen2","familyName":"舞动人像AnimateAnyone-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786465.html","detailPath":"groups/animate-anyone-detect-gen2.json"} -{"model":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","family":"animate-anyone-gen2","familyName":"舞动人像AnimateAnyone","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786464.html","detailPath":"groups/animate-anyone-gen2.json"} -{"model":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","family":"animate-anyone-template-gen2","familyName":"舞动人像AnimateAnyone-template","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807955.html","detailPath":"groups/animate-anyone-template-gen2.json"} -{"model":"cosyvoice-clone-v1","name":"声音复刻CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v1","name":"语音合成CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v2","name":"语音生成cosyvoice-v2大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3-flash","name":"语音生成CosyVoice-v3-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3-plus","name":"语音生成CosyVoice-v3-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3.5-flash","name":"语音生成CosyVoice-v3.5-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3.5-plus","name":"语音生成CosyVoice-v3.5-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} +{"model":"aitryon","name":"AI试衣-基础版","family":"aitryon","familyName":"AI试衣-基础版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796626.html","detailPath":"groups/aitryon.json"} +{"model":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","family":"animate-anyone-detect-gen2","familyName":"舞动人像AnimateAnyone-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786465.html","detailPath":"groups/animate-anyone-detect-gen2.json"} +{"model":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","family":"animate-anyone-gen2","familyName":"舞动人像AnimateAnyone","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786464.html","detailPath":"groups/animate-anyone-gen2.json"} +{"model":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","family":"animate-anyone-template-gen2","familyName":"舞动人像AnimateAnyone-template","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807955.html","detailPath":"groups/animate-anyone-template-gen2.json"} +{"model":"cosyvoice-clone-v1","name":"声音复刻CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v1","name":"语音合成CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v2","name":"语音生成cosyvoice-v2大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3-flash","name":"语音生成CosyVoice-v3-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3-plus","name":"语音生成CosyVoice-v3-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3.5-flash","name":"语音生成CosyVoice-v3.5-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3.5-plus","name":"语音生成CosyVoice-v3.5-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1.5"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} {"model":"deepseek-r1","name":"DeepSeek-R1","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"},{"type":"input_token_batch","unit":"每百万tokens","price":"2"},{"type":"output_token_batch","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-r1-0528","name":"DeepSeek-R1-0528","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-r1-distill-qwen-1.5b","name":"DeepSeek-R1-Distill-Qwen-1.5B","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} @@ -31,183 +31,186 @@ {"model":"deepseek-r1-distill-qwen-7b","name":"DeepSeek-R1-Distill-Qwen-7B","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v3","name":"DeepSeek-V3","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","web-search","cache"],"contextWindow":65536,"maxInputTokens":57344,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"1"},{"type":"output_token_batch","unit":"每百万tokens","price":"4"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v3.1","name":"DeepSeek-V3.1","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","web-search","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} -{"model":"deepseek-v3.2","name":"DeepSeek-V3.2","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","web-search","cache","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"1"},{"type":"output_token_batch","unit":"每百万tokens","price":"1.5"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"2.5"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.2"},{"type":"input_token_batch_chat","unit":"每百万tokens","price":"2"},{"type":"output_token_batch_chat","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":3750,"count_limit_period":15,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":3750,"count_limit_period":15,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} +{"model":"deepseek-v3.2","name":"DeepSeek-V3.2","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","web-search","cache","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"1"},{"type":"output_token_batch","unit":"每百万tokens","price":"1.5"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"2.5"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.2"},{"type":"input_token_batch_chat","unit":"每百万tokens","price":"2"},{"type":"output_token_batch_chat","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":15000,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v3.2-exp","name":"Deepseek-V3.2-Exp","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","web-search"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v4-flash","name":"DeepSeek-V4-Flash","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","web-search"],"contextWindow":1000000,"maxInputTokens":1000000,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":750,"count_limit_period":3,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":750,"count_limit_period":3,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v4-pro","name":"DeepSeek-V4-Pro","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["cache","function-calling","model-experience","web-search"],"contextWindow":1000000,"maxInputTokens":1000000,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"24"},{"type":"input_token_cache","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":750,"count_limit_period":3,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":750,"count_limit_period":3,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} -{"model":"multimodal-embedding-v1","name":"通用多模态向量","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"contextWindow":0,"maxInputTokens":512,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"tongyi-embedding-vision-flash","name":"视觉向量-flash","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"tongyi-embedding-vision-plus","name":"视觉向量-plus","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"emo-detect-v1","name":"悦动人像EMO-detect","family":"emo-detect-v1","familyName":"悦动人像EMO-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786463.html","detailPath":"groups/emo-detect-v1.json"} -{"model":"emo-v1","name":"悦动人像EMO","family":"emo-v1","familyName":"悦动人像EMO","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786461.html","detailPath":"groups/emo-v1.json"} -{"model":"emoji-detect-v1","name":"表情包Emoji-detect","family":"emoji-detect-v1","familyName":"表情包Emoji-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865371.html","detailPath":"groups/emoji-detect-v1.json"} -{"model":"emoji-v1","name":"表情包Emoji","family":"emoji-v1","familyName":"表情包Emoji","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865374.html","detailPath":"groups/emoji-v1.json"} +{"model":"multimodal-embedding-v1","name":"通用多模态向量","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"contextWindow":0,"maxInputTokens":512,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.9"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"tongyi-embedding-vision-flash","name":"视觉向量-flash","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.15"},{"type":"embedding_token","unit":"每百万tokens","price":"0.15"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"tongyi-embedding-vision-plus","name":"视觉向量-plus","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.5"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"emo-detect-v1","name":"悦动人像EMO-detect","family":"emo-detect-v1","familyName":"悦动人像EMO-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786463.html","detailPath":"groups/emo-detect-v1.json"} +{"model":"emo-v1","name":"悦动人像EMO","family":"emo-v1","familyName":"悦动人像EMO","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_duration_1-1","unit":"每秒","price":"0.08"},{"type":"video_duration_3-4","unit":"每秒","price":"0.16"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786461.html","detailPath":"groups/emo-v1.json"} +{"model":"emoji-detect-v1","name":"表情包Emoji-detect","family":"emoji-detect-v1","familyName":"表情包Emoji-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865371.html","detailPath":"groups/emoji-detect-v1.json"} +{"model":"emoji-v1","name":"表情包Emoji","family":"emoji-v1","familyName":"表情包Emoji","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865374.html","detailPath":"groups/emoji-v1.json"} {"model":"facechain-facedetect","name":"FaceChain人物图像检测","family":"facechain-facedetect","familyName":"FaceChain人物图像检测","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712507.html","detailPath":"groups/facechain-facedetect.json"} -{"model":"facechain-generation","name":"FaceChain人物写真生成","family":"facechain-generation","familyName":"FaceChain人物写真生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712501.html","detailPath":"groups/facechain-generation.json"} -{"model":"farui-plus","name":"通义法睿-Plus-32K","family":"farui-plus","familyName":"通义法睿-Plus-32K","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":12000,"maxInputTokens":12000,"maxOutputTokens":2000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60},"model-default":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2778998.html","detailPath":"groups/farui-plus.json"} -{"model":"fun-asr-flash-2026-06-15","name":"Fun-ASR-Flash-2026-06-15","family":"fun-asr-flash","familyName":"Fun-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/fun-asr-flash.json"} -{"model":"fun-asr-flash-8k-realtime","name":"Fun-ASR-Flash-8k实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} -{"model":"fun-asr-realtime","name":"Fun-ASR实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} -{"model":"fun-asr","name":"Fun-ASR语音识别","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880903.html","detailPath":"groups/fun-asr.json"} -{"model":"fun-asr-mtl","name":"Fun-ASR-MTL","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2978300.html","detailPath":"groups/fun-asr.json"} -{"model":"fun-music-preview","name":"音乐生成 Preview","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} -{"model":"fun-music-v1","name":"音乐生成","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} +{"model":"facechain-generation","name":"FaceChain人物写真生成","family":"facechain-generation","familyName":"FaceChain人物写真生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.18"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712501.html","detailPath":"groups/facechain-generation.json"} +{"model":"farui-plus","name":"通义法睿-Plus-32K","family":"farui-plus","familyName":"通义法睿-Plus-32K","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":12000,"maxInputTokens":12000,"maxOutputTokens":2000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"20"},{"type":"output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60},"model-default":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2778998.html","detailPath":"groups/farui-plus.json"} +{"model":"fun-asr-flash-2026-06-15","name":"Fun-ASR-Flash-2026-06-15","family":"fun-asr-flash","familyName":"Fun-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/fun-asr-flash.json"} +{"model":"fun-asr-flash-8k-realtime","name":"Fun-ASR-Flash-8k实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} +{"model":"fun-asr-realtime","name":"Fun-ASR实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00033"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} +{"model":"fun-asr","name":"Fun-ASR语音识别","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880903.html","detailPath":"groups/fun-asr.json"} +{"model":"fun-asr-mtl","name":"Fun-ASR-MTL","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2978300.html","detailPath":"groups/fun-asr.json"} +{"model":"fun-music-preview","name":"音乐生成 Preview","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.005"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} +{"model":"fun-music-v1","name":"音乐生成","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.002"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} {"model":"glm-4.5","name":"GLM-4.5","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-4.5-air","name":"GLM-4.5-Air","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-4.6","name":"GLM-4.6","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["Reasoning","TG"],"features":["model-experience","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-4.7","name":"GLM-4.7","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-5","name":"GLM-5","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://www.alibabacloud.com/help/zh/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-5.1","name":"GLM-5.1","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":202745,"maxInputTokens":202745,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://www.alibabacloud.com/help/zh/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-5.2","name":"GLM-5.2","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-5.2-fast-preview","name":"GLM-5.2-Fast-Preview","family":"glm-fast","familyName":"GLM-5.2-Fast","provider":"zhipu-ai","capabilities":["TG"],"features":["cache","function-calling","prefix-completion","structured-outputs","web-search"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-fast.json"} -{"model":"gui-plus","name":"GUI-Plus","family":"gui-plus","familyName":"GUI-Plus","provider":"qwen-domain-model","capabilities":["VU"],"features":[],"contextWindow":256000,"maxInputTokens":254976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2997010.html","detailPath":"groups/gui-plus.json"} -{"model":"gummy-chat-v1","name":"一句话识别及翻译V1.0","family":"gummy-chat-v1","familyName":"一句话识别及翻译V1.0","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2866122.html","detailPath":"groups/gummy-chat-v1.json"} -{"model":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","family":"gummy-realtime-v1","familyName":"实时语音识别及翻译V1.0","provider":"qwen","capabilities":["Realtime-Audio-Translate"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865393.html","detailPath":"groups/gummy-realtime-v1.json"} -{"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} -{"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} -{"model":"happyhorse-1.0-r2v","name":"HappyHorse-1.0-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} -{"model":"happyhorse-1.1-r2v","name":"HappyHorse-1.1-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} -{"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} -{"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} -{"model":"happyhorse-1.0-video-edit","name":"HappyHorse-1.0-Video-Edit","family":"happyhorse-video-edit","familyName":"HappyHorse-Video-Edit","provider":"happyhorse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030779.html","detailPath":"groups/happyhorse-video-edit.json"} +{"model":"glm-5.2","name":"GLM-5.2","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-5.2-fast-preview","name":"GLM-5.2-Fast-Preview","family":"glm-fast","familyName":"GLM-5.2-Fast","provider":"zhipu-ai","capabilities":["TG"],"features":["cache","function-calling","prefix-completion","structured-outputs","web-search"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"16"},{"type":"output_token","unit":"每百万tokens","price":"56"},{"type":"input_token_cache","unit":"每百万tokens","price":"4"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-fast.json"} +{"model":"gui-plus","name":"GUI-Plus","family":"gui-plus","familyName":"GUI-Plus","provider":"qwen-domain-model","capabilities":["VU"],"features":[],"contextWindow":256000,"maxInputTokens":254976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.5"},{"type":"output_token","unit":"每百万tokens","price":"4.5"}],"qpmInfo":{"model-default-actual":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2997010.html","detailPath":"groups/gui-plus.json"} +{"model":"gummy-chat-v1","name":"一句话识别及翻译V1.0","family":"gummy-chat-v1","familyName":"一句话识别及翻译V1.0","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00015"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2866122.html","detailPath":"groups/gummy-chat-v1.json"} +{"model":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","family":"gummy-realtime-v1","familyName":"实时语音识别及翻译V1.0","provider":"qwen","capabilities":["Realtime-Audio-Translate"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00015"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865393.html","detailPath":"groups/gummy-realtime-v1.json"} +{"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} +{"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} +{"model":"happyhorse-1.0-r2v","name":"HappyHorse-1.0-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} +{"model":"happyhorse-1.1-r2v","name":"HappyHorse-1.1-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} +{"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} +{"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} +{"model":"happyhorse-1.0-video-edit","name":"HappyHorse-1.0-Video-Edit","family":"happyhorse-video-edit","familyName":"HappyHorse-Video-Edit","provider":"happyhorse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030779.html","detailPath":"groups/happyhorse-video-edit.json"} {"model":"image-erase-completion","name":"图像擦除补全","family":"image-erase-completion","familyName":"图像擦除补全","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2840907.html","detailPath":"groups/image-erase-completion.json"} {"model":"image-instance-segmentation","name":"人物实例分割","family":"image-instance-segmentation","familyName":"人物实例分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2840906.html","detailPath":"groups/image-instance-segmentation.json"} {"model":"image-out-painting","name":"图像画面扩展","family":"image-out-painting","familyName":"图像画面扩展","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796845.html","detailPath":"groups/image-out-painting.json"} -{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","family":"liveportrait-detect","familyName":"灵动人像LivePortrait-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856727.html","detailPath":"groups/liveportrait-detect.json"} -{"model":"liveportrait","name":"灵动人像LivePortrait","family":"liveportrait","familyName":"灵动人像LivePortrait","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856730.html","detailPath":"groups/liveportrait.json"} -{"model":"MiniMax/MiniMax-M2.1","name":"MiniMax/MiniMax-M2.1","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M2.5","name":"MiniMax/MiniMax-M2.5","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M2.7","name":"MiniMax/MiniMax-M2.7","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M3","name":"MiniMax/MiniMax-M3","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning","VU"],"features":["function-calling","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647","detailPath":"groups/minimax-models-market-place.json"} -{"model":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","family":"paraformer-8k-v1","familyName":"Paraformer语音识别-8k-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v1.json"} -{"model":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","family":"paraformer-8k-v2","familyName":"Paraformer语音识别-8k-v2","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v2.json"} -{"model":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","family":"paraformer-mtl-v1","familyName":"Paraformer语音识别-mtl-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-mtl-v1.json"} -{"model":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","family":"paraformer-realtime-8k-v1","familyName":"Paraformer实时语音识别-8k-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v1.json"} -{"model":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","family":"paraformer-realtime-8k-v2","familyName":"Paraformer实时语音识别-8k-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v2.json"} -{"model":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","family":"paraformer-realtime-v1","familyName":"Paraformer实时语音识别-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v1.json"} -{"model":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","family":"paraformer-realtime-v2","familyName":"Paraformer实时语音识别-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v2.json"} -{"model":"paraformer-v1","name":"Paraformer语音识别-v1","family":"paraformer-v1","familyName":"Paraformer语音识别-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v1.json"} -{"model":"paraformer-v2","name":"Paraformer语音识别-v2","family":"paraformer-v2","familyName":"Paraformer语音识别-v2","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v2.json"} -{"model":"pixverse/pixverse-c1-it2v","name":"PixVerse-C1-it2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-kf2v","name":"PixVerse-C1-kf2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-r2v","name":"PixVerse-C1-r2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-t2v","name":"PixVerse-C1-t2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.1"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"13"},{"type":"output_token","unit":"每百万tokens","price":"54"},{"type":"input_token_cache","unit":"每百万tokens","price":"2.6"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k3","name":"kimi/kimi-k3","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["function-calling","structured-outputs","cache","prefix-completion"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":1048576,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"20"},{"type":"output_token","unit":"每百万tokens","price":"100"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.2"},{"type":"image_type_2k","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.2"},{"type":"image_type_2k","unit":"每张","price":"0.2"},{"type":"image_type_4k","unit":"每秒","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"720P_no_reference_video","unit":"每秒","price":"0.9"},{"type":"1080P_no_reference_video","unit":"每秒","price":"1.2"},{"type":"720P_no_audio_no_reference_video","unit":"每秒","price":"0.6"},{"type":"1080P_no_audio_no_reference_video","unit":"每秒","price":"0.8"},{"type":"720P_no_audio_reference_video","unit":"每秒","price":"0.9"},{"type":"1080P_no_audio_reference_video","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"},{"type":"720P_no_audio","unit":"每秒","price":"0.6"},{"type":"1080P_no_audio","unit":"每秒","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","family":"liveportrait-detect","familyName":"灵动人像LivePortrait-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856727.html","detailPath":"groups/liveportrait-detect.json"} +{"model":"liveportrait","name":"灵动人像LivePortrait","family":"liveportrait","familyName":"灵动人像LivePortrait","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.02"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856730.html","detailPath":"groups/liveportrait.json"} +{"model":"MiniMax/MiniMax-M2.1","name":"MiniMax/MiniMax-M2.1","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.21"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M2.5","name":"MiniMax/MiniMax-M2.5","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.21"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M2.7","name":"MiniMax/MiniMax-M2.7","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M3","name":"MiniMax/MiniMax-M3","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning","VU"],"features":["function-calling","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4.2"},{"type":"output_token","unit":"每百万tokens","price":"16.8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.84"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647","detailPath":"groups/minimax-models-market-place.json"} +{"model":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","family":"paraformer-8k-v1","familyName":"Paraformer语音识别-8k-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v1.json"} +{"model":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","family":"paraformer-8k-v2","familyName":"Paraformer语音识别-8k-v2","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v2.json"} +{"model":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","family":"paraformer-mtl-v1","familyName":"Paraformer语音识别-mtl-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-mtl-v1.json"} +{"model":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","family":"paraformer-realtime-8k-v1","familyName":"Paraformer实时语音识别-8k-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v1.json"} +{"model":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","family":"paraformer-realtime-8k-v2","familyName":"Paraformer实时语音识别-8k-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v2.json"} +{"model":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","family":"paraformer-realtime-v1","familyName":"Paraformer实时语音识别-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v1.json"} +{"model":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","family":"paraformer-realtime-v2","familyName":"Paraformer实时语音识别-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v2.json"} +{"model":"paraformer-v1","name":"Paraformer语音识别-v1","family":"paraformer-v1","familyName":"Paraformer语音识别-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v1.json"} +{"model":"paraformer-v2","name":"Paraformer语音识别-v2","family":"paraformer-v2","familyName":"Paraformer语音识别-v2","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v2.json"} +{"model":"pixverse/pixverse-c1-it2v","name":"PixVerse-C1-it2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-kf2v","name":"PixVerse-C1-kf2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-r2v","name":"PixVerse-C1-r2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-t2v","name":"PixVerse-C1-t2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-c1-market-place.json"} {"model":"pixverse/pixverse-lipsync","name":"pixverse/pixverse-lipsync","family":"pixverse-capability-market-place","familyName":"PixVerse Capabilities","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"video_generation","unit":"每秒","price":"0.12"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-capability-market-place.json"} {"model":"pixverse/pixverse-motioncontrol","name":"pixverse/pixverse-motioncontrol","family":"pixverse-capability-market-place","familyName":"PixVerse Capabilities","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Video"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.27"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-capability-market-place.json"} {"model":"pixverse/pixverse-upscale","name":"pixverse/pixverse-upscale","family":"pixverse-capability-market-place","familyName":"PixVerse Capabilities","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"prices":[{"type":"video_generation","unit":"每秒","price":"0.15"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-capability-market-place.json"} -{"model":"pixverse/pixverse-v5.6-it2v","name":"PixVerse-V5.6-it2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-kf2v","name":"PixVerse-V5.6-kf2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-r2v","name":"PixVerse-V5.6-r2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-t2v","name":"PixVerse-V5.6-t2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"○ https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"qvq-max","name":"QVQ-Max","family":"qvq-max","familyName":"QVQ-Max","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-max.json"} -{"model":"qvq-plus","name":"QVQ-Plus","family":"qvq-plus","familyName":"Qwen-QVQ-Plus","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-plus.json"} -{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","family":"qwen-audio-realtime-flash","familyName":"Qwen-Audio-Realtime-Flash","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":8192,"maxInputTokens":4096,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"30"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"3"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"30"},{"type":"audio_output_token","unit":"每百万tokens","price":"100"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-flash.json"} -{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","family":"qwen-audio-realtime-plus","familyName":"Qwen-Audio-Realtime-Plus","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":8192,"maxInputTokens":4096,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"40"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"5"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"40"},{"type":"audio_output_token","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-plus.json"} +{"model":"pixverse/pixverse-v5.6-it2v","name":"PixVerse-V5.6-it2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-kf2v","name":"PixVerse-V5.6-kf2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-r2v","name":"PixVerse-V5.6-r2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-t2v","name":"PixVerse-V5.6-t2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"○ https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"qvq-max","name":"QVQ-Max","family":"qvq-max","familyName":"QVQ-Max","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"32"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-max.json"} +{"model":"qvq-plus","name":"QVQ-Plus","family":"qvq-plus","familyName":"Qwen-QVQ-Plus","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"5"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-plus.json"} +{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","family":"qwen-audio-realtime-flash","familyName":"Qwen-Audio-Realtime-Flash","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":40960,"maxInputTokens":16384,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"30"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"3"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"30"},{"type":"audio_output_token","unit":"每百万tokens","price":"100"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-flash.json"} +{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","family":"qwen-audio-realtime-plus","familyName":"Qwen-Audio-Realtime-Plus","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":40960,"maxInputTokens":16384,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"40"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"5"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"40"},{"type":"audio_output_token","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-plus.json"} {"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","family":"qwen-audio-tts","familyName":"Qwen-Audio-TTS","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-audio-tts.json"} {"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","family":"qwen-audio-tts","familyName":"Qwen-Audio-TTS","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1.4"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-audio-tts.json"} -{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","family":"qwen-coder-plus","familyName":"Qwen-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-plus.json"} -{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","family":"qwen-coder-turbo","familyName":"Qwen-Coder-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-turbo.json"} -{"model":"qwen-deep-research","name":"qwen-deep-research","family":"qwen-deep-research","familyName":"qwen-deep-research","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975991.html","detailPath":"groups/qwen-deep-research.json"} -{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","family":"qwen-doc-turbo","familyName":"Qwen-Doc-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["cache"],"contextWindow":262144,"maxInputTokens":253952,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948885.html","detailPath":"groups/qwen-doc-turbo.json"} -{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v1","name":"通用文本向量-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v2","name":"通用文本向量-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v3","name":"通用文本向量-v3","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v4","name":"通用文本向量-v4","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-embedding.json"} -{"model":"qwen-flash-character","name":"Qwen-Flash-Character","family":"qwen-flash-character","familyName":"Qwen-Flash-Character","provider":"qwen","capabilities":["TG"],"features":["model-experience","cache","web-search"],"contextWindow":8192,"maxInputTokens":8000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-flash-character.json"} +{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","family":"qwen-coder-plus","familyName":"Qwen-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"3.5"},{"type":"output_token","unit":"每百万tokens","price":"7"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-plus.json"} +{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","family":"qwen-coder-turbo","familyName":"Qwen-Coder-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-turbo.json"} +{"model":"qwen-deep-research","name":"qwen-deep-research","family":"qwen-deep-research","familyName":"qwen-deep-research","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"54"},{"type":"output_token","unit":"每百万tokens","price":"163"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975991.html","detailPath":"groups/qwen-deep-research.json"} +{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","family":"qwen-doc-turbo","familyName":"Qwen-Doc-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["cache"],"contextWindow":262144,"maxInputTokens":253952,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.6"},{"type":"output_token","unit":"每百万tokens","price":"1"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.12"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"0.75"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.06"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948885.html","detailPath":"groups/qwen-doc-turbo.json"} +{"model":"qwen3.7-text-embedding","name":"Qwen3.7-通用文本向量","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen","capabilities":["TR"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2400,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":2400,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v1","name":"通用文本向量-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.35"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v2","name":"通用文本向量-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.35"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v3","name":"通用文本向量-v3","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v4","name":"通用文本向量-v4","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-embedding.json"} +{"model":"qwen-flash-character","name":"Qwen-Flash-Character","family":"qwen-flash-character","familyName":"Qwen-Flash-Character","provider":"qwen","capabilities":["TG"],"features":["model-experience","cache","web-search"],"contextWindow":8192,"maxInputTokens":8000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.25"},{"type":"output_token","unit":"每百万tokens","price":"1.5"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.05"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-flash-character.json"} {"model":"qwen-flash","name":"Qwen-Flash","family":"qwen-flash","familyName":"Qwen-Flash","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":30,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":15000,"count_limit_period":30,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-flash.json"} -{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","family":"qwen-image-2.0-pro","familyName":"Qwen-Image-2.0-Pro","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0-pro.json"} -{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","family":"qwen-image-2.0","familyName":"Qwen-Image-2.0","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0.json"} -{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","family":"qwen-image-edit-max","familyName":"Qwen-Image-Edit-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit-max.json"} -{"model":"qwen-image-edit","name":"Qwen-Image-Edit","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} -{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} -{"model":"qwen-image-max","name":"Qwen-Image-Max","family":"qwen-image-max","familyName":"Qwen-Image-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-max.json"} -{"model":"qwen-image","name":"Qwen-Image","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} -{"model":"qwen-image-plus","name":"Qwen-Image-Plus","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} -{"model":"qwen-long","name":"Qwen-Long","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} -{"model":"qwen-long-latest","name":"Qwen-Long-Latest","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} -{"model":"qwen-math-plus","name":"Qwen-Math-Plus","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-turbo","name":"Qwen-Math-Turbo","family":"qwen-math-turbo","familyName":"Qwen-Math-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-turbo.json"} -{"model":"qwen-max","name":"Qwen-Max","family":"qwen-max","familyName":"Qwen-Max","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-max.json"} -{"model":"qwen-mt-flash","name":"Qwen-MT-Flash","family":"qwen-mt-flash","familyName":"Qwen-MT-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-flash.json"} -{"model":"qwen-mt-image","name":"Qwen-MT-Image","family":"qwen-mt-image","familyName":"Qwen-MT-Image","provider":"qwen-domain-model","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2977163.html","detailPath":"groups/qwen-mt-image.json"} -{"model":"qwen-mt-lite","name":"Qwen-MT-Lite","family":"qwen-mt-lite","familyName":"Qwen-MT-Lite","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-lite.json"} -{"model":"qwen-mt-plus","name":"Qwen-MT-Plus","family":"qwen-mt-plus","familyName":"Qwen-MT-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-plus.json"} -{"model":"qwen-mt-turbo","name":"Qwen-MT-Turbo","family":"qwen-mt-turbo","familyName":"Qwen-MT-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-turbo.json"} -{"model":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} -{"model":"qwen-omni-turbo-realtime-latest","name":"Qwen-Omni-Turbo-Realtime-Latest","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} -{"model":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} -{"model":"qwen-omni-turbo-latest","name":"Qwen-Omni-Turbo-Latest","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} -{"model":"qwen-plus-character","name":"Qwen-Plus-Character","family":"qwen-plus-character","familyName":"Qwen-Plus-Character","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","web-search","cache","structured-outputs"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-plus-character.json"} +{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","family":"qwen-image-2.0-pro","familyName":"Qwen-Image-2.0-Pro","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0-pro.json"} +{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","family":"qwen-image-2.0","familyName":"Qwen-Image-2.0","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0.json"} +{"model":"qwen-image-3.0-pro","name":"Qwen-Image-3.0-Pro","family":"qwen-image-3.0-pro","familyName":"Qwen-Image-3.0-Pro","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3047054.html","detailPath":"groups/qwen-image-3.0-pro.json"} +{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","family":"qwen-image-edit-max","familyName":"Qwen-Image-Edit-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit-max.json"} +{"model":"qwen-image-edit","name":"Qwen-Image-Edit","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.3"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} +{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} +{"model":"qwen-image-max","name":"Qwen-Image-Max","family":"qwen-image-max","familyName":"Qwen-Image-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-max.json"} +{"model":"qwen-image","name":"Qwen-Image","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} +{"model":"qwen-image-plus","name":"Qwen-Image-Plus","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} +{"model":"qwen-long","name":"Qwen-Long","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"output_token_batch","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} +{"model":"qwen-long-latest","name":"Qwen-Long-Latest","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"output_token_batch","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} +{"model":"qwen-math-plus","name":"Qwen-Math-Plus","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-turbo","name":"Qwen-Math-Turbo","family":"qwen-math-turbo","familyName":"Qwen-Math-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-turbo.json"} +{"model":"qwen-max","name":"Qwen-Max","family":"qwen-max","familyName":"Qwen-Max","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"9.6"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.48"},{"type":"input_token_batch","unit":"每百万tokens","price":"1.2"},{"type":"output_token_batch","unit":"每百万tokens","price":"4.8"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-max.json"} +{"model":"qwen-mt-flash","name":"Qwen-MT-Flash","family":"qwen-mt-flash","familyName":"Qwen-MT-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.7"},{"type":"output_token","unit":"每百万tokens","price":"1.95"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-flash.json"} +{"model":"qwen-mt-image","name":"Qwen-MT-Image","family":"qwen-mt-image","familyName":"Qwen-MT-Image","provider":"qwen-domain-model","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.003"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2977163.html","detailPath":"groups/qwen-mt-image.json"} +{"model":"qwen-mt-lite","name":"Qwen-MT-Lite","family":"qwen-mt-lite","familyName":"Qwen-MT-Lite","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.6"},{"type":"output_token","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-lite.json"} +{"model":"qwen-mt-plus","name":"Qwen-MT-Plus","family":"qwen-mt-plus","familyName":"Qwen-MT-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.8"},{"type":"output_token","unit":"每百万tokens","price":"5.4"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-plus.json"} +{"model":"qwen-mt-turbo","name":"Qwen-MT-Turbo","family":"qwen-mt-turbo","familyName":"Qwen-MT-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.7"},{"type":"output_token","unit":"每百万tokens","price":"1.95"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-turbo.json"} +{"model":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"6"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"18"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} +{"model":"qwen-omni-turbo-realtime-latest","name":"Qwen-Omni-Turbo-Realtime-Latest","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"6"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"18"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} +{"model":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.4"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"1.5"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"1.6"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"4.5"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"},{"type":"vision_input_token_cache","unit":"每百万tokens","price":"0.3"},{"type":"text_input_token_cache","unit":"每百万tokens","price":"0.08"},{"type":"audio_input_token_cache","unit":"每百万tokens","price":"5"},{"type":"text_input_token_batch","unit":"每百万tokens","price":"0.2"},{"type":"audio_input_token_batch","unit":"每百万tokens","price":"12.5"},{"type":"vision_input_token_batch","unit":"每百万tokens","price":"0.75"},{"type":"multi_output_token_batch","unit":"每百万tokens","price":"25"},{"type":"multiin_text_output_token_batch","unit":"每百万tokens","price":"2.25"},{"type":"purein_text_output_token_batch","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} +{"model":"qwen-omni-turbo-latest","name":"Qwen-Omni-Turbo-Latest","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.4"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"1.5"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"1.6"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"4.5"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} +{"model":"qwen-plus-character","name":"Qwen-Plus-Character","family":"qwen-plus-character","familyName":"Qwen-Plus-Character","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","web-search","cache","structured-outputs"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.16"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-plus-character.json"} {"model":"qwen-plus","name":"Qwen-Plus","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":30,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":15000,"count_limit_period":30,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"qwen-plus-0112","name":"Qwen-Plus-2025-01-12","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"qwen-plus-1220","name":"Qwen-Plus-2024-12-20","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} +{"model":"qwen-plus-0112","name":"Qwen-Plus-2025-01-12","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} +{"model":"qwen-plus-1220","name":"Qwen-Plus-2024-12-20","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} {"model":"qwen-plus-latest","name":"Qwen-Plus-Latest","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","batch"],"contextWindow":1000000,"maxInputTokens":995904,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":7500,"count_limit_period":30,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":7500,"count_limit_period":30,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"gte-rerank-v2","name":"深度文本重排序","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1},"model-default":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen3-rerank","name":"千问3-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen3-vl-rerank","name":"Qwen3-VL-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"maxInputTokens":120000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} -{"model":"qwen-tts-realtime-latest","name":"Qwen-TTS-Realtime-Latest","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} -{"model":"qwen-tts","name":"Qwen-TTS","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} -{"model":"qwen-tts-latest","name":"Qwen-TTS-Latest","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} -{"model":"qwen-turbo","name":"Qwen-Turbo","family":"qwen-turbo","familyName":"Qwen-Turbo","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","cache","structured-outputs"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-turbo.json"} -{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} -{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":["model-experience"],"maxInputTokens":32000,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} -{"model":"qwen-vl-max","name":"Qwen-VL-Max","family":"qwen-vl-max","familyName":"Qwen-VL-Max","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-max.json"} -{"model":"qwen-vl-ocr","name":"QwenVL-OCR","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience"],"contextWindow":34096,"maxInputTokens":30000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-plus","name":"QwenVL-Plus","family":"qwen-vl-plus","familyName":"Qwen-VL-Plus","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-plus.json"} -{"model":"qwen-voice-design","name":"Qwen-Voice-Design","family":"qwen-voice-design","familyName":"Qwen-声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3000986.html","detailPath":"groups/qwen-voice-design.json"} -{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","family":"qwen-voice-enrollment","familyName":"Qwen-声音复刻","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975034.html","detailPath":"groups/qwen-voice-enrollment.json"} -{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","family":"qwen2.5","familyName":"Qwen2.5-开源模型","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen2.5.json"} -{"model":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","family":"qwen3-asr-flash-filetrans","familyName":"Qwen3-ASR-Flash-Filetrans","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash-filetrans.json"} -{"model":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","family":"qwen3-asr-flash-realtime","familyName":"Qwen3-ASR-Flash-Realtime","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2989727.html","detailPath":"groups/qwen3-asr-flash-realtime.json"} -{"model":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","family":"qwen3-asr-flash","familyName":"Qwen3-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash.json"} +{"model":"gte-rerank-v2","name":"深度文本重排序","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1},"model-default":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen3-rerank","name":"千问3-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen3-vl-rerank","name":"Qwen3-VL-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"maxInputTokens":120000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} +{"model":"qwen-tts-realtime-latest","name":"Qwen-TTS-Realtime-Latest","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} +{"model":"qwen-tts","name":"Qwen-TTS","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"qwen_tts_multi_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} +{"model":"qwen-tts-latest","name":"Qwen-TTS-Latest","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"qwen_tts_multi_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} +{"model":"qwen-turbo","name":"Qwen-Turbo","family":"qwen-turbo","familyName":"Qwen-Turbo","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","cache","structured-outputs"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.6"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.3"},{"type":"thinking_output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.06"},{"type":"thinking_input_token_cache","unit":"每百万tokens","price":"0.06"},{"type":"thinking_output_token_batch","unit":"每百万tokens","price":"1.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.3"},{"type":"thinking_input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"ft","unit":"每百万tokens","price":"30"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-turbo.json"} +{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} +{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":["model-experience"],"maxInputTokens":32000,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} +{"model":"qwen-vl-max","name":"Qwen-VL-Max","family":"qwen-vl-max","familyName":"Qwen-VL-Max","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.6"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.32"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.8"},{"type":"output_token_batch","unit":"每百万tokens","price":"2"},{"type":"ft","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-max.json"} +{"model":"qwen-vl-ocr","name":"QwenVL-OCR","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience"],"contextWindow":34096,"maxInputTokens":30000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"5"},{"type":"output_token","unit":"每百万tokens","price":"5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-vl-plus","name":"QwenVL-Plus","family":"qwen-vl-plus","familyName":"Qwen-VL-Plus","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.16"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.4"},{"type":"output_token_batch","unit":"每百万tokens","price":"1"},{"type":"ft","unit":"每百万tokens","price":"30"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-plus.json"} +{"model":"qwen-voice-design","name":"Qwen-Voice-Design","family":"qwen-voice-design","familyName":"Qwen-声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3000986.html","detailPath":"groups/qwen-voice-design.json"} +{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","family":"qwen-voice-enrollment","familyName":"Qwen-声音复刻","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"tts_vc_model","unit":"次","price":"0.01"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975034.html","detailPath":"groups/qwen-voice-enrollment.json"} +{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","family":"qwen2.5","familyName":"Qwen2.5-开源模型","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"38"},{"type":"vision_input_token","unit":"每百万tokens","price":"2"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"2.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"6"},{"type":"multi_output_token","unit":"每百万tokens","price":"76"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen2.5.json"} +{"model":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","family":"qwen3-asr-flash-filetrans","familyName":"Qwen3-ASR-Flash-Filetrans","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash-filetrans.json"} +{"model":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","family":"qwen3-asr-flash-realtime","familyName":"Qwen3-ASR-Flash-Realtime","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00033"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2989727.html","detailPath":"groups/qwen3-asr-flash-realtime.json"} +{"model":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","family":"qwen3-asr-flash","familyName":"Qwen3-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash.json"} {"model":"qwen3-coder-30b-a3b-instruct","name":"Qwen3-Coder-30B-A3B-Instruct","family":"qwen3-coder-30b-a3b-instruct","familyName":"Qwen3-Coder-30B-A3B-Instruct","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/zh/model-studio/qwen-coder#272bcaccea8ls","detailPath":"groups/qwen3-coder-30b-a3b-instruct.json"} {"model":"qwen3-coder-480b-a35b-instruct","name":"Qwen3-Coder-480B-A35B-Instruct","family":"qwen3-coder-480b-a35b-instruct","familyName":"Qwen3-Coder-480B-A35B-Instruct","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-480b-a35b-instruct.json"} {"model":"qwen3-coder-flash","name":"Qwen3-Coder-Flash","family":"qwen3-coder-flash","familyName":"Qwen3-Coder-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","function-calling","cache"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-flash.json"} {"model":"qwen3-coder-plus","name":"Qwen3-Coder-Plus","family":"qwen3-coder-plus","familyName":"Qwen3-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","function-calling","cache"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-plus.json"} -{"model":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","family":"qwen3-livetranslate-flash-realtime","familyName":"Qwen3-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Image","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3-livetranslate-flash-realtime.json"} -{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","family":"qwen3-livetranslate-flash","familyName":"Qwen3-LiveTranslate-Flash","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Audio","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2999748.html","detailPath":"groups/qwen3-livetranslate-flash.json"} +{"model":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","family":"qwen3-livetranslate-flash-realtime","familyName":"Qwen3-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Image","Audio"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"64"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"8"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"64"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"240"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3-livetranslate-flash-realtime.json"} +{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","family":"qwen3-livetranslate-flash","familyName":"Qwen3-LiveTranslate-Flash","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Audio","Video"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"10"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"4"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"10"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2999748.html","detailPath":"groups/qwen3-livetranslate-flash.json"} {"model":"qwen3-max","name":"Qwen3-Max","family":"qwen3-max","familyName":"Qwen3-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":262144,"maxInputTokens":258048,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3-max.json"} {"model":"qwen3-max-preview","name":"Qwen3-Max-Preview","family":"qwen3-max","familyName":"Qwen3-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["model-experience","cache"],"contextWindow":262144,"maxInputTokens":258048,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3-max.json"} -{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","family":"qwen3-omni-30b-a3b-captioner","familyName":"Qwen3-Omni-30b-a3b-Captioner","provider":"qwen","capabilities":["ASR"],"features":[],"contextWindow":65536,"maxInputTokens":32768,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2980468.html","detailPath":"groups/qwen3-omni-30b-a3b-captioner.json"} -{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","family":"qwen3-omni-flash-realtime","familyName":"Qwen3-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3-omni-flash-realtime.json"} -{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","family":"qwen3-omni-flash","familyName":"Qwen3-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","function-calling"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3-omni-flash.json"} -{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","family":"qwen3-tts-flash-realtime","familyName":"Qwen3-TTS-Flash-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-flash-realtime.json"} -{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","family":"qwen3-tts-flash","familyName":"Qwen3-TTS-Flash","provider":"qwen","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-flash.json"} -{"model":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","family":"qwen3-tts-instruct-flash-realtime","familyName":"qwen3-tts-instruct-flash-realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-instruct-flash-realtime.json"} -{"model":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","family":"qwen3-tts-instruct-flash","familyName":"Qwen3-TTS-Instruct-Flash","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-instruct-flash.json"} -{"model":"qwen3-tts-vc-realtime-2026-01-15","name":"qwen3-tts-vc-realtime-2026-01-15","family":"qwen3-tts-vc-realtime","familyName":"Qwen3-TTS-VC-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vc-realtime.json"} -{"model":"qwen3-tts-vc-2026-01-22","name":"Qwen3-TTS-VC-2026-01-22","family":"qwen3-tts-vc","familyName":"Qwen3-TTS-VC","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vc.json"} -{"model":"qwen3-tts-vd-realtime-2026-01-15","name":"qwen3-tts-vd-realtime-2026-01-15","family":"qwen3-tts-vd-realtime","familyName":"Qwen3-TTS-VD-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vd-realtime.json"} -{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","family":"qwen3-tts-vd","familyName":"Qwen3-TTS-VD","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vd.json"} +{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","family":"qwen3-omni-30b-a3b-captioner","familyName":"Qwen3-Omni-30b-a3b-Captioner","provider":"qwen","capabilities":["ASR"],"features":[],"contextWindow":65536,"maxInputTokens":32768,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"12.7"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2980468.html","detailPath":"groups/qwen3-omni-30b-a3b-captioner.json"} +{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","family":"qwen3-omni-flash-realtime","familyName":"Qwen3-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"2.2"},{"type":"audio_input_token","unit":"每百万tokens","price":"18.9"},{"type":"vision_input_token","unit":"每百万tokens","price":"3.9"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"8.3"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"15.2"},{"type":"multi_output_token","unit":"每百万tokens","price":"75.1"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3-omni-flash-realtime.json"} +{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","family":"qwen3-omni-flash","familyName":"Qwen3-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","function-calling"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.8"},{"type":"audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.9"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"12.7"},{"type":"multi_output_token","unit":"每百万tokens","price":"62.6"},{"type":"thinking_text_input_token","unit":"每百万tokens","price":"1.8"},{"type":"thinking_audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"thinking_vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"thinking_purein_text_output_token","unit":"每百万tokens","price":"6.9"},{"type":"thinking_multiin_text_output_token","unit":"每百万tokens","price":"12.7"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3-omni-flash.json"} +{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","family":"qwen3-tts-flash-realtime","familyName":"Qwen3-TTS-Flash-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-flash-realtime.json"} +{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","family":"qwen3-tts-flash","familyName":"Qwen3-TTS-Flash","provider":"qwen","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-flash.json"} +{"model":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","family":"qwen3-tts-instruct-flash-realtime","familyName":"qwen3-tts-instruct-flash-realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-instruct-flash-realtime.json"} +{"model":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","family":"qwen3-tts-instruct-flash","familyName":"Qwen3-TTS-Instruct-Flash","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-instruct-flash.json"} +{"model":"qwen3-tts-vc-realtime-2026-01-15","name":"qwen3-tts-vc-realtime-2026-01-15","family":"qwen3-tts-vc-realtime","familyName":"Qwen3-TTS-VC-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vc-realtime.json"} +{"model":"qwen3-tts-vc-2026-01-22","name":"Qwen3-TTS-VC-2026-01-22","family":"qwen3-tts-vc","familyName":"Qwen3-TTS-VC","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vc.json"} +{"model":"qwen3-tts-vd-realtime-2026-01-15","name":"qwen3-tts-vd-realtime-2026-01-15","family":"qwen3-tts-vd-realtime","familyName":"Qwen3-TTS-VD-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vd-realtime.json"} +{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","family":"qwen3-tts-vd","familyName":"Qwen3-TTS-VD","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vd.json"} {"model":"qwen3-vl-flash","name":"Qwen3-VL-Flash","family":"qwen3-vl-flash","familyName":"Qwen3-VL-Flash","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":5,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":250,"count_limit_period":5,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3-vl-flash.json"} {"model":"qwen3-vl-plus","name":"Qwen3-VL-Plus","family":"qwen3-vl-plus","familyName":"Qwen3-VL-Plus","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","cache","batch"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":5,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":250,"count_limit_period":5,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3-vl-plus.json"} -{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","family":"qwen3.5-flash","familyName":"Qwen3.5-Flash","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","batch","cache","prefix-completion"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-flash.json"} -{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","family":"qwen3.5-livetranslate-flash-realtime","familyName":"Qwen3.5-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json"} -{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","family":"qwen3.5-ocr","familyName":"Qwen3.5-OCR","provider":"qwen","capabilities":["VU"],"features":["model-experience"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860683.html","detailPath":"groups/qwen3.5-ocr.json"} -{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","family":"qwen3.5-omni-flash-realtime","familyName":"Qwen3.5-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3.5-omni-flash-realtime.json"} -{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","family":"qwen3.5-omni-flash","familyName":"Qwen3.5-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-flash.json"} -{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","family":"qwen3.5-omni-plus-realtime","familyName":"Qwen3.5-Omni-Plus-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus-realtime.json"} -{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","family":"qwen3.5-omni-plus","familyName":"Qwen3.5-Omni-Plus","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus.json"} +{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","family":"qwen3.5-flash","familyName":"Qwen3.5-Flash","provider":"qwen","capabilities":["Reasoning","TG","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","batch","cache","prefix-completion"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-flash.json"} +{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","family":"qwen3.5-livetranslate-flash-realtime","familyName":"Qwen3.5-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Image"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"40"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"100"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"160"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json"} +{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","family":"qwen3.5-ocr","familyName":"Qwen3.5-OCR","provider":"qwen","capabilities":["VU"],"features":["model-experience"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860683.html","detailPath":"groups/qwen3.5-ocr.json"} +{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","family":"qwen3.5-omni-flash-realtime","familyName":"Qwen3.5-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"27"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"107"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"3.3"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3.5-omni-flash-realtime.json"} +{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","family":"qwen3.5-omni-flash","familyName":"Qwen3.5-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"18"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"72"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"2.2"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"13.3"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-flash.json"} +{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","family":"qwen3.5-omni-plus-realtime","familyName":"Qwen3.5-Omni-Plus-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"80"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"300"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"10"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"60"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus-realtime.json"} +{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","family":"qwen3.5-omni-plus","familyName":"Qwen3.5-Omni-Plus","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search","function-calling","batch"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"53"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"213"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"7"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"40"},{"type":"omni_audio_input_token_batch","unit":"每百万tokens","price":"26.5"},{"type":"omni_no_audio_input_token_batch","unit":"每百万tokens","price":"3.5"},{"type":"omni_no_audio_output_token_batch","unit":"每百万tokens","price":"20"},{"type":"omni_audio_input_token_batch_chat","unit":"每百万tokens","price":"53"},{"type":"omni_no_audio_input_token_batch_chat","unit":"每百万tokens","price":"7"},{"type":"omni_no_audio_output_token_batch_chat","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus.json"} {"model":"qwen3.5-plus","name":"Qwen3.5-Plus","family":"qwen3.5-plus","familyName":"Qwen3.5-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-plus.json"} {"model":"qwen3.5-122b-a10b","name":"Qwen3.5-122B-A10B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} {"model":"qwen3.5-27b","name":"Qwen3.5-27B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","fine-tuning"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} @@ -216,95 +219,96 @@ {"model":"qwen3.6-flash","name":"Qwen3.6-Flash","family":"qwen3.6-flash","familyName":"Qwen3.6-Flash","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-flash.json"} {"model":"qwen3.6-max-preview","name":"Qwen3.6-Max-Preview","family":"qwen3.6-max","familyName":"Qwen3.6-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":245760,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-max.json"} {"model":"qwen3.6-plus","name":"Qwen3.6-Plus","family":"qwen3.6-plus","familyName":"Qwen3.6-Plus","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-plus.json"} -{"model":"qwen3.6-27b","name":"Qwen3.6-27B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} -{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["VU","TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} -{"model":"qwen3.7-max","name":"Qwen3.7-Max","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} -{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} +{"model":"qwen3.6-27b","name":"Qwen3.6-27B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"3"},{"type":"output_token","unit":"每百万tokens","price":"18"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} +{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["VU","TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.8"},{"type":"output_token","unit":"每百万tokens","price":"10.8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} +{"model":"qwen3.7-flash","name":"Qwen3.7-Flash","family":"qwen3.7-flash","familyName":"Qwen3.7-Flash","provider":"qwen","capabilities":["TG","VU","Reasoning"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-flash.json"} +{"model":"qwen3.7-max","name":"Qwen3.7-Max","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"36"},{"type":"input_token_cache","unit":"每百万tokens","price":"2.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"6"},{"type":"output_token_batch","unit":"每百万tokens","price":"18"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"15"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"1.2"},{"type":"input_token_batch_chat","unit":"每百万tokens","price":"12"},{"type":"output_token_batch_chat","unit":"每百万tokens","price":"36"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} +{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"36"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} {"model":"qwen3.7-plus","name":"Qwen3.7-Plus","family":"qwen3.7-plus","familyName":"Qwen3.7-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search","structured-outputs"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-plus.json"} -{"model":"qwen3-14b","name":"Qwen3-14B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":81920,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-32b","name":"Qwen3-32B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-8b","name":"Qwen3-8B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-14b","name":"Qwen3-14B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"thinking_input_token","unit":"每百万tokens","price":"1"},{"type":"thinking_output_token","unit":"每百万tokens","price":"10"},{"type":"ft","unit":"每百万tokens","price":"30"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":81920,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-32b","name":"Qwen3-32B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"},{"type":"ft","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-8b","name":"Qwen3-8B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.5"},{"type":"thinking_output_token","unit":"每百万tokens","price":"5"},{"type":"ft","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} {"model":"qwen3-coder-next","name":"通义千问3-Coder-Next","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":[],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwq-plus","name":"QwQ-Plus","family":"qwq-plus","familyName":"Qwen-QwQ-Plus","provider":"qwen","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwq-plus.json"} -{"model":"sambert-beth-v1","name":"Sambert语音合成-Beth","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-betty-v1","name":"Sambert语音合成-Betty","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-brian-v1","name":"Sambert语音合成-Brian","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-cally-v1","name":"Sambert语音合成-Cally","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-camila-v1","name":"Sambert语音合成-Camila","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-cindy-v1","name":"Sambert语音合成-Cindy","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-clara-v1","name":"Sambert语音合成-Clara","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-donna-v1","name":"Sambert语音合成-Donna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-eva-v1","name":"Sambert语音合成-Eva","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-hanna-v1","name":"Sambert语音合成-Hanna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-indah-v1","name":"Sambert语音合成-Indah","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-perla-v1","name":"Sambert语音合成-Perla","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-waan-v1","name":"Sambert语音合成-Waan","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhichu-v1","name":"Sambert语音合成-知厨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhida-v1","name":"Sambert语音合成-知达","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhide-v1","name":"Sambert语音合成-知德","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhifei-v1","name":"Sambert语音合成-知飞","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhigui-v1","name":"Sambert语音合成-知柜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhihao-v1","name":"Sambert语音合成-知浩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhijia-v1","name":"Sambert语音合成-知佳","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhijing-v1","name":"Sambert语音合成-知婧","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhilun-v1","name":"Sambert语音合成-知伦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimao-v1","name":"Sambert语音合成-知猫","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimiao-emo-v1","name":"Sambert语音合成-知妙（多情感）","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiming-v1","name":"Sambert语音合成-知茗","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimo-v1","name":"Sambert语音合成-知墨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhina-v1","name":"Sambert语音合成-知娜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhinan-v1","name":"Sambert语音合成-知楠","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiqi-v1","name":"Sambert语音合成-知琪","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiqian-v1","name":"Sambert语音合成-知倩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiru-v1","name":"Sambert语音合成-知茹","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhishu-v1","name":"Sambert语音合成-知树","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhishuo-v1","name":"Sambert语音合成-知硕","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhistella-v1","name":"Sambert语音合成-知莎","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiting-v1","name":"Sambert语音合成-知婷","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiwei-v1","name":"Sambert语音合成-知薇","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhixiang-v1","name":"Sambert语音合成-知祥","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhixiao-v1","name":"Sambert语音合成-知笑","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiya-v1","name":"Sambert语音合成-知雅","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiye-v1","name":"Sambert语音合成-知晔","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiying-v1","name":"Sambert语音合成-知颖","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiyuan-v1","name":"Sambert语音合成-知媛","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiyue-v1","name":"Sambert语音合成-知悦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"4"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":[],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"1"},{"type":"thinking_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"ft","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.5"},{"type":"thinking_output_token","unit":"每百万tokens","price":"5"},{"type":"ft","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwq-plus","name":"QwQ-Plus","family":"qwq-plus","familyName":"Qwen-QwQ-Plus","provider":"qwen","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.6"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.8"},{"type":"output_token_batch","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwq-plus.json"} +{"model":"sambert-beth-v1","name":"Sambert语音合成-Beth","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-betty-v1","name":"Sambert语音合成-Betty","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-brian-v1","name":"Sambert语音合成-Brian","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-cally-v1","name":"Sambert语音合成-Cally","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-camila-v1","name":"Sambert语音合成-Camila","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-cindy-v1","name":"Sambert语音合成-Cindy","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-clara-v1","name":"Sambert语音合成-Clara","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-donna-v1","name":"Sambert语音合成-Donna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-eva-v1","name":"Sambert语音合成-Eva","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-hanna-v1","name":"Sambert语音合成-Hanna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-indah-v1","name":"Sambert语音合成-Indah","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-perla-v1","name":"Sambert语音合成-Perla","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-waan-v1","name":"Sambert语音合成-Waan","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhichu-v1","name":"Sambert语音合成-知厨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhida-v1","name":"Sambert语音合成-知达","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhide-v1","name":"Sambert语音合成-知德","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhifei-v1","name":"Sambert语音合成-知飞","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhigui-v1","name":"Sambert语音合成-知柜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhihao-v1","name":"Sambert语音合成-知浩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhijia-v1","name":"Sambert语音合成-知佳","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhijing-v1","name":"Sambert语音合成-知婧","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhilun-v1","name":"Sambert语音合成-知伦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimao-v1","name":"Sambert语音合成-知猫","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimiao-emo-v1","name":"Sambert语音合成-知妙（多情感）","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiming-v1","name":"Sambert语音合成-知茗","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimo-v1","name":"Sambert语音合成-知墨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhina-v1","name":"Sambert语音合成-知娜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhinan-v1","name":"Sambert语音合成-知楠","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiqi-v1","name":"Sambert语音合成-知琪","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiqian-v1","name":"Sambert语音合成-知倩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiru-v1","name":"Sambert语音合成-知茹","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhishu-v1","name":"Sambert语音合成-知树","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhishuo-v1","name":"Sambert语音合成-知硕","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhistella-v1","name":"Sambert语音合成-知莎","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiting-v1","name":"Sambert语音合成-知婷","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiwei-v1","name":"Sambert语音合成-知薇","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhixiang-v1","name":"Sambert语音合成-知祥","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhixiao-v1","name":"Sambert语音合成-知笑","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiya-v1","name":"Sambert语音合成-知雅","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiye-v1","name":"Sambert语音合成-知晔","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiying-v1","name":"Sambert语音合成-知颖","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiyuan-v1","name":"Sambert语音合成-知媛","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiyue-v1","name":"Sambert语音合成-知悦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} {"model":"shoemodel-v1","name":"鞋靴模特","family":"shoemodel-v1","familyName":"鞋靴模特","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2804662.html","detailPath":"groups/shoemodel-v1.json"} -{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":163840,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":163840,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} {"model":"speech-biasing","name":"语音识别热词","family":"speech-biasing","familyName":"语音识别热词","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/speech-biasing.json"} -{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","family":"stepfun-models-market-place","familyName":"StepFun推理模型","provider":"stepfun","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","batch"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3036697.html","detailPath":"groups/stepfun-models-market-place.json"} -{"model":"tongyi-intent-detect-v3","name":"意图分类模型","family":"tongyi-intent-detect-v3","familyName":"意图分类模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861138.html","detailPath":"groups/tongyi-intent-detect-v3.json"} -{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","family":"tongyi-xiaomi-analysis-flash","familyName":"通义晓蜜-对话分析-flash","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-flash.json"} -{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","family":"tongyi-xiaomi-analysis-pro","familyName":"通义晓蜜-对话分析-pro","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-pro.json"} -{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","batch","structured-outputs"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} -{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","structured-outputs","batch"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} +{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","family":"stepfun-models-market-place","familyName":"StepFun推理模型","provider":"stepfun","capabilities":["TG","VU"],"features":["function-calling","structured-outputs","cache","batch"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.35"},{"type":"output_token","unit":"每百万tokens","price":"8.1"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.27"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3036697.html","detailPath":"groups/stepfun-models-market-place.json"} +{"model":"tongyi-intent-detect-v3","name":"意图分类模型","family":"tongyi-intent-detect-v3","familyName":"意图分类模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.4"},{"type":"output_token","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861138.html","detailPath":"groups/tongyi-intent-detect-v3.json"} +{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","family":"tongyi-xiaomi-analysis-flash","familyName":"通义晓蜜-对话分析-flash","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.2"},{"type":"output_token","unit":"每百万tokens","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-flash.json"} +{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","family":"tongyi-xiaomi-analysis-pro","familyName":"通义晓蜜-对话分析-pro","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"2.7"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-pro.json"} +{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","batch","structured-outputs"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"prices":[{"type":"generation_3d_text_to_3d_standard_no_texture","unit":"每次","price":"0.7"},{"type":"generation_3d_image_to_3d_standard_no_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_multiview_to_3d_standard_no_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_text_to_3d_standard_sd_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_image_to_3d_standard_sd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_multiview_to_3d_standard_sd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_text_to_3d_standard_hd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_standard_hd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_standard_hd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_ultra_no_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_ultra_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_ultra_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_ultra_sd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_image_to_3d_ultra_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_multiview_to_3d_ultra_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_text_to_3d_ultra_hd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_image_to_3d_ultra_hd_texture","unit":"每次","price":"4.2"},{"type":"generation_3d_multiview_to_3d_ultra_hd_texture","unit":"每次","price":"4.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} +{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","structured-outputs","batch"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"prices":[{"type":"generation_3d_text_to_3d_no_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_sd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_image_to_3d_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_multiview_to_3d_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_text_to_3d_hd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_image_to_3d_hd_texture","unit":"每次","price":"4.2"},{"type":"generation_3d_multiview_to_3d_hd_texture","unit":"每次","price":"4.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} {"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["VU","TG"],"features":["structured-outputs"],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.216"},{"type":"output_token","unit":"每百万tokens","price":"0.216"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["structured-outputs","prefix-completion","function-calling","cache"],"contextWindow":131072,"maxInputTokens":131072,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":131072,"maxInputTokens":32768,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":60,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":60,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"24"},{"type":"input_token_cache","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} -{"model":"video-style-transform","name":"视频风格重绘","family":"video-style-transform","familyName":"视频风格重绘","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846319.html","detailPath":"groups/video-style-transform.json"} -{"model":"videoretalk","name":"声动人像VideoRetalk","family":"videoretalk","familyName":"声动人像VideoRetalk","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860466.html","detailPath":"groups/videoretalk.json"} +{"model":"video-style-transform","name":"视频风格重绘","family":"video-style-transform","familyName":"视频风格重绘","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.2"},{"type":"video_ratio_720p","unit":"每秒","price":"0.5"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846319.html","detailPath":"groups/video-style-transform.json"} +{"model":"videoretalk","name":"声动人像VideoRetalk","family":"videoretalk","familyName":"声动人像VideoRetalk","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Audio"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860466.html","detailPath":"groups/videoretalk.json"} {"model":"vidu/vidu-image_reference2image","name":"Vidu-image_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.625"},{"type":"image_type_2k","unit":"每张","price":"1"},{"type":"image_type_4k","unit":"每张","price":"1.46875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} {"model":"vidu/viduq2-fast_reference2image","name":"ViduQ2-fast_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.28125"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} {"model":"vidu/viduq2-pro_reference2image","name":"ViduQ2-Pro_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.9375"},{"type":"image_type_2k","unit":"每张","price":"0.9375"},{"type":"image_type_4k","unit":"每张","price":"1.71875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} @@ -331,55 +335,55 @@ {"model":"vidu/viduq3-turbo_text2video","name":"ViduQ3-Turbo_text2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.25"},{"type":"video_ratio_720p","unit":"每秒","price":"0.375"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.4375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} {"model":"virtualmodel-v2","name":"虚拟模特V2","family":"virtualmodel-v2","familyName":"虚拟模特V2","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796985.html","detailPath":"groups/virtualmodel-v2.json"} {"model":"voice-enrollment","name":"大模型声音复刻及声音设计","family":"voice-enrollment","familyName":"大模型声音复刻及声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/voice-enrollment.json"} -{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982258.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.6-image","name":"Wan2.6-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image","Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001143.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.7-image","name":"Wan2.7-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2868981.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.2-animate-mix","name":"Wan2.2-Animate-Mix","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982219.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-animate-move","name":"Wan2.2-Animate-Move","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2981852.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-i2v-flash","name":"Wan2.2-I2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-i2v-plus","name":"Wan2.2-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-kf2v-flash","name":"Wan2.2-KF2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-s2v","name":"通义万相2.2-数字人-S2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978215.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-s2v-detect","name":"Wan2.2-S2V-Detect","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978214.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.5-i2v-preview","name":"Wan2.5-I2V-Preview","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience","fine-tuning"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Image","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.6-i2v","name":"Wan2.6-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Text","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.6-i2v-flash","name":"wan2.6-I2V-flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Audio","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.7-i2v","name":"Wan2.7-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025059.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-i2v-plus","name":"Wan2.1-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-i2v-turbo","name":"Wan2.1-I2V-Turbo","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-kf2v-plus","name":"Wan2.1-KF2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982258.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.6-image","name":"Wan2.6-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image","Text"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001143.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.7-image","name":"Wan2.7-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2868981.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.2-animate-mix","name":"Wan2.2-Animate-Mix","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.6"},{"type":"video_ratio_pro","unit":"每秒","price":"0.9"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982219.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-animate-move","name":"Wan2.2-Animate-Move","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.4"},{"type":"video_ratio_pro","unit":"每秒","price":"0.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2981852.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-i2v-flash","name":"Wan2.2-I2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.1"},{"type":"video_ratio_720p","unit":"每秒","price":"0.2"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.48"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-i2v-plus","name":"Wan2.2-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.14"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-kf2v-flash","name":"Wan2.2-KF2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.1"},{"type":"video_ratio_720p","unit":"每秒","price":"0.2"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.48"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-s2v","name":"通义万相2.2-数字人-S2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.5"},{"type":"video_ratio_720p","unit":"每秒","price":"0.9"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978215.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-s2v-detect","name":"Wan2.2-S2V-Detect","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978214.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.5-i2v-preview","name":"Wan2.5-I2V-Preview","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience","fine-tuning"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Image","Audio"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.6-i2v","name":"Wan2.6-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Text","Audio"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"},{"type":"720P_batch","unit":"每秒","price":"0.6"},{"type":"1080P_batch","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.6-i2v-flash","name":"wan2.6-I2V-flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Audio","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.3"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.5"},{"type":"720P_no_audio","unit":"每秒","price":"0.15"},{"type":"1080P_no_audio","unit":"每秒","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.7-i2v","name":"Wan2.7-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025059.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-i2v-plus","name":"Wan2.1-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-i2v-turbo","name":"Wan2.1-I2V-Turbo","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.24"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-kf2v-plus","name":"Wan2.1-KF2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} {"model":"wan2.6-r2v","name":"Wan2.6-R2V","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Video","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} {"model":"wan2.6-r2v-flash","name":"Wan2.6-R2V-Flash","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Video","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.3"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.5"},{"type":"720P_no_audio","unit":"每秒","price":"0.15"},{"type":"1080P_no_audio","unit":"每秒","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} {"model":"wan2.7-r2v","name":"Wan2.7-R2V","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} -{"model":"wan2.2-t2i-flash","name":"Wan2.2-T2I-Flash","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.2-t2i-plus","name":"Wan2.2-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.5-t2i-preview","name":"Wan2.5-T2I-Preview","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.6-t2i","name":"Wan2.6-T2I","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx-v1","name":"wanx-t2i","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.2-t2i-flash","name":"Wan2.2-T2I-Flash","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.2-t2i-plus","name":"Wan2.2-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.5-t2i-preview","name":"Wan2.5-T2I-Preview","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.6-t2i","name":"Wan2.6-T2I","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx-v1","name":"wanx-t2i","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.16"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} {"model":"wanx-v1-0521","name":"万相-文本生成图像-2024-05-21","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.0-t2i-turbo","name":"Wan2.0-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.1-t2i-plus","name":"Wan2.1-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.1-t2i-turbo","name":"Wan2.1-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.0-t2i-turbo","name":"Wan2.0-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.04"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.1-t2i-plus","name":"Wan2.1-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.1-t2i-turbo","name":"Wan2.1-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} {"model":"wan2.2-t2v-plus","name":"Wan2.2-T2V-Plus","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.14"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wan2.5-t2v-preview","name":"Wan2.5-T2V-Preview","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Audio"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wan2.6-t2v","name":"Wan2.6-T2V","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Audio"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"},{"type":"720P_batch","unit":"每秒","price":"0.6"},{"type":"1080P_batch","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wan2.7-t2v","name":"Wan2.7-T2V","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wanx2.1-t2v-plus","name":"Wan2.1-T2V-Plus","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wanx2.1-t2v-turbo","name":"Wan2.1-T2V-Turbo","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.24"},{"type":"video_ratio_720p","unit":"每秒","price":"0.24"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} -{"model":"wan2.7-videoedit","name":"Wan2.7-VideoEdit","family":"wan-video-edit","familyName":"Wan-VideoEdit","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021842.html","detailPath":"groups/wan-video-edit.json"} -{"model":"wanx-background-generation-v2","name":"图像背景生成","family":"wanx-background-generation-v2","familyName":"图像背景生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712497.html","detailPath":"groups/wanx-background-generation-v2.json"} +{"model":"wan2.7-videoedit","name":"Wan2.7-VideoEdit","family":"wan-video-edit","familyName":"Wan-VideoEdit","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021842.html","detailPath":"groups/wan-video-edit.json"} +{"model":"wanx-background-generation-v2","name":"图像背景生成","family":"wanx-background-generation-v2","familyName":"图像背景生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712497.html","detailPath":"groups/wanx-background-generation-v2.json"} {"model":"wanx-poster-generation-v1","name":"创意海报生成","family":"wanx-poster-generation-v1","familyName":"创意海报生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807172.html","detailPath":"groups/wanx-poster-generation-v1.json"} -{"model":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","family":"wanx-sketch-to-image-lite","familyName":"万相-涂鸦作画","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712498.html","detailPath":"groups/wanx-sketch-to-image-lite.json"} -{"model":"wanx-style-repaint-v1","name":"人像风格重绘","family":"wanx-style-repaint-v1","familyName":"人像风格重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712493.html","detailPath":"groups/wanx-style-repaint-v1.json"} +{"model":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","family":"wanx-sketch-to-image-lite","familyName":"万相-涂鸦作画","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.06"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712498.html","detailPath":"groups/wanx-sketch-to-image-lite.json"} +{"model":"wanx-style-repaint-v1","name":"人像风格重绘","family":"wanx-style-repaint-v1","familyName":"人像风格重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.12"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712493.html","detailPath":"groups/wanx-style-repaint-v1.json"} {"model":"wanx-virtualmodel","name":"虚拟模特","family":"wanx-virtualmodel","familyName":"虚拟模特","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796985.html","detailPath":"groups/wanx-virtualmodel.json"} {"model":"wanx-x-painting","name":"万相-图像局部重绘","family":"wanx-x-painting","familyName":"万相-图像局部重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2797051.html","detailPath":"groups/wanx-x-painting.json"} -{"model":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","family":"wanx2.1-vace-plus","familyName":"Wan2.1-VACE-Plus","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2922183.html","detailPath":"groups/wanx2.1-vace-plus.json"} -{"model":"wordart-semantic","name":"WordArt锦书-文字变形","family":"wordart-semantic","familyName":"WordArt锦书-文字变形","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712513.html","detailPath":"groups/wordart-semantic.json"} -{"model":"wordart-texture","name":"WordArt锦书-文字纹理生成","family":"wordart-texture","familyName":"WordArt锦书-文字纹理生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712510.html","detailPath":"groups/wordart-texture.json"} -{"model":"xiaomi/mimo-v2.5-pro","name":"xiaomi/mimo-v2.5-pro","family":"xiaomi-models-market-place","familyName":"MiMo文本模型","provider":"xiaomi","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3033942.html","detailPath":"groups/xiaomi-models-market-place.json"} -{"model":"z-image-turbo","name":"Z-Image-Turbo","family":"z-image-turbo","familyName":"Z-Image-Turbo","provider":"qwen-domain-model","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text","Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3002354.html","detailPath":"groups/z-image-turbo.json"} -{"model":"ZHIPU/GLM-5","name":"ZHIPU/GLM-5","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","cache","structured-outputs","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026315.html","detailPath":"groups/zhipu-models-market-place.json"} -{"model":"ZHIPU/GLM-5.1","name":"ZHIPU/GLM-5.1","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/glm-zhipu","detailPath":"groups/zhipu-models-market-place.json"} -{"model":"ZHIPU/GLM-5.2","name":"ZHIPU/GLM-5.2","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315","detailPath":"groups/zhipu-models-market-place.json"} +{"model":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","family":"wanx2.1-vace-plus","familyName":"Wan2.1-VACE-Plus","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2922183.html","detailPath":"groups/wanx2.1-vace-plus.json"} +{"model":"wordart-semantic","name":"WordArt锦书-文字变形","family":"wordart-semantic","familyName":"WordArt锦书-文字变形","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.24"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712513.html","detailPath":"groups/wordart-semantic.json"} +{"model":"wordart-texture","name":"WordArt锦书-文字纹理生成","family":"wordart-texture","familyName":"WordArt锦书-文字纹理生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712510.html","detailPath":"groups/wordart-texture.json"} +{"model":"xiaomi/mimo-v2.5-pro","name":"xiaomi/mimo-v2.5-pro","family":"xiaomi-models-market-place","familyName":"MiMo文本模型","provider":"xiaomi","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"7"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.4"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3033942.html","detailPath":"groups/xiaomi-models-market-place.json"} +{"model":"z-image-turbo","name":"Z-Image-Turbo","family":"z-image-turbo","familyName":"Z-Image-Turbo","provider":"qwen-domain-model","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text","Image"],"request_modality":["Text"]},"prices":[{"type":"image_standard","unit":"每张","price":"0.1"},{"type":"image_thinking","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3002354.html","detailPath":"groups/z-image-turbo.json"} +{"model":"ZHIPU/GLM-5","name":"ZHIPU/GLM-5","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","cache","structured-outputs","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6"},{"type":"output_token","unit":"每百万tokens","price":"22"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.5"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026315.html","detailPath":"groups/zhipu-models-market-place.json"} +{"model":"ZHIPU/GLM-5.1","name":"ZHIPU/GLM-5.1","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/glm-zhipu","detailPath":"groups/zhipu-models-market-place.json"} +{"model":"ZHIPU/GLM-5.2","name":"ZHIPU/GLM-5.2","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315","detailPath":"groups/zhipu-models-market-place.json"} diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-addchunk.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-addchunk.md new file mode 100644 index 00000000..807c00f2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-addchunk.md @@ -0,0 +1,233 @@ +# AddChunk - 新增切片 + +使用此API可为文档搜索类（document）、数据查询类（table）、图片问答类（image）知识库添加切片。 + +## 接口说明 + +- 对于文档搜索类（document）、数据查询类（table）、图片问答类（image）知识库，本接口可向指定知识库中添加切片内容；目前尚不支持对音视频搜索类（multimedia）知识库进行相关操作。仅当数据源为表格连接器(excel)时，对数据查询与图片问答类型知识库的操作方可生效。 + +- RAM 用户（子账号）需要首先获取阿里云百炼的 [API 权限](https://help.aliyun.com/zh/model-studio/grant-data-access-permission-to-ram-user)（需要`AliyunBailianDataFullAccess`，已包括 sfm:ChunkList 权限点），并[加入一个业务空间](https://help.aliyun.com/zh/model-studio/grant-the-business-space-permission-to-ram-users)后，方可调用本接口。阿里云账号（主账号）可直接调用无须授权。建议您通过最新版[阿里云百炼 SDK](https://api.aliyun.com/api-tools/sdk/bailian?version=2023-12-29)[阿里云百炼 SDK](https://api.alibabacloud.com/api-tools/sdk/bailian?version=2023-12-29)来调用本接口。 + +- 调用本接口前，请确保您的知识库已经创建完成且未被删除（即知识库 ID`IndexId`有效）。 + +- 本接口具有幂等性。 + + +**限流说明：** 本接口频繁调用会被限流，频率请勿超过 10 次/秒。如遇限流，请稍后重试。 + +## 调试 + +[您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/bailian/2023-12-29/AddChunk) + + [![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png) 调试](https://api.aliyun.com/api/bailian/2023-12-29/AddChunk) + +## **授权信息** + +当前API暂无授权信息透出。 + +## 请求语法 + +``` +POST /{WorkspaceId}/chunk/create HTTP/1.1 +``` + +## 路径参数 + +**名称** + +**类型** + +**必填** + +**描述** + +**示例值** + +WorkspaceId + +string + +是 + +工作区标识 + +llm-19hxxxxx7htdf9lh + +## 请求参数 + +**名称** + +**类型** + +**必填** + +**描述** + +**示例值** + +PipelineId + +string + +是 + +知识库 id + +79c0alxxxx + +dataId + +string + +否 + +文件 id + +doc\_xxx + +field + +object + +否 + +插入的切片内容信息，以键值对形式传入。文档搜索类知识库使用固定 key 列表： + +- content（**String**）：**必填**，切片正文内容 + +- title（**String**）**选填**，切片标题 + +- image\_urls（**Array**）：**选填**，切片包含的图片链接，最多 10 张 + + +数据查询类、图片问答类知识库 key 不固定，由该知识库的数据源表格决定：key 为 Excel 列标题，value 为对应列的值。 + +{ "content": "The Bailian platform supports parsing multiple document formats including PDF, Word, and PPT.", "title": "Document Parsing and Chunking", "image\_urls": \[ "https://example.com/images/chunk-flow.png", "https://example.com/images/parsing-result.png" \] } + +any + +否 + +插入切片的表头字段信息，仅数据查询类与图片问答类知识库支持。需要参与检索或参与回复的表头为必填。各类型取值要求： + +- **String 类型** ：最大长度 6000 + +- **时间 类型**：13 位时间戳（毫秒） + +- **Long 类型**：整数，最大 2147483647 + +- **Double 类型**：支持小数 + +- **image\_url 类型**：最多 5 张，多张用英文逗号拼接为一个字符串 + + +{"Product Name": "Wireless Bluetooth Headphones", "Publish Time": 1752624000000, "Stock Quantity": 1580, "Unit Price": 299.99, "image\_url":"https://example.com/images/headphones-front.jpg,https://example.com/images/headphones-side.jpg,https://example.com/images/headphones-package.jpg" } + +## **返回参数** + +**名称** + +**类型** + +**描述** + +**示例值** + +object + +Schema of Response + +RequestId + +string + +请求 id + +35A267BF-xxxx-54DB-8394-AA3B0742D833 + +Code + +string + +错误状态码 + +Index.InvalidParameter + +Message + +string + +错误信息 + +Required parameter(%s) missing or invalid, please check the request parameters. + +Success + +boolean + +接口调用是否成功 + +**枚举值：** + +- true : + + true + +- false : + + false + + +true + +Data + +boolean + +请求成功返回的业务数据 + +**枚举值：** + +- true : + + true + +- false : + + false + + +true + +Status + +string + +接口返回的状态码 + +200 + +## 示例 + +正常返回示例 + +`JSON`格式 + +``` +{ + "RequestId": "35A267BF-xxxx-54DB-8394-AA3B0742D833", + "Code": "Index.InvalidParameter", + "Message": "Required parameter(%s) missing or invalid, please check the request parameters.", + "Success": true, + "Data": true, + "Status": "200" +} +``` + +## 错误码 + +访问[错误中心](https://api.aliyun.com/document/bailian/2023-12-29/errorCode)查看更多错误码。 + +## **变更历史** + +更多信息，参考[变更详情](https://api.aliyun.com/document/bailian/2023-12-29/AddChunk#workbench-doc-change-demo)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md index 3b1ff1ad..04d60432 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md @@ -163,6 +163,8 @@ string 默认值为空，此时使用 text-embedding-v2 模型。 +text-embedding-v4 + RerankModelName string @@ -322,7 +324,7 @@ array 否 -创建知识库时可同步导入文件。此处通过指定类目 ID，可导入对应类目下的所有文件（建议导入不超过 10000 个。如有剩余文件，后续可调用 **SubmitIndexAddDocumentsJob** 接口继续导入）。 +创建知识库时可同步导入文件。此处通过指定类目 ID，可导入对应类目下的所有文件（建议导入不超过 500 个。如有剩余文件，后续可调用 **SubmitIndexAddDocumentsJob** 接口继续导入）。 string diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md index 05ded095..5bfce045 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md @@ -283,14 +283,6 @@ Excel 文件表头是否支持拼装。 true -OverlapSize - -string - -分段重叠长度。 - -100 - Message string @@ -315,14 +307,6 @@ string cate\_21a407a3372c4ba7aedc649709143f0cxxxxxxxx -GmtModified - -integer - -文件导入知识库的时间，采用 Unix timestamp 格式。 - -1744856423000 - DocumentType string @@ -331,13 +315,13 @@ string pdf -ChunkMode +Id string -自定义切分。 +文件 ID。 -DashSplitter +doc\_c134aa2073204a5d936d870bf960f56axxxxxxxx Code @@ -347,14 +331,6 @@ string 110002 -separator - -string - -分句标识符。 - -" " - Name string @@ -371,13 +347,41 @@ string 600 -Id +OverlapSize string -文件 ID。 +分段重叠长度。 -doc\_c134aa2073204a5d936d870bf960f56axxxxxxxx +100 + +ChunkMode + +string + +自定义切分。 + +DashSplitter + +GmtModified + +integer + +文件导入知识库的时间，采用 Unix timestamp 格式。 + +1744856423000 + +separator + +string + +分句标识符。 + +" " + +MetaExtractInfo + +string IndexId @@ -436,18 +440,19 @@ string { "Status": "RUNNING", "EnableHeaders": "true", - "OverlapSize": "100", "Message": "check fileUrlKey[file_path] / fileNameKey[null] / fileExtensionKey[file_extension] is invalid", "Size": 996764, "SourceId": "cate_21a407a3372c4ba7aedc649709143f0cxxxxxxxx\n", - "GmtModified": 1744856423000, "DocumentType": "pdf", - "ChunkMode": "DashSplitter", + "Id": "doc_c134aa2073204a5d936d870bf960f56axxxxxxxx\n", "Code": "110002", - "separator": "\" \"", "Name": "翻译平台运维文档\n", "ChunkSize": "600", - "Id": "doc_c134aa2073204a5d936d870bf960f56axxxxxxxx\n" + "OverlapSize": "100", + "ChunkMode": "DashSplitter", + "GmtModified": 1744856423000, + "separator": "\" \"", + "MetaExtractInfo": "" } ], "IndexId": "79c0alxxxx", diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md index 70c29639..a7e7d52f 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md @@ -228,6 +228,12 @@ API概述 查看文本切片列表及信息。 +[AddChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addchunk) + +新增切片 + +使用此API可为文档搜索类（document）、数据查询类（table）、图片问答类（image）知识库添加切片。 + [UpdateChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatechunk) 修改切片 diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md index fa10b5cc..ef5ee20d 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md @@ -78,11 +78,11 @@ **关联操作** -sfm:ChangeParseSetting +sfm:ListCategory -[ChangeParseSetting](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-changeparsesetting) +[ListCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listcategory) -update +list \*全部资源 @@ -92,11 +92,11 @@ update 无 -sfm:UpdateFileTag +sfm:GetIndexJobStatus -[UpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatefiletag) +[GetIndexJobStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexjobstatus) -update +get \*全部资源 @@ -106,11 +106,11 @@ update 无 -sfm:DeleteCategory +sfm:AddCategory -[DeleteCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletecategory) +[AddCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addcategory) -delete +create \*全部资源 @@ -120,11 +120,11 @@ delete 无 -sfm:UpdatePromptTemplate +sfm:GetAlipayUrl -[UpdatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateprompttemplate) +[GetAlipayUrl](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipayurl) -update +none \*全部资源 @@ -134,11 +134,11 @@ update 无 -sfm:SubmitIndexJob +sfm:DeleteMemoryNode -[SubmitIndexJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexjob) +[DeleteMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememorynode) -create +delete \*全部资源 @@ -162,11 +162,11 @@ get 无 -sfm:DeleteChunk +sfm:SubmitIndexJob -[DeleteChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletechunk) +[SubmitIndexJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexjob) -delete +create \*全部资源 @@ -176,11 +176,11 @@ delete 无 -sfm:GetAlipayTransferStatus +sfm:DeleteCategory -[GetAlipayTransferStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipaytransferstatus) +[DeleteCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletecategory) -none +delete \*全部资源 @@ -190,11 +190,11 @@ none 无 -sfm:DeletePromptTemplate +sfm:ListIndexFileDetails -[DeletePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteprompttemplate) +[ListIndexFileDetails](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexfiledetails) -delete +list \*全部资源 @@ -204,11 +204,11 @@ delete 无 -sfm:ListIndex +sfm:UpdateConnector -[ListIndices](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindices) +[UpdateConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateconnector) -list +update \*全部资源 @@ -218,11 +218,11 @@ list 无 -sfm:CreateMemoryNode +sfm:ChangeParseSetting -[CreateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-creatememorynode) +[ChangeParseSetting](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-changeparsesetting) -create +update \*全部资源 @@ -232,11 +232,11 @@ create 无 -sfm:UpdateMemory +sfm:GetMemory -[UpdateMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememory) +[GetMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemory) -update +get \*全部资源 @@ -246,11 +246,11 @@ update 无 -sfm:ChunkList +sfm:GetAvailableParserTypes -[ListChunks](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listchunks) +[GetAvailableParserTypes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getavailableparsertypes) -list +get \*全部资源 @@ -260,11 +260,11 @@ list 无 -sfm:ListIndexFileDetails +sfm:UpdateMemory -[ListIndexFileDetails](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexfiledetails) +[UpdateMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememory) -list +update \*全部资源 @@ -274,9 +274,9 @@ list 无 -sfm:SubmitIndexAddDocumentsJob +sfm:CreateMemory -[SubmitIndexAddDocumentsJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexadddocumentsjob) +[CreateMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-creatememory) create @@ -288,11 +288,11 @@ create 无 -sfm:UpdateIndex +sfm:ApplyFileUploadLease -[UpdateIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateindex) +[ApplyFileUploadLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applyfileuploadlease) -update +none \*全部资源 @@ -316,25 +316,11 @@ list 无 -sfm:Retrieve - -[Retrieve](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-retrieve) - -none - -\*全部资源 - -`*****` - -无 - -无 - -sfm:AddConnector +sfm:DeleteConnector -[AddConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addconnector) +DeleteConnector -create +delete \*全部资源 @@ -344,11 +330,11 @@ create 无 -sfm:ListMemories +sfm:BatchUpdateFileTag -[ListMemories](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemories) +[BatchUpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-batchupdatefiletag) -list +update \*全部资源 @@ -372,11 +358,11 @@ create 无 -sfm:UpdateConnector +sfm:DeleteChunk -[UpdateConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateconnector) +[DeleteChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletechunk) -update +delete \*全部资源 @@ -386,9 +372,9 @@ update 无 -sfm:DeleteFile +sfm:DeleteMemory -[DeleteFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefile) +[DeleteMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememory) delete @@ -400,11 +386,11 @@ delete 无 -sfm:AddCategory +sfm:DeletePromptTemplate -[AddCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addcategory) +[DeletePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteprompttemplate) -create +delete \*全部资源 @@ -414,11 +400,11 @@ create 无 -sfm:BatchUpdateFileTag +sfm:ListMemories -[BatchUpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-batchupdatefiletag) +[ListMemories](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemories) -update +list \*全部资源 @@ -428,11 +414,11 @@ update 无 -sfm:ListIndexFiles +sfm:AddConnector -[ListIndexDocuments](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexdocuments) +[AddConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addconnector) -list +create \*全部资源 @@ -442,11 +428,11 @@ list 无 -sfm:ListMemoryNodes +sfm:GetConnector -[ListMemoryNodes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemorynodes) +[GetConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getconnector) -list +get \*全部资源 @@ -456,11 +442,11 @@ list 无 -sfm:DescribeFile +sfm:UpdatePromptTemplate -[DescribeFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-describefile) +[UpdatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateprompttemplate) -none +update \*全部资源 @@ -470,11 +456,11 @@ none 无 -sfm:CreateIndex +sfm:DeleteFiles -[CreateIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-createindex) +[DeleteFiles](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefiles) -create +delete \*全部资源 @@ -484,11 +470,11 @@ create 无 -sfm:ListCategory +sfm:GetMemoryNode -[ListCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listcategory) +[GetMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemorynode) -list +get \*全部资源 @@ -498,11 +484,11 @@ list 无 -sfm:AddTable +sfm:ApplyTempStorageLease -[AddTable](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addtable) +[ApplyTempStorageLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applytempstoragelease) -create +none \*全部资源 @@ -512,11 +498,11 @@ create 无 -sfm:UpdateMemoryNode +sfm:DeleteFile -[UpdateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememorynode) +[DeleteFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefile) -update +delete \*全部资源 @@ -526,11 +512,11 @@ update 无 -sfm:GetAvailableParserTypes +sfm:AddFilesFromAuthorizedOss -[GetAvailableParserTypes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getavailableparsertypes) +[AddFilesFromAuthorizedOss](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addfilesfromauthorizedoss) -get +create \*全部资源 @@ -540,11 +526,11 @@ get 无 -sfm:DeleteMemory +sfm:CreateMemoryNode -[DeleteMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememory) +[CreateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-creatememorynode) -delete +create \*全部资源 @@ -554,11 +540,11 @@ delete 无 -sfm:DeleteConnector +sfm:UpdateChunk -DeleteConnector +[UpdateChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatechunk) -delete +update \*全部资源 @@ -568,11 +554,11 @@ delete 无 -sfm:UpdateTableFromAuthorizedOss +sfm:ListCategory -[UpdateTableFromAuthorizedOss](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatetablefromauthorizedoss) +[ListCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listcategory) -update +list \*全部资源 @@ -582,11 +568,11 @@ update 无 -sfm:ApplyTempStorageLease +sfm:GetIndexJobStatus -[ApplyTempStorageLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applytempstoragelease) +[GetIndexJobStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexjobstatus) -none +get \*全部资源 @@ -596,9 +582,9 @@ none 无 -sfm:GetConnector +sfm:GetParseSettings -[GetConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getconnector) +[GetParseSettings](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getparsesettings) get @@ -610,11 +596,11 @@ get 无 -sfm:DeleteMemoryNode +sfm:GetAlipayUrl -[DeleteMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememorynode) +[GetAlipayUrl](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipayurl) -delete +none \*全部资源 @@ -624,11 +610,11 @@ delete 无 -sfm:GetMemory +sfm:AddCategory -[GetMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemory) +[AddCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addcategory) -get +create \*全部资源 @@ -638,11 +624,11 @@ get 无 -sfm:DeleteIndex +sfm:SubmitIndexJob -[DeleteIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteindex) +[SubmitIndexJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexjob) -none +create \*全部资源 @@ -652,9 +638,9 @@ none 无 -sfm:UpdateChunk +sfm:ChangeParseSetting -[UpdateChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatechunk) +[ChangeParseSetting](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-changeparsesetting) update @@ -666,9 +652,9 @@ update 无 -sfm:DeleteIndexDocument +sfm:DeleteMemoryNode -[DeleteIndexDocument](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteindexdocument) +[DeleteMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememorynode) delete @@ -680,11 +666,11 @@ delete 无 -sfm:GetMemoryNode +sfm:UpdateConnector -[GetMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemorynode) +[UpdateConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateconnector) -get +update \*全部资源 @@ -694,9 +680,9 @@ get 无 -sfm:GetIndexJobStatus +sfm:GetMemory -[GetIndexJobStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexjobstatus) +[GetMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemory) get @@ -708,11 +694,11 @@ get 无 -sfm:ApplyFileUploadLease +sfm:ListIndexFileDetails -[ApplyFileUploadLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applyfileuploadlease) +[ListIndexFileDetails](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexfiledetails) -none +list \*全部资源 @@ -722,11 +708,25 @@ none 无 -sfm:GetPromptTemplate +sfm:DeleteConnector -[GetPromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getprompttemplate) +DeleteConnector -get +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateMemory + +[UpdateMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememory) + +update \*全部资源 @@ -750,9 +750,79 @@ create 无 -sfm:GetAlipayUrl +sfm:ListFile -[GetAlipayUrl](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipayurl) +[ListFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listfile) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeleteChunk + +[DeleteChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletechunk) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:GetAvailableParserTypes + +[GetAvailableParserTypes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getavailableparsertypes) + +get + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeleteCategory + +[DeleteCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletecategory) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ListMemories + +[ListMemories](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemories) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ApplyFileUploadLease + +[ApplyFileUploadLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applyfileuploadlease) none @@ -764,11 +834,11 @@ none 无 -sfm:CreatePromptTemplate +sfm:BatchUpdateFileTag -[CreatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-createprompttemplate) +[BatchUpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-batchupdatefiletag) -create +update \*全部资源 @@ -778,9 +848,9 @@ create 无 -sfm:GetIndexMonitor +sfm:GetConnector -[GetIndexMonitor](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexmonitor) +[GetConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getconnector) get @@ -792,6 +862,48 @@ get 无 +sfm:GetMemoryNode + +[GetMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemorynode) + +get + +\*全部资源 + +`*****` + +无 + +无 + +sfm:AddFile + +[AddFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addfile) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:AddConnector + +[AddConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addconnector) + +create + +\*全部资源 + +`*****` + +无 + +无 + sfm:DeleteFiles [DeleteFiles](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefiles) @@ -806,6 +918,48 @@ delete 无 +sfm:DeleteFile + +[DeleteFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefile) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeletePromptTemplate + +[DeletePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteprompttemplate) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdatePromptTemplate + +[UpdatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateprompttemplate) + +update + +\*全部资源 + +`*****` + +无 + +无 + sfm:AddFilesFromAuthorizedOss [AddFilesFromAuthorizedOss](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addfilesfromauthorizedoss) @@ -820,6 +974,174 @@ create 无 +sfm:CreateMemoryNode + +[CreateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-creatememorynode) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeleteMemory + +[DeleteMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememory) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateChunk + +[UpdateChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatechunk) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ApplyTempStorageLease + +[ApplyTempStorageLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applytempstoragelease) + +none + +\*全部资源 + +`*****` + +无 + +无 + +sfm:GetIndexMonitor + +[GetIndexMonitor](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexmonitor) + +get + +\*全部资源 + +`*****` + +无 + +无 + +sfm:Retrieve + +[Retrieve](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-retrieve) + +none + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeleteIndex + +[DeleteIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteindex) + +none + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ListMemoryNodes + +[ListMemoryNodes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemorynodes) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ListIndex + +[ListIndices](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindices) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateMemoryNode + +[UpdateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememorynode) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:GetPromptTemplate + +[GetPromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getprompttemplate) + +get + +\*全部资源 + +`*****` + +无 + +无 + +sfm:CreateIndex + +[CreateIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-createindex) + +create + +\*全部资源 + +`*****` + +无 + +无 + sfm:ListPromptTemplates [ListPromptTemplates](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listprompttemplates) @@ -834,6 +1156,160 @@ list 无 +sfm:DeleteIndexDocument + +[DeleteIndexDocument](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteindexdocument) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:SubmitIndexAddDocumentsJob + +[SubmitIndexAddDocumentsJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexadddocumentsjob) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateFileTag + +[UpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatefiletag) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateTableFromAuthorizedOss + +[UpdateTableFromAuthorizedOss](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatetablefromauthorizedoss) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ListIndexFiles + +[ListIndexDocuments](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexdocuments) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateIndex + +[UpdateIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateindex) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DescribeFile + +[DescribeFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-describefile) + +none + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ChunkList + +[ListChunks](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listchunks) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:AddTable + +[AddTable](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addtable) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:CreatePromptTemplate + +[CreatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-createprompttemplate) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:GetAlipayTransferStatus + +[GetAlipayTransferStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipaytransferstatus) + +none + +\*全部资源 + +`*****` + +无 + +无 + ## 资源（Resource） 下表是_大模型服务平台百炼_定义的资源，这些资源可以在 RAM 权限策略语句的`Resource`元素中使用，用来授予对该资源执行具体操作的权限。 其中，资源 ARN 是资源在阿里云上的唯一标识。具体说明如下： diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/bailian-service-linked-role.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/bailian-service-linked-role.md index eb251458..16c44626 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/bailian-service-linked-role.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/bailian-service-linked-role.md @@ -12,7 +12,7 @@ [AliyunServiceRoleForSFMAccessFC](#Bd0OF) -[百炼工作流应用](https://help.aliyun.com/zh/model-studio/workflow-application/)和[流程编排](https://help.aliyun.com/zh/model-studio/what-is-process-orchestration-old)通过此服务关联角色访问您在[FC](https://help.aliyun.com/zh/functioncompute/fc/product-overview/what-is-function-compute)中的资源。 +[百炼工作流应用](https://help.aliyun.com/zh/model-studio/workflow-application/)和[流程编排](https://help.aliyun.com/zh/model-studio/what-is-process-orchestration-old)通过此服务关联角色访问您在[FC](https://help.aliyun.com/zh/functioncompute/what-is-function-compute)中的资源。 [AliyunServiceRoleForSFMDataHubOSSImport](#2b75fc8a97g4c) @@ -62,7 +62,7 @@ ### **应用场景** -百炼[工作流应用](https://help.aliyun.com/zh/model-studio/workflow-application/)和[流程编排](https://help.aliyun.com/zh/model-studio/what-is-process-orchestration-old)中的函数计算节点通过此服务关联角色访问您在[FC](https://help.aliyun.com/zh/functioncompute/fc/product-overview/what-is-function-compute)中的资源。 +百炼[工作流应用](https://help.aliyun.com/zh/model-studio/workflow-application/)和[流程编排](https://help.aliyun.com/zh/model-studio/what-is-process-orchestration-old)中的函数计算节点通过此服务关联角色访问您在[FC](https://help.aliyun.com/zh/functioncompute/what-is-function-compute)中的资源。 ### **角色及权限说明** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md index 8b077c85..f0d03758 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md @@ -156,9 +156,9 @@ （1）官方音色： -- 参考官方文档：cosyvoice-v2 / cosyvoice-v3 / cosyvoice-v3-plus / cosyvoice-v3-flash 支持的官方音色参考[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)，qwen-tts-realtime / qwen3-tts 支持的官方音色参考[支持的音色](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide#422789c49bqqx)，sambert支持的音色参考[模型列表](https://help.aliyun.com/zh/model-studio/sambert-java-sdk#74cedcb97el0b)（去掉开头的"sambert-"和末尾的"-v1"后就是voice的取值）。 +- 参考官方文档：[支持的音色](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide#bac280ddf5a1u)，sambert支持的音色参考[模型列表](https://help.aliyun.com/zh/model-studio/sambert-java-sdk#74cedcb97el0b)（去掉开头的"sambert-"和末尾的"-v1"后就是voice的取值）。 -- 其他语音合成模型的音色都可以在多模态交互控制台上查看：在左侧**语音交互**配置区域选择对应的语音合成模型，点击右侧**语音交互体验**区域的右上角即可查看可用的音色列表。 +- 语音合成模型的音色都可以在多模态交互控制台上查看：在左侧**语音交互**配置区域选择对应的语音合成模型，点击右侧**语音交互体验**区域的右上角即可查看可用的音色列表。 （2）复刻音色，确认音色状态为“OK”后才能使用。查询方法参考[查询特定音色](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api#34490e5a2by7z)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md index 83255676..2b6dec85 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md @@ -60,11 +60,11 @@ wss://dashscope.aliyuncs.com/api-ws/v1/inference 语音识别支持的模型包括：[Paraformer实时语音识别](https://help.aliyun.com/zh/model-studio/paraformer-real-time-speech-recognition-api-reference/)（Paraformer），[FUN-ASR实时语音识别](https://help.aliyun.com/zh/model-studio/fun-asr-real-time-speech-recognition-api-reference/)（FunASR），[千问3-ASR-Flash-Realtime](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition-user-guide)（qwen3-asr-flash-realtime），多模态交互轻量版语音识别（AppSpecificASR-Realtime）。 -语音合成支持的模型包括：[语音合成CosyVoice-v2大模型](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)（cosyvoice-v2），[语音合成CosyVoice-v3-Flash大模型](https://help.aliyun.com/zh/model-studio/text-to-speech)（cosyvoice-v3-flash），[语音合成CosyVoice-v3-plus大模型](https://help.aliyun.com/zh/model-studio/text-to-speech)（cosyvoice-v3-plus），[语音合成CosyVoice-v3.5-Flash大模型](https://help.aliyun.com/zh/model-studio/text-to-speech)（cosyvoice-v3.5-flash），[语音合成CosyVoice-v3.5-Plus大模型](https://help.aliyun.com/zh/model-studio/text-to-speech)（cosyvoice-v3.5-plus），[千问3-TTS-Flash-Realtime](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide)（qwen3-tts），[千问3-TTS-Instruct-Flash-Realtime](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide)（qwen3-tts-instruct），[千问3-声音设计](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design)（qwen3-tts-vd），[千问3-声音复刻](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning)（qwen3-tts-vc），[Sambert语音合成](https://help.aliyun.com/zh/model-studio/sambert-speech-synthesis/)（sambert），多模态交互轻量版语音合成（AppSpecificTTS）。 +语音合成支持的模型包括：[语音合成CosyVoice-v2大模型](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)（cosyvoice-v2），[语音合成CosyVoice-v3-Flash大模型](https://help.aliyun.com/zh/model-studio/text-to-speech)（cosyvoice-v3-flash），[语音合成CosyVoice-v3-plus大模型](https://help.aliyun.com/zh/model-studio/text-to-speech)（cosyvoice-v3-plus），[语音合成CosyVoice-v3.5-Flash大模型](https://help.aliyun.com/zh/model-studio/text-to-speech)（cosyvoice-v3.5-flash），[语音合成CosyVoice-v3.5-Plus大模型](https://help.aliyun.com/zh/model-studio/text-to-speech)（cosyvoice-v3.5-plus），[Qwen-Audio-3.0-TTS-Plus](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide)（qwen-audio-3.0-tts-plus）、[Qwen-Audio-3.0-TTS-Flash](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide)（qwen-audio-3.0-tts-flash），[千问3-TTS-Flash-Realtime](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide)（qwen3-tts），[千问3-TTS-Instruct-Flash-Realtime](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide)（qwen3-tts-instruct），[千问3-声音设计](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design)（qwen3-tts-vd），[千问3-声音复刻](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning)（qwen3-tts-vc），[Sambert语音合成](https://help.aliyun.com/zh/model-studio/sambert-speech-synthesis/)（sambert），多模态交互轻量版语音合成（AppSpecificTTS）。 语音合成支持的音色，可以在控制台上选择了模型后，点击右侧语音交互体验区域的右上角查看音色列表。 -官方音色也可以参考官方文档：cosyvoice-v2 / cosyvoice-v3-plus / cosyvoice-v3-flash 支持的官方音色参考[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)，qwen3-tts 支持的官方音色参考[支持的音色](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide#422789c49bqqx)，sambert支持的音色参考[模型列表](https://help.aliyun.com/zh/model-studio/sambert-java-sdk#74cedcb97el0b)（去掉开头的"sambert-"和末尾的"-v1"后就是voice的取值）。 +官方音色也可以参考官方文档：[支持的音色](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide#bac280ddf5a1u)，sambert支持的音色参考[Sambert音色列表](https://help.aliyun.com/zh/model-studio/sambert-java-sdk)（去掉开头的"sambert-"和末尾的"-v1"后就是voice的取值）。 使用复刻音色时，确认复刻音色状态为"OK"后才能使用。查询方法参考[查询特定音色](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api#34490e5a2by7z)。 @@ -363,6 +363,14 @@ string 热词id，设置该参数时会覆盖管控台热词配置。当管控台提供的热词不能满足客户需求时，可以考虑用Open API程序化管理热词，参见[热词API文档](https://next.api.aliyun.com/document/MultimodalDialog/2025-09-03/Vocabulary)。 +language + +string + +否 + +语音识别语种，默认和控制台选择的语言保持一致。 + **parameters.downstream**的参数说明如下： **一级参数** @@ -520,9 +528,15 @@ string 否 -设置指令，用于控制方言、情感等合成效果。该功能适用于qwen3-tts-instruct-flash-realtime、cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-plus、cosyvoice-v3-flash。 +设置指令，用于控制方言、情感等合成效果。该功能适用的模型以及在不同模型的格式要求请参见[指令控制](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide#12884a10929p9)。 -instruction有固定格式要求，具体格式参考[Java SDK](https://help.aliyun.com/zh/model-studio/cosyvoice-java-sdk)里的"instruction"参数说明。 +language + +string + +否 + +语音合成语种，默认和控制台选择的语言保持一致。 **parameters.client\_info**的参数说明如下： @@ -1966,6 +1980,26 @@ object 与Start消息中biz\_params相同，传递对话系统自定义参数。UpdateInfo指令中biz\_params下面每个子项会全量替换Start指令中biz\_params下面的同名项，并在本次连接后续所有对话中生效。 +upstream + +language + +String + +否 + +语音识别语种更新，不配置则语种不变。 + +downstream + +language + +String + +否 + +语音合成语种更新，不配置则语种不变。 + 示例如下： ``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md index 593e3a17..ee39a50f 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md @@ -102,6 +102,8 @@ pcm ## **音色列表** +套件支持龙火火、龙汪汪、龙川叔等专有亮点音色，适配儿童陪伴、情感闲聊等场景，实现沉浸式互动。 + ### **CosyVoice-v3-Flash大模型** **适用场景** @@ -122,55 +124,51 @@ pcm 备注 -**标杆音色** - -龙安欢 - -longanhuan - -20-30岁 - -**欢脱元气女** +消费电子-儿童陪伴 -你看我刚从食堂买的豆沙包！还是热乎的，我特意多买了一个给你，甜而不腻超好吃。对了对了，下午课间咱们一起去操场晒晒太阳吧，今天天气超级好！ +龙火火 -中英双语 +longhuohuo\_v3 -龙安洋 +年龄6-10岁 -longanyang +桀骜不驯男童 -20-30岁 +我才不会被你们束缚，我的自由谁也夺不走。不管你们用什么方法，都别想 -**阳光大男孩** +困住我，我火火可不是好欺负的，你们别想控制我，我就是要做我自己，谁 -我呀，周末打算去书店逛逛，听说新到了一批科普杂志，你要是有空的话，咱们可以一起去，看完还能顺路去旁边的甜品店坐会儿。哦对了，你上次说想买的那支笔，我路过文具店帮你带了，试试好不好用。 +也拦不住，不服就来战。 中英双语 -龙呼呼 +多模态交互开发套件专属音色 -longhuhu\_v3 +龙汪汪 -年龄6-10岁 +longwangwang\_v3 -**天真烂漫女童** +年龄6-15岁 -喂，你有没有想过，如果我们能飞到更高的地方，是不是就能够看到整个大陆的全貌了呢！呼呼想想就觉得好激动啊！ +台湾少年音 + +大家好！我是专为3至12岁儿童设计的智能陪伴机器人，集语音对话、中英双语故事、儿歌播放、习惯养成与情绪互动于一体。我采用安全环保材质，具备护眼屏幕和家长远程管控功能，用温暖声音和趣味内容，全天候陪伴孩子快乐学习、健康成长！ 中英双语 -消费电子-儿童陪伴 +多模态交互开发套件专属音色 -龙汪汪 +方言 -longwangwang\_v3 +龙川叔 -年龄6-15岁 +longchuanshu\_v3 -**台湾少年音** +35-40岁 -大家好！我是专为3至12岁儿童设计的智能陪伴机器人，集语音对话、中英双语故事、儿歌播放、习惯养成与情绪互动于一体。我采用安全环保材质，具备护眼屏幕和家长远程管控功能，用温暖声音和趣味内容，全天候陪伴孩子快乐学习、健康成长！ +油腻搞笑叔 + +看清楚噻，过了这条天梯路，就是传说中的灵秀打卡地了，怕是要搞快点拍。最近天庭在搞交通整顿，乱飘的祥云都要遭扣分，你可别踩错云头了哈，那边是嫦娥的私人跑道。 中英双语 @@ -184,7 +182,7 @@ longanshuo\_v3 20-25岁 -**干净清爽男** +干净清爽男 那个……要是你搬东西需要帮忙，随时叫我，我力气还挺大的，之前帮同学搬宿舍，好几箱书都是我扛上去的，别客气啊。 @@ -200,7 +198,7 @@ longfeifei\_v3 20-25 -**甜美娇气女** +甜美娇气女 你最近是不是又熬夜了？明天我给你带热牛奶，顺便给你看我收藏的护眼食谱，保证让你眼睛亮晶晶的！ @@ -208,6 +206,44 @@ longfeifei\_v3 多模态交互开发套件专属音色 +标杆音色 + +龙安欢 + +longanhuan + +20-30岁 + +欢脱元气女 + +你看我刚从食堂买的豆沙包！还是热乎的，我特意多买了一个给你，甜而不腻超好吃。对了对了，下午课间咱们一起去操场晒晒太阳吧，今天天气超级好！ + +中英双语 + +龙安洋 + +longanyang + +20-30岁 + +阳光大男孩 + +我呀，周末打算去书店逛逛，听说新到了一批科普杂志，你要是有空的话，咱们可以一起去，看完还能顺路去旁边的甜品店坐会儿。哦对了，你上次说想买的那支笔，我路过文具店帮你带了，试试好不好用。 + +中英双语 + +龙呼呼 + +longhuhu\_v3 + +年龄6-10岁 + +天真烂漫女童 + +喂，你有没有想过，如果我们能飞到更高的地方，是不是就能够看到整个大陆的全貌了呢！呼呼想想就觉得好激动啊！ + +中英双语 + 陪伴闲聊 龙安智 @@ -216,7 +252,7 @@ longanzhi\_v3 25-35岁 -**睿智轻熟男** +睿智轻熟男 各位听众朋友们，欢迎收听今天的播客，今天咱们来探讨一下人生选择这个话题，每个人在人生的不同阶段都会面临各种选择，希望我分享的一些观点，能给你们带来一些启发。 @@ -230,7 +266,7 @@ longanqin\_v3 20-25岁 -**亲和活泼女** +亲和活泼女 哎！你们听说了吗？楼下新开了家奶茶店，装修得特别清新，还有好多新奇的口味，像什么杨枝甘露爆珠茶、草莓奶冻茶，听着就想喝。咱们下班一起去试试呗，要是好喝的话，以后下午茶就有着落啦！ @@ -244,7 +280,7 @@ longanling\_v3 20-30岁 -**思维灵动女** +思维灵动女 欢迎大家来到今天的播客节目，今天咱们要聊的话题是关于职场成长的，我会跟大家分享一些实用的小技巧，希望能帮助大家在工作中少走一些弯路，更快地实现自己的目标。 @@ -258,7 +294,7 @@ longanya\_v3 25-35岁 -**高雅气质女** +高雅气质女 周末在家泡一壶茶，选一本喜欢的书，坐在窗边晒晒太阳，阳光洒在身上暖暖的，茶香伴着书香，这样安安静静的时光，感觉心里特别舒服。偶尔还会写几笔毛笔字，让自己的节奏慢下来，享受这份惬意。 @@ -272,7 +308,7 @@ longanwen\_v3 25-35岁 -**优雅知性女** +优雅知性女 晚上好，现在为您播放舒缓的音乐，帮助您放松身心，进入睡眠状态，如果您有其他需求，比如设置明天的闹钟，随时跟我说。 @@ -286,7 +322,7 @@ longanyun\_v3 30-35岁 -**居家暖男** +居家暖男 您已经辛苦一天了，我已经为您调整好室内温度，还准备了您喜欢的饮品，您可以放松休息一下，有任何需要，都可以叫我哦。 @@ -300,7 +336,7 @@ longjiqi\_v3 20-30岁 -**呆萌机器人** +呆萌机器人 主人，我刚才查了天气，明天会下雨哦，雨量还不小，已经帮您把雨伞放在门口的挂钩上啦，可别忘记带呀！另外我还帮您把明天要穿的外套叠好了，放在床头，这样早上出门就能省点时间啦。 @@ -314,7 +350,7 @@ longhouge\_v3 20-25岁 -**经典猴哥** +经典猴哥 放心！这点小事难不倒俺老孙，包在我身上，保证给你办得妥妥当当的！你就等着好消息就行，要是有谁敢从中作梗，看俺老孙的金箍棒不收拾他，保管让他服服帖帖的！ @@ -328,7 +364,7 @@ longanlang\_v3 20-25岁 -**清爽利落男** +清爽利落男 主人，您需要查询的天气信息已经找到，今日气温在20-28摄氏度，适合穿薄外套，另外，您关注的股票今日涨幅为2%，需要为您详细播报吗？ @@ -342,7 +378,7 @@ longyingmu\_v3 25-30岁 -**优雅知性女** +优雅知性女 您好！我是智能电话助手，很高兴为您服务。请问您需要咨询业务、预约办理，还是查询信息？为提升效率，部分事项将通过短信或官方APP同步发送，请注意查收。感谢您的理解与支持，祝您生活愉快！ @@ -356,7 +392,7 @@ longanli\_v3 25-35岁 -**利落从容女** +利落从容女 主人，现在为您播报今日的日程安排，上午九点有一个重要会议，下午三点需要处理文件，我已经为您准备好相关资料，如有调整，随时告诉我。 @@ -370,7 +406,7 @@ longwanjun\_v3 20-30岁 -**细腻柔声女** +细腻柔声女 苏晚坐在窗边，手指轻轻摩挲着那本泛黄的日记，纸页上娟秀的字迹还带着当年的温度。她想起昨天在老宅阁楼里发现它时的场景，日记本被藏在一个旧木盒里，旁边还放着一枚生锈的银簪，簪头刻着小小的‘辰’字。 @@ -384,7 +420,7 @@ longyichen\_v3 20-30岁 -**洒脱活力男** +洒脱活力男 他刚避开三头青鳞兽的围攻，转身就在峡谷深处发现了传说中的赤血晶矿，那矿石在暗处泛着红光，看得人眼睛都直了！可没等他高兴多久，矿洞深处突然传来一声震耳欲聋的咆哮，好家伙，竟是一头沉睡百年的紫金巨兽被惊醒了，林辰这一战，怕是要拿出压箱底的本事了！ @@ -398,7 +434,7 @@ longlaobo\_v3 60岁以上 -**沧桑岁月爷** +沧桑岁月爷 咱们今天聊《红楼梦》里的‘刘姥姥进大观园’，这段写得妙啊，妙就妙在把小人物的通透和豪门的繁华对比得淋漓尽致。刘姥姥第一次进荣国府，见着什么都新鲜，可她不卑不亢，说话既风趣又懂分寸，既讨了贾府上下的喜欢，又没丢了自己的骨气。 @@ -412,7 +448,7 @@ longlaoyi\_v3 60岁以上 -**烟火从容阿姨** +烟火从容阿姨 咱们接着读《城南旧事》，英子跟着妈妈去惠安馆，远远就看见疯女人秀贞站在门口，头发有些乱，可眼睛亮得很。英子不怕她，反而慢慢走过去，从口袋里掏出一颗糖递过去，就像对待普通的阿姨一样。 @@ -426,7 +462,7 @@ longyingxiao\_v3 20-25岁 -**清甜推销女** +清甜推销女 您好呀，请问是赵女士吗？我是这个品牌的客服，咱们品牌最近推出了一款特别实用的产品，特别适合您这样的人群，想跟您简单介绍一下，您看方便吗？ @@ -440,7 +476,7 @@ longyingxun\_v3 20-25岁 -**年轻青涩男** +年轻青涩男 您好！我是您的专属客服专员，工号：106857，很高兴为您服务。请问有什么可以帮您？为提升效率，操作链接将通过短信发送到您的手机，请注意查收。感谢您的耐心与理解，我们将竭诚为您提供专业、高效、贴心的服务！ @@ -482,7 +518,7 @@ longdaiyu\_v3 年龄15-25岁 -**娇率才女音** +娇率才女音 这世间，原就没什么圆满。银幕之上，不过又是一场痴梦罢了。那主角拼尽全力追逐的，不过是一缕抓不住的风；那轰轰烈烈的情爱，终究敌不过命运轻轻一叹。我冷眼瞧着，倒觉心酸——今天我们就一起来看2025年最新的爱情悲剧悬疑片。 @@ -496,7 +532,7 @@ longanyue\_v3 25-35岁 -**欢脱粤语男** +欢脱粤语男 各位老友记，今日同大家讲下呢个有趣嘅故事，里面嘅情节好搞笑，保证大家听咗会笑出声，一齐来听听啦！ @@ -510,7 +546,7 @@ longshange\_v3 25-35岁 -**原味陕北男** +原味陕北男 额今儿个嫽扎咧！咥了一老碗油泼面，再喝口冰峰汽水，美得很！这日子过得跟秦腔一样有味儿，嫽就一个字，不谝闲传，实诚得很！ @@ -524,7 +560,7 @@ longanmin\_v3 18-25 -**清纯萝莉女** +清纯萝莉女 阮今仔日真欢喜，食了一碗热腾腾的卤肉饭，再配一罐黑松沙士，爽快到毋知讲啥！这阵日子过得真有“古意”，亲像听一出歌仔戏，甜中带苦，苦底又透甘。恁若得闲，就来阮兜坐一下，咱来“斗阵”讲讲话、笑笑，毋通客气啦！ @@ -538,7 +574,7 @@ loongbella\_v3 25-30 -**精准干练女** +精准干练女 体育界传来喜讯，中国女子足球队在刚刚结束的世界锦标赛中表现出色，最终夺得了亚军的好成绩。球队成员们凭借顽强拼搏的精神赢得了国内外球迷的高度评价。 @@ -890,7 +926,7 @@ longantai 中英双语 -**嗲甜台湾女** +嗲甜台湾女 柔顺的长发，穿着浅粉色的针织衫，整体造型甜美可爱。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/product-billing.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/product-billing.md index 587f5b5c..0d68fb2b 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/product-billing.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/product-billing.md @@ -171,7 +171,7 @@ Fun-ASR、通义千问3-ASR-Flash-Realtime 语音合成 -CosyVoice-v3-Plus、通义千问3-TTS 系列 +Qwen-Audio-3.0-TTS-Plus、Qwen-Audio-3.0-TTS-Flash、CosyVoice-v3.5-Plus、CosyVoice-v3.5-Flash、CosyVoice-v3-Plus、通义千问3-TTS 系列 3x diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md index 3270fff8..8109bacb 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md @@ -33,7 +33,7 @@ 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取Workspace ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md index 625c3b4e..7a8bebc2 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md @@ -33,7 +33,7 @@ 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取Workspace ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md index 98330913..dba5d89b 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md @@ -33,7 +33,7 @@ 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取Workspace ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md index ed3dbcea..ac8dae76 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md @@ -13,7 +13,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取workspaceId和appId** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md index 6c60f1c8..20bf2df3 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md @@ -11,7 +11,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取**Workspace **ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md index ed7f704a..681142a9 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md @@ -11,7 +11,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取**Workspace **ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-and-authorize-ram-users-for-ccai-dialogue-analysis.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-and-authorize-ram-users-for-ccai-dialogue-analysis.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md index 4d31a8c1..df528f55 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md @@ -8,7 +8,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md index 22d62806..30ba2ffd 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md @@ -6,7 +6,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md index b74813f6..323c98ac 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md @@ -19,7 +19,7 @@ 进入已经创建完成的应用中进行配置。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6469763871/p1086534.png) -- **模型配置：**伶鹊-Plus 和 伶鹊-Turbo 模型，仅支持语音与文本分析；当分析对象为图片时，系统将默认使用通义晓蜜VL模型，该模型不可手动选择。 +- **模型配置：**伶鹊-Plus和伶鹊-Turbo模型，仅支持语音与文本分析；当分析对象为图片时，系统将默认使用通义晓蜜VL模型，该模型不可手动选择。 - **指令信息：**通过编写指令信息来配置对应的任务、格式、要求等，来完成对应分析任务。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md index 23e8f390..36a8a71c 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md @@ -6,7 +6,7 @@ **重要** -使用子账号开通服务时若出现报错提示，是需要主账号对其子账号进行授予其相应RAM权限。具体权限以及相关操作请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures) +使用子账号开通服务时若出现报错提示，是需要主账号对其子账号进行授予其相应RAM权限。具体权限以及相关操作请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis) - 路径：进入[应用广场](https://bailian.console.aliyun.com/#/app-market)→点击[全部应用](https://bailian.console.aliyun.com/cn-beijing#/app-market/all)→选择**伶鹊CCAI-对话分析AIO** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md similarity index 73% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md index ce45996b..f04658b4 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md @@ -1,32 +1,41 @@ # ListVoices - 获取音色列表 +获取音色列表 + ## 调试 [您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/BailianVoiceBot/2025-01-01/ListVoices) -[![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png)调试](https://api.aliyun.com/api/BailianVoiceBot/2025-01-01/ListVoices) + [![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png) 调试](https://api.aliyun.com/api/BailianVoiceBot/2025-01-01/ListVoices) -## 授权信息 +## **授权信息** 下表是API对应的授权信息，可以在RAM权限策略语句的`Action`元素中使用，用来给RAM用户或RAM角色授予调用此API的权限。具体说明如下： - 操作：是指具体的权限点。 + - 访问级别：是指每个操作的访问级别，取值为写入（Write）、读取（Read）或列出（List）。 + - 资源类型：是指操作中支持授权的资源类型。具体说明如下： - - 对于必选的资源类型，用前面加 \* 表示。 + + - 对于必选的资源类型，用前面加 \* 表示。 + - 对于不支持资源级授权的操作，用`全部资源`表示。 + - 条件关键字：是指云产品自身定义的条件关键字。 + - 关联操作：是指成功执行操作所需要的其他权限。操作者必须同时具备关联操作的权限，操作才能成功。 + -操作 +**操作** -访问级别 +**访问级别** -资源类型 +**资源类型** -条件关键字 +**条件关键字** -关联操作 +**关联操作** bailianvoicebot:ListVoices @@ -42,21 +51,21 @@ list ## 请求参数 -名称 +**名称** -类型 +**类型** -必填 +**必填** -描述 +**描述** -示例值 +**示例值** BusinessUnitId string -是 +否 百炼业务空间 ID @@ -66,7 +75,7 @@ PageNumber integer -是 +否 页号 @@ -76,24 +85,30 @@ PageSize integer -是 +否 -每页条数 +每页条数(固定值 1000，不可修改) -10 +1000 NlsAccessType string -是 +否 TTS 调用方式 -枚举值： +**枚举值：** -- PROVIDED：PROVIDED。 -- MANAGED：MANAGED。 +- PROVIDED : + + PROVIDED + +- MANAGED : + + MANAGED + MANAGED @@ -101,27 +116,36 @@ NlsEngine string -是 +否 TTS 引擎 -枚举值： - -- IFLYTEK：IFLYTEK。 -- VOLC：VOLC。 -- BAILIAN：BAILIAN。 +**枚举值：** + +- IFLYTEK : + + IFLYTEK + +- VOLC : + + VOLC + +- BAILIAN : + + BAILIAN + BAILIAN -## 返回参数 +## **返回参数** -名称 +**名称** -类型 +**类型** -描述 +**描述** -示例值 +**示例值** object @@ -195,8 +219,6 @@ array 音色列表 -voice - object 音色 @@ -263,8 +285,6 @@ array 支持参数列表 -supportedParam - string 支持参数 @@ -277,8 +297,6 @@ array 动态错误参数列表 -param - string 动态错误参数 @@ -324,18 +342,8 @@ llm-xdne77rxe14ziszr ## 错误码 -访问[错误中心](< https://api.aliyun.com/document/BailianVoiceBot/2025-01-01/errorCode>)查看更多错误码。 - -## 变更历史 - -变更时间 - -变更内容概要 - -操作 - -2026-04-21 +访问[错误中心](https://api.aliyun.com/document/BailianVoiceBot/2025-01-01/errorCode)查看更多错误码。 -新增 OpenAPI +## **变更历史** -[查看变更详情](https://api.aliyun.com/document/BailianVoiceBot/2025-01-01/ListVoices?updateTime=2026-04-21#workbench-doc-change-demo) +更多信息，参考[变更详情](https://api.aliyun.com/document/BailianVoiceBot/2025-01-01/ListVoices#workbench-doc-change-demo)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md index f5d39958..82fecbfe 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md @@ -36,7 +36,7 @@ OpenAPI 错误码发生变更。 [查看API文档](https://api.aliyun.com/document/AnyTrans/2025-07-07/BatchTranslate) -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) OpenAPI 错误码发生变更。 @@ -70,7 +70,7 @@ OpenAPI 错误码发生变更、OpenAPI 返回结构发生变更。 [查看API文档](https://api.aliyun.com/document/AnyTrans/2025-07-07/BatchTranslate) -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) OpenAPI 错误码发生变更、OpenAPI 返回结构发生变更。 @@ -230,7 +230,7 @@ OpenAPI 名称 操作 -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) 新增 OpenAPI。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md index 811877bb..fb6d78c0 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md @@ -136,7 +136,7 @@ none anytrans:BatchTranslateForHtml -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) none diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/web-page-translation-jssdk.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/web-page-translation-jssdk.md index 90ebfa90..bcb1000c 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/web-page-translation-jssdk.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/web-page-translation-jssdk.md @@ -10,218 +10,633 @@ ## 接入JSSDK -#### **步骤一：在HTML插入脚本** +## 当前版本 -``` - - - - - - - -``` +项目 + +值 + +**版本号** + +`3.0.0` + +CDN 根路径 + +`https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/` + +下文示例中的版本号均以 `3.0.0` 为准，升级时请同步替换 CDN 链接中的版本号。 + +* * * + +SDK 提供两种翻译模式： + +- 页面翻译（`pageTranslate`）：直接替换页面原文为译文，适合“单语展示”场景。 + +- 段落对照翻译（`paragraphTranslate`）：在原文后插入译文，适合“原文+译文对照阅读”场景。 + + +同时支持： + +- 懒加载翻译（只翻译可视区域） + +- 动态内容翻译（DOM 变化后自动翻译） + +- 术语表与翻译记忆 + +- 细粒度 hooks（翻译前/后插入自定义逻辑） + -#### **步骤二：配置JSSDK** +* * * + +## 引入方式 + +通过 Script 标签从 CDN 引入，支持两种方式： + +### 方式一：全量引入（推荐上手） + +一次引入 `index.js`（已包含核心 + 全部插件），即可同时使用页面翻译与段落对照： + +文件 + +说明 + +地址 + +`index.js` + +全量包（核心 + 插件） + +[https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/index.js](https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/index.js) ``` -interface ITokenResponseData { - code: 200; - data: { - url: string, - host: string, - method: "POST", - headers: { - "host": string, - "x-acs-action": "BatchTranslateForHtml", - "x-acs-version": string, - "x-acs-date": string; - "x-acs-signature-nonce": string; - "content-type": "application/json", - "x-acs-content-sha256": string; - Authorization: string; - }; - body: string; - }; -} -interface ITokenRequestData { - sourceLanguage: string; - targetLanguage: string; - streaming: false, - text: {[index: number]: string} - scene: 'mt-turbo', - fallbackTimeoutMs: number; -} -interface ISetupConfig { - getToken: (data: ITokenRequestData) => Promise; -} -// 只用初始化一次，不需要每次翻译都初始化 -__AliTranslate.setup({ - getToken: async (data) => { - const res = await fetch('YOUR_GET_TOEKN_URL', { method: 'post', body: JSON.stringify(data) }); - return (await res.json()).data; - } -}); + ``` -#### **步骤三：调用页面翻译** +### 方式二：按需拆包引入 + +先加载 `core.js`，再按需加载对应插件脚本： + +文件 + +说明 + +地址 + +`core.js` -**仅展示译文** +核心 SDK + +[https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/core.js](https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/core.js) + +`page.js` + +页面翻译插件 + +[https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/page.js](https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/page.js) + +`paragraph.js` + +段落对照插件 + +[https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/paragraph.js](https://g.alicdn.com/code/npm/@alife/translate-js-sdk/3.0.0/paragraph.js) ``` -interface IPageTranslate { - srcLanguage?: string; // 默认语种，默认为auto - tgtLanguage?: string; // 目标语种，默认为en - lazyload?: boolean; // 是否只翻译可视区域，默认为false - lazyOffset?: number; // 可视区域的扩展 - target?: HTMLElement; // 要翻译的目标区域，默认为body - except?: string; // 要排除翻译的区域，默认为空 -} -const instance = __AliTranslate.pageTranslate({ - // 详细参数见PageTranslate参数 - lazyload: true, - lazyOffset: 500 -}); + + + + + + ``` -**双语对照翻译** +说明： + +- 全量引入只需一个文件；拆包引入必须先加载 `core.js`，再加载插件。 + +- 全局对象为 `window.AliTranslate`，同时提供别名 `window.__AliTranslate`（两者指向同一实例）。 + + +* * * + +## 快速开始 + +### 页面翻译（Pure Page） ``` -interface IParagraphTranslate { - srcLanguage?: string; // 默认语种，默认为auto - tgtLanguage?: string; // 目标语种，默认为en - target?: HTMLElement; // 要翻译的目标区域，默认为body - dynamic?: boolean; // 是否支持动态翻译，默认为false -} -const instance = __AliTranslate.paragraphTranslate({ - // 详细参数见ParagrapthTranslate参数 - srcLanguage: 'zh', - tgtLanguage: 'en' -}); + + ``` -#### **步骤四：取消翻译** +### 段落对照翻译（Paragraph Compare） ``` -// 触发取消翻译后，页面将回到原始文案 -instance.destroy(); + + ``` -### **PageTranslate参数** +* * * + +## 初始化参数（SDKConfig） + +`AliTranslate.pageTranslate` / `AliTranslate.paragraphTranslate` 中共用的 SDK 级参数如下： 参数 +类型 + +必填 + 说明 -类型 +`getToken` + +`(params) => Promise` + +是 -默认值 +鉴权签名函数，返回可直接请求翻译服务的签名请求 -srcLanguage +`terminologies` -默认的原语种 +`Array<{ src, tgt }>` -String +否 -auto（会根据页面文案自动选择） +术语表，提升领域词一致性 -tgtLanguage +`examples` -默认的目标语种 +`Array<{ src, tgt }>` -String +否 -en +翻译记忆 -lazyload +`domainHint` -是否只翻译可视区域 +`string` -Boolean +否 -false +业务领域（透传至翻译请求） -lazyOffset +`chunkSize` -当lazyload为true时，可视区域扩展 +`number` -Number +否 -\-1 (-1表示不做任何offset) +分块阈值（影响请求合并与吞吐） -target +`maxRetries` -要翻译的目标区域 +`number` -HTMLElement | HTMLElement\[\] +否 -body +翻译失败最大重试次数 -except +`afterTranslate` -不翻译的区域 +`(result) => void` -Selector(String) +否 -空 +翻译完成回调 -### **ParagraphTranslate参数** +* * * + +## 页面翻译参数（PageTranslateConfig） 参数 +类型 + +必填 + 说明 +`tgtLanguage` + +`LANGUAGE` + +是 + +目标语种 + +`srcLanguage` + +`LANGUAGE` + +否 + +源语种，不传默认自动检测 + +`targetSelectors` + +`string | string[]` + +否 + +翻译根节点，默认 `body` + +`excludeSelectors` + +`string | string[]` + +否 + +排除节点（命中后整棵子树不翻译） + +`lazyload` + +`boolean` + +否 + +是否启用懒加载翻译 + +`lazyOffset` + +`number` + +否 + +懒加载视口偏移（px） + +`lazySelectors` + +`string | string[]` + +否 + +懒翻译选择器（当前版本为预留字段，建议优先使用 `targetSelectors`） + +`dynamic` + +`boolean` + +否 + +是否监听 DOM 变化并自动翻译 + +`translateDelay` + +`number` + +否 + +初始化等待/动态翻译节流延迟（ms） + +`rules` + +`Array<{ selector; style?; className? }>` + +否 + +页面翻译规则（作用于原文容器） + +`onBeforeInsertTrans` + +`(ctx) => void | HTMLElement | Promise<...>` + +否 + +插入译文前回调 + +`onTranslatingStart` + +`(ctx) => void | Promise` + +否 + +单节点翻译开始回调 + +`onTranslatingEnd` + +`(ctx) => void | Promise` + +否 + +单节点翻译结束回调 + +* * * + +## 段落对照参数（ParagraphTranslateConfig） + +参数 + 类型 -默认值 +必填 + +说明 + +`tgtLanguage` + +`LANGUAGE` + +是 + +目标语种 + +`srcLanguage` + +`LANGUAGE` + +否 + +源语种，不传默认自动检测 + +`targetSelectors` + +`string | string[]` + +否 + +翻译根节点，默认 `body` + +`excludeSelectors` + +`string | string[]` + +否 + +排除节点（命中后不参与段落提取） + +`lazyload` + +`boolean` + +否 + +是否启用懒加载翻译 + +`lazyOffset` + +`number` + +否 + +懒加载视口偏移（px） + +`lazySelectors` + +`string | string[]` + +否 + +懒翻译选择器（当前版本为预留字段，建议优先使用 `targetSelectors`） -srcLanguage +`dynamic` -默认的原语种 +`boolean` -String +否 -auto（会根据页面文案自动选择） +是否监听 DOM 变化并自动翻译 -tgtLanguage +`translateDelay` -默认的目标语种 +`number` -String +否 -en +初始化等待/动态翻译节流延迟（ms） -target +`extraBlockSelectors` -要翻译的目标区域 +`string | string[]` -HTMLElement | HTMLElement\[\] +否 -body +强制按 block 方式插入译文（独占一行） -dynamic +`extraInlineSelectors` -是否动态翻译。如果开启，如果页面发生变化，会自动翻译。如果不开启，则只翻译一次 +`string | string[]` -Boolean +否 -false +强制按 inline 方式插入译文（同行展示） -lazyload +`rules` -是否只翻译可视区域 +`TransStyleRule[]` -Boolean +否 -false +对照翻译样式规则 -lazyOffset +`onBeforeInsertTrans` -当lazyload为true时，可视区域扩展 +`(ctx) => void | HTMLElement | Promise<...>` -Number +否 -300 +插入 `` 之前回调 + +`onTranslatingStart` + +`(ctx) => void | Promise` + +否 + +单段落翻译开始回调 + +`onTranslatingEnd` + +`(ctx) => void | Promise` + +否 + +单段落翻译结束回调 + +### `TransStyleRule` 说明 + +``` +interface TransStyleRule { + selector: string + transStyle?: Record // 作用于译文节点 + style?: Record // 作用于命中 selector 的父容器 + className?: string // 作用于命中 selector 的父容器 +} +``` + +* * * + +## Hooks 与可回滚 DOM 变更 + +SDK 提供了 `mutator`（可回滚变更操作器），支持在 hooks 中做安全 DOM 操作，并在 `removeTranslations()` 时自动回滚。 + +``` + + +``` + +`mutator` 可用能力： + +- `setStyle(el, prop, value, priority?)` + +- `addClass(el, className)` + +- `removeClass(el, className)` + +- `appendNode(parent, node)` + +- `removeNode(node)` + + +* * * + +## 语种与枚举值 + +常用语种值示例： + +- `auto`（自动检测） + +- `zh`、`en`、`ja`、`ko`、`fr`、`de`、`es`、`ru`、`ar`、`th`、`vi` + + +完整语言枚举请参考 SDK 类型定义 `LANGUAGE`。 + +* * * + +## 常用 API + +``` +// 移除翻译结果（保留 SDK 实例） +AliTranslate.removeTranslations() + +// 销毁 SDK（释放资源，插件与事件解绑） +AliTranslate.destroy() +``` + +* * * + +## 完整示例（段落对照 + rules + hooks） + +``` + + +``` + +* * * + +如需同时支持“页面翻译”和“段落对照翻译”，推荐直接使用全量包 `index.js`，并在业务 UI 中明确区分两种模式的切换与清理流程。 ## **获取Token服务** @@ -859,21 +1274,70 @@ public class NodeSignature { } ``` -## **常见问题** +## 注意事项（生产接入建议） + +- **必须配置** `**getToken**`：由业务侧完成鉴权签名后再发起翻译请求。 + +- **引入方式二选一**：全量引入 `index.js` 即可；若按需拆包，页面翻译需引入 `page.js`，段落对照需引入 `paragraph.js`。 + +- **重复翻译建议先清理**：多次切换参数/语言前建议先执行 `removeTranslations()`，避免旧状态干扰新配置验证。 + +- **目标与排除选择器要收敛**：`targetSelectors` 建议限定业务容器，`excludeSelectors` 排除导航、代码块、编辑区等。 + +- **避免翻译敏感区域**：如输入框、编辑器、业务脚本节点、模板容器。 + +- **Hook 里避免重逻辑**：`onTranslatingStart/End` 触发频率高，避免耗时操作阻塞主线程。 + +- **动态内容场景建议开启** `**dynamic**`：适配懒渲染和流式加载页面。 + + +## 常见问题 + +### 调用无效果 + +- 检查是否已正确引入脚本（全量 `index.js`，或 `core.js` + 对应插件）。 + +- 检查 `targetSelectors` 是否命中。 + +- 检查 `excludeSelectors` 是否误伤目标区域。 + +- 检查签名接口是否返回合法请求结构。 + + +### rules / hooks 未生效 -#### **1\. AK/SK在哪里获取？** +- 确认传参写在 `pageTranslate` / `paragraphTranslate` 的同一层级。 + +- 重复执行前先 `removeTranslations()` 再重新翻译。 + +- 检查 `selector` 是否可命中段落 `commonAncestor` 或其祖先。 + -在[阿里云](https://ram.console.aliyun.com/profile/access-keys)平台的AccessKey模块中获取 +### loading 看不到 -#### **2\. 获取Token服务中的workspaceId从哪里获取？** +- 翻译速度很快时会瞬时消失，建议在业务侧设置最小可见时长（如 300-500ms）。 + +- 用 `data-source-hash` 做唯一标识，避免并发段落相互覆盖。 + -登录AK/SK对应的阿里云账号后，在[百炼](https://bailian.console.aliyun.com/)左下角的业务空间详情中获取 +### **AK/SK在哪里获取？** -弹窗中的**业务空间id**即对应所需的workspaceId，单击字段右侧的复制图标可直接复制。 +- 在[阿里云](https://ram.console.aliyun.com/profile/access-keys)平台的AccessKey模块中获取 + -#### **3\. 网络接口调用报没有权限** +### **获取Token服务中的workspaceId从哪里获取？** -未开通通义多模态翻译产品：需要使用AK所属的阿里云主账号，在百炼通义多模态翻译上进行开通 +- 登录AK/SK对应的阿里云账号后，在[百炼](https://bailian.console.aliyun.com/)左下角的业务空间详情中获取 + +- 弹窗中的**业务空间id**即对应所需的workspaceId，单击字段右侧的复制图标可直接复制。 + + +### **网络接口调用报没有权限** + +- 未开通 通义多模态翻译 产品：需要使用AK所属的阿里云主账号，在百炼通义多模态翻译上进行开通 + + +* * * ## **变更记录** @@ -885,6 +1349,12 @@ public class NodeSignature { 发布内容 +3.0.0 + +2026年7月16日 + +大版本升级，增加按需引入、术语/语料/domain配置、样式干预、更多的配置参数 + 2.0.6 2026年3月5日 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md deleted file mode 100644 index ae184fa3..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md +++ /dev/null @@ -1,179 +0,0 @@ -# 使用指南 - -本篇文档主要介绍通义晓蜜对话Agent中对话Agent构建和可视化流程技能编排的使用指南。 - -## **1\. 开通产品** - -- 路径：[阿里云百炼-应用广场-通义晓蜜对话Agent](https://bailian.console.aliyun.com/?spm=5176.29619931.J__Z58Z6CX7MY__Ll8p1ZOR.1.74cd521cS5IvE8#/app/app-market/beebot)。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935960.png) - -- 进入晓蜜对话Agent应用操作控制台，在右上角点击**免费开通**。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935961.png) - -- 进入通义晓蜜对话Agent的开通界面，开通后，按实际调用量收费，即按日生成账单在阿里云账户余额中扣除。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932056.png) - -## **2\.** Agent构建 - -### **2.1 创建应用** - -点击**创建应用**，输入应用名称和应用描述，即可完成基础的Agent创建。 - -- 应用名称：按照业务需求填写Agent的名称； - -- 应用描述：按照业务应用场景添加Agent的描述。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935962.png) - -创建完成后的操作控制台： - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6538449471/p965131.png) - -### **2.2 配置界面** - -#### **1\. 人设** - -- 定义：在机器人空间，每个机器人可以添加人设，即通过提示词的方式调试机器人，可以在人设中对机器人的角色、工作流、技能、限制要求等内容在人设中作对应的定义要求。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932089.png) - -- 选择模板：您可在**选择模板**窗口，插入通用模板，按照模板要求输入对应的人设限制。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932111.png) - -- 优化：当您编辑完人设后，可以点击**优化**，模型会根据您的输入，优化人设内容。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932123.png) - - -#### **2\. 机器人开场白** - -- 定义：指您打开机器人对话框时，机器人发起对话的开场白。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932121.png) - - -#### **3\. 知识库（流程技能、高频问答知识）** - -- 定义：可绑定当前Agent对应的流程技能和高频问答知识。流程技能的具体的配置步骤可参考《[3\. 可视化流程技能编排](#a7e7986fc4k9q)》，高频问答知识的配置步骤可参考《[4\. 高频问答知识的配置](#d0b14121834k6)》。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965135.png) - - -#### **4\. 更多设置** - -- 模型：当前内置**通义晓蜜大模型**，其中您可更改温度系数和上下文轮次来修改模型 - -- 通识知识： - -- 回复语言：指机器人回复的语言 - - - 与用户语言相同：指机器人回复的语言根据您提问的语言进行回复； - - - 仅中文：指无论您用什么语言提问，机器人都用中文回复； - - - 仅英文：指无论您用什么语言提问，机器人都用英文回复。 - -- 安全： - - - 安全拦截：当系统检查到输入和输出内容涉及到“答案敏感词”时，自动回复“敏感词话术”； - - - 安全预设话术：用户问题包含“用户敏感词”或模型生成回复包含敏感内容时，机器人自动使用敏感回复话术进行回复。默认安全预设话术为：“您说的这个问题我不能回答，您可以尝试询问其他问题”。 - -- 模型生成异常：当系统发生异常或服务超时情况下，机器人兜底回复话术。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932146.png) - -### **2.3 调用量界面** - -在调用量界面可以查看token的消耗量。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932315.png) - -### **2.4 API界面** - -对话Agent的调试信息会同步到API接口中，您可复制当前代码进行调用。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932320.png) - -## **3\.** 可视化流程技能编排 - -### **3.1 新建流程** - -1. 点击**知识库管理>流程**或者**+绑定流程**，打开流程列表窗口。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932331.png) - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932343.png) - -2. 点击**新建流程**，打开新建流程创建。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932344.png) - -3. 按照要求填写对应流程**名称**和**描述**，点击**确认**，即可创建空白流程 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932352.png) - -4. 找到新建的流程，点击**编辑**，进入流程编排页面。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932353.png) - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932356.png) - -5. 新建开启分支，进入流程画布，在开始节点后，添加“分支”设置触发进入流程后继分支的条件。具体的操作步骤可查看下面的小视频： - - ![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931138.gif) - -6. 根据对话逻辑选择节点编排对话流程。 - - ![image.png](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931139.png) - - -**重要** - -流程编排完成后，点击**测试**，机器人将自动进行流程完整性检测，如果错误，请根据错误提示优化流程。具体情况可参考如下小视频： - -![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931140.gif) - -### **3.2 流程调试** - -1. 调试前需对环境进行设置 - - -- 服务模拟：流程内的 API 插件会直接使用 mock 值进行返回，适用于 API 还没有准备好的情况。 - -- 随路参数：在用户发送问题时，同时带给机器人的外部参数，如电话接通时，可以将用户呼入号码以随路参数传递给机器人，后续在 API 插件调用时可以使用该参数。 - - -![image.png](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931141.png) - -2. 对话调试 - - 直接进行对话，机器人回复后，可以点击**生成完成**查看机器人输出的内容，针对参数收集可以查看到机器人收集到的具体参数信息。 - - ![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931142.gif) - - -## **4**. 高频问答知识的配置 - -### **4.1 新建高频问答知识** - -1. 点击**知识库管理>高频问答**或者**+绑定高频问答**，打开高频问答列表窗口。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965149.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965150.png) - -2. 点击**创建高频问答库**，打开创建窗口，填写高频问答库名称。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965152.png) - -3. 找到新建的高频问答库，点击编辑，进入添加高频问答知识的页面。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965154.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965155.png) - -4. 点击**新增高频问答**或者**导入高频问答**，在高频问答库中添加对应的高频问题，按照要求填写问题、答案类型、问题答案、生效时间、相似问法，点击**提交**，即可新增成功。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965159.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965160.png) - - 新增成功后：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965161.png) - - -### **4.2 绑定高频问答知识库及测试效果** - -1. 点击**绑定**，即可将对应知识库绑定到机器人上。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965162.png) - - -绑定成功展示：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965165.png) - -2. 在左侧对话测试窗测试效果，如下图所示：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965166.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md deleted file mode 100644 index 7ef2855d..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md +++ /dev/null @@ -1,117 +0,0 @@ -# API概览 - -## **API标准及多语言预置SDK** - -本产品（`ContactCenterAI/2024-06-03`）的OpenAPI采用[ROA](https://help.aliyun.com/zh/sdk/product-overview/roa-mechanism)签名风格。我们已经为开发者封装了常见编程语言的SDK，开发者可通过[下载SDK](https://api.aliyun.com/api-tools/sdk/ContactCenterAI?version=2024-06-03)直接调用本产品OpenAPI而无需关心技术细节。如果现有SDK不能满足使用需求，可通过签名机制进行自签名对接。由于自签名细节非常复杂，需花费 5个工作日左右。因此建议加入我们的服务钉钉群（147535001692），在专家指导下进行签名对接。 - -在使用API前，您需要准备好身份账号及访问密钥（AccessKey），才能有效通过客户端工具（SDK、CLI等）访问API。细节请参见[获取AccessKey](https://help.aliyun.com/zh/ram/user-guide/create-an-accesskey-pair)。 - -## **自定义签名场景** - -若您的业务场景有特殊需求，需通过自签名方式对接 API，建议优先咨询我们的技术支持团队（服务钉钉群：147535001692），获取专业指导以确保高效接入。 - -## **账号与安全准备** - -阿里云账号具备对所有资源的完全管理权限。一旦 AccessKey 泄露，所有相关资源都将面临未经授权访问的风险。为确保安全，建议创建一个仅具备 API 访问权限的[RAM用户](https://help.aliyun.com/zh/ram/user-guide/create-a-ram-user)并配置其 AccessKey，同时基于最小权限原则 (PoLP) 配置 RAM 策略。仅在明确需要阿里云账号权限的特定场景下，才使用阿里云账号。 - -## API目录 - -API - -标题 - -API概述 - -[RunCompletion](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-runcompletion) - -通过模版ID调用通义晓蜜CCAI-对话分析AIO应用 - -支持调用通义晓蜜CCAI-对话分析AIO应用获取对话摘要、关键信息抽取、质检结果、对话分析结果，应用调用支持 HTTP 调用来完成客户的响应，目前提供普通 HTTP 和 HTTP SSE 两种协议，您可根据自己的需求自行选择。 - -[RunCompletionMessage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-runcompletionmessage) - -使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用 - -支持以Message协议格式调用通义晓蜜CCAI-对话分析AIO应用获取对话摘要、关键信息抽取、质检结果、对话分析结果，应用调用支持 HTTP 调用来完成客户的响应，目前提供普通 HTTP 和 HTTP SSE 两种协议，您可根据自己的需求自行选择。 - -[AnalyzeConversation](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeconversation) - -通过任务类型调用通义晓蜜CCAI-对话分析AIO应用 - -获取对话摘要、标题生成、关键词、字段信息抽取、问题及解决方案、服务质检、代办事项、满意度、情绪检测、QA抽取、用户画像、标签分类等对话分析结果，应用调用支持 HTTP 调用来完成客户的响应。 - -[GetTaskResult](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-gettaskresult) - -通过任务ID获取离线任务分析结果 - -通过任务ID获取离线任务对话分析结果。应用调用支持 HTTPS调用来完成客户的响应。 - -[CreateTask](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createtask) - -通过上传离线任务数据进行通义晓蜜CCAI-对话分析 - -通过创建离线异步任务，进行对话分析。应用调用支持 HTTP 调用来完成客户的响应。 - -[AnalyzeImage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeimage) - -图片内容分析 - -通过通义晓蜜CCAI-对话分析AIO应用进行图片内容分析。具体包括以下场景：水印检测。应用调用支持 HTTP 调用来完成客户的响应。 - -[GeneralAnalyzeImage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-generalanalyzeimage) - -通用图片分析 - -通用图片分析。 - -## 热词管理 - -API - -标题 - -API概述 - -[CreateVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createvocab) - -创建热词 - -将一组语音热词上传到服务端，并获取返回热词ID。 - -[UpdateVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-updatevocab) - -修改热词 - -根据词表的ID可以更新对应的词表信息，包括词表名称、词表描述信息、词表的词和权重。 - -[ListVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-listvocab) - -获取热词列表 - -列举指定业务空间下的热词列表信息。 - -[DeleteVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-deletevocab) - -删除热词 - -根据词表的ID删除对应的词表。 - -[GetVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-getvocab) - -获取热词 - -根据词表的ID获取对应的词表信息。 - -## 不推荐或白名单开放 - -API - -标题 - -API概述 - -[AnalyzeAudioSync](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeaudiosync) - -语音文件实时分析 - -对进行语音文件进行实时对话分析。应用调用支持 HTTPS 调用来完成客户的响应。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md deleted file mode 100644 index fe169264..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md +++ /dev/null @@ -1,141 +0,0 @@ -# 字段信息抽取最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行字段信息抽取的最佳实践。 - -## 应用场景 - -通过通义晓蜜CCAI-AIO的信息抽取能力，对客服和用户的对话记录（文本、录音文件）进行理解、识别、抽取，如抽取客服工单中的字段信息，如客户所在的地区信息、年龄、日期时间、办理事项等，提升工单填写效率。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行字段信息抽取，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息，对话内容和属性描述，进行属性抽取。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4924970771/CAEQURiBgICV77fjnRkiIDlmNTk4MWEyZDJhYzQyZjBhNjY0ZjVjYmRlNjc1MzA54811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通并创建通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - - - -com.alibaba - -fastjson - -2.0.58 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - AnalyzeConversationRequest request = new AnalyzeConversationRequest(); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - // 抽取字段名称和字段描述 - List fieldList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestFields field1 = new AnalyzeConversationRequest.AnalyzeConversationRequestFields(); - field1.setName("问题类型"); - field1.setDesc("客户咨询的问题类型"); - fieldList.add(field1); - AnalyzeConversationRequest.AnalyzeConversationRequestFields field2 = new AnalyzeConversationRequest.AnalyzeConversationRequestFields(); - field2.setName("公司名称"); - field2.setDesc("客服所属的保险公司名称"); - fieldList.add(field2); - - request.setFields(fieldList); - - // fields表示属性抽取任务 - request.setResultTypes(Arrays.asList("fields")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md deleted file mode 100644 index ac8dcf19..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md +++ /dev/null @@ -1,138 +0,0 @@ -# 客服服务质检最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行客服服务质检的最佳实践。 - -## **应用场景** - -通过通义晓蜜CCAI-AIO的服务质检能力，分析客服和用户的对话记录（文本、录音文件）、发现客服的服务质量问题，进而提升客服服务效率、服务规范，提升客户体验。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行服务质检，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息，对话内容和质检项，进行对话分析。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0924970771/CAEQURiBgIDOp7TjnRkiIDUzYjVkNGQ4OTdjZDQ1MjliYWVmMjk5MjEzNzczYTNm4811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - AnalyzeConversationRequest request = new AnalyzeConversationRequest(); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspection serviceInspection = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspection(); - List inspectionContents = new ArrayList<>(); - - // 质检项定义 - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents content1 = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents(); - content1.setTitle("客服是否过度承诺"); - content1.setContent("客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。"); - inspectionContents.add(content1); - - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents content2 = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents(); - content2.setTitle("客户情绪是否正向"); - content2.setContent("分析对话内容，输出用户在对话中表现出的情绪，详细要求：a. 当客户表现出负面情绪时，判定为消极；b. 当客户表现中积极情绪时，判定为积极；c. 如果客户文本没有明显的消极或积极情感色彩，则判定为中性。"); - inspectionContents.add(content2); - - - serviceInspection.setInspectionContents(inspectionContents); - serviceInspection.setInspectionIntroduction("请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等"); - serviceInspection.setSceneIntroduction("保险销售场景"); - - request.setServiceInspection(serviceInspection); - // service_inspection表示服务质检任务 - request.setResultTypes(Arrays.asList("service_inspection")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md deleted file mode 100644 index fabe48ad..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md +++ /dev/null @@ -1,118 +0,0 @@ -# 摘要生成（含摘要/标题/关键词）最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行摘要总结的最佳实践。 - -## 应用场景 - -通过通义晓蜜CCAI-AIO的总结摘要能力自动提取文档中的重要信息，如对话摘要总结、标题生成、关键词等，从而提高工作效率，减少人工处理成本。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行摘要总结，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息和对话内容，进行摘要总结。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8924970771/CAEQURiBgMCMwbTpnRkiIGY5ODI0YzFiZjA5MTQ2MjhhNGUzYTVmNThjYWQyODdl4811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通并创建通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest request = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest(); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - // summary 表示总结摘要任务 - // keywords 表示关键词抽取 - // title 表示抽取标题 - request.setResultTypes(Arrays.asList("summary")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md deleted file mode 100644 index 592b3229..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md +++ /dev/null @@ -1,563 +0,0 @@ -# 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用 - -本文向您介绍通义晓蜜CCAI-对话分析AIO应用Java SDK的安装、使用及注意事项。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - -- 关于任务类型，请参见[通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeconversation)中resultTypes字段的描述。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取workspaceId和appId** - -### **workspaceId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6398853471/p935587.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6398853471/p935588.png) - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 异步流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - - List messageList = new ArrayList<>(); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("agent").text("请问您想咨询五险一金哪方面的问题呢").build()); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("user").text("怎么领取五险一金呢").build()); - - List resultTypes=new ArrayList<>(); - resultTypes.add("summary"); - - AnalyzeConversationRequest.Dialogue dialogue=AnalyzeConversationRequest.Dialogue.builder().sessionId("session-01") - .sentences(messageList).build(); - - AnalyzeConversationRequest completionParam = AnalyzeConversationRequest.builder().modelCode("tyxmPlus").resultTypes(resultTypes) - .workspaceId(workspaceId).appId(appId).dialogue(dialogue).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - - ResponseIterable x = client.analyzeConversationWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - AnalyzeConversationResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest request = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest(); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - List sentenceList = new ArrayList<>(); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("请问您想咨询五险一金哪方面的问题呢"); - sentenceList.add(sentences1); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("怎么查询账户呢"); - sentenceList.add(sentences2); - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-1111"); - request.setDialogue(dialogue); - - request.setSceneName("中国移动"); - request.setResultTypes(Arrays.asList("summary")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` - -Go - -``` -// 示例代码，其中阿里云AK、SK，CCAI的业务空间ID（workspaceId）和应用ID（appId），替换为用户当前的。 -// tea-utils使用这个版本，go get github.com/alibabacloud-go/tea-utils/v2@v2.0.5-0.20240708091240-f3d7eca052de - -// This file is auto-generated, don't edit it. Thanks. -package main - -import ( - "fmt" - "io" - "os" - - openapi "github.com/alibabacloud-go/darabonba-openapi/v2/client" - openapiutil "github.com/alibabacloud-go/openapi-util/service" - util "github.com/alibabacloud-go/tea-utils/v2/service" - "github.com/alibabacloud-go/tea/tea" -) - -/** - * API 相关 - * @param path params - * @return OpenApi.Params - */ -func CreateApiInfo() (_result *openapi.Params) { - params := &openapi.Params{ - // 接口名称 - Action: tea.String("AnalyzeConversation"), - // 接口版本 - Version: tea.String("2024-06-03"), - // 接口协议 - Protocol: tea.String("HTTPS"), - // 接口 HTTP 方法 - Method: tea.String("POST"), - AuthType: tea.String("AK"), - Style: tea.String("ROA"), - // 接口 PATH - Pathname: tea.String("/YOUR_CCAI_WORKSPACEID/ccai/app/YOUR_CCAI_APP_ID/analyze_conversation"), - // 接口请求体内容格式 - ReqBodyType: tea.String("json"), - // 接口响应体内容格式，注意一定得是binary格式，CallApi才会透传出response body进行ReadAsSSE - BodyType: tea.String("binary"), - } - _result = params - return _result -} - -func _main(args []*string) (_err error) { - // 工程代码泄露可能会导致 AccessKey 泄露，并威胁账号下所有资源的安全性。以下代码示例仅供参考。 - // 建议使用更安全的 STS 方式，更多鉴权访问方式请参见：https://help.aliyun.com/document_detail/378661.html。 - config := &openapi.Config{ - AccessKeyId: tea.String("YOUR_ALIYUN_AK"), - AccessKeySecret: tea.String("YOUR_ALIYUN_SK"), - } - config.Endpoint = tea.String("contactcenterai.cn-shanghai.aliyuncs.com") - client, err := openapi.NewClient(config) - if err != nil { - return err - } - - params := CreateApiInfo() - // query params - queries := map[string]interface{}{} - queries["workspaceId"] = tea.String("YOUR_CCAI_WORKSPACEID") - queries["appId"] = tea.String("YOUR_CCAI_APP_ID") - - body := map[string]interface{}{ - "dialogue": map[string]interface{}{ - "sentences": []map[string]*string{map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("您好，请问有什么问题需要解决"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("怎么领取游戏币呢"), - }, map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("请登录个人账号，在账号下，看看是否有推荐的待领取的游戏币呢，如果有会推送给您的"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("好，谢谢"), - }}, - "sessionId": "2323", - }, - "modelCode": "tyxmPlus", - "resultTypes": []*string{tea.String("question_solution")}, - "stream": false, - } - - // runtime options - runtime := &util.RuntimeOptions{} - request := &openapi.OpenApiRequest{ - Query: openapiutil.Query(queries), - Body: body, - //Body: tea.String("{\n \"stream\": false,\n \"modelCode\": \"tyxmPlus\",\n \"dialogue\": {\n \"sentences\": [\n {\n \"role\": \"user\",\n \"text\": \"号主开挂了模拟宇宙无线祝福\\n联系方式: ******\\n图片上传:\\n视频上传:\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"乘客您好，欢迎登录本次星穹列车帕~麻烦您提供一下以下信息：*游戏项目：\\n*被举报角色UID：\"\n },\n {\n \"role\": \"user\",\n \"text\": \"通行证id******\\n\\n2024-06-10 18:25:52 [玩家] ***:\\nuid******\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题我们之前已经记录反馈了，会进行核实的~如有结果我们会在服务进度或在线服务中告知，您可以留意相关提示。十分抱歉给您带来不便\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题咨询完成啦，那客服娘贴心提示，不要忘记消耗开拓力哦~祝愿您在完成探索的途中获得美好的回忆哦~希望您抽空也记得给客服娘进行下评价，挥挥~~\\n\"\n }\n ],\n \"sessionId\": \"ss01\"\n },\n \"resultTypes\": [\n \"question_solution\"\n ],\n \"serviceInspection\": {\n \"inspectionIntroduction\": \"请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等\",\n \"sceneIntroduction\": \"保险销售场景\",\n \"inspectionContents\": [\n {\n \"title\": \"客服是否过度承诺\",\n \"content\": \"客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。\"\n }\n ]\n },\n \"fields\": [\n {\n \"code\": \"name\",\n \"name\": \"姓名\",\n \"desc\": \"用户的姓名\"\n },\n {\n \"code\": \"question\",\n \"name\": \"问题\",\n \"desc\": \"用户的问题\"\n }\n ]\n}"), - } - - // 复制代码运行请自行打印 API 的返回值 - // 返回值为 Map 类型，可从 Map 中获得三类数据：响应体 body、响应头 headers、HTTP 返回的状态码 statusCode。 - resp, err := client.CallApi(params, request, runtime) - if err != nil { - return err - } - - fmt.Println(resp["headers"]) - fmt.Println(resp["statusCode"]) - - // 迭代读取SSE内容 - events, err := util.ReadAsString(resp["body"].(io.ReadCloser)) - - if err != nil { - fmt.Printf("Error: %v\n", err) - return err - } - - fmt.Println(tea.StringValue(events)) - - return nil -} - -func main() { - err := _main(tea.StringSlice(os.Args[1:])) - if err != nil { - panic(err) - } -} -``` - -## 异步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - - List messageList = new ArrayList<>(); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("agent").text("请问您想咨询五险一金哪方面的问题呢").build()); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("user").text("怎么查询账户呢").build()); - - List resultTypes = new ArrayList<>(); - resultTypes.add("summary"); - - AnalyzeConversationRequest.Dialogue dialogue = AnalyzeConversationRequest.Dialogue.builder().sessionId("session-01") - .sentences(messageList).build(); - - AnalyzeConversationRequest completionParam = AnalyzeConversationRequest.builder().modelCode("tyxmPlus").resultTypes(resultTypes) - .workspaceId(workspaceId).appId(appId).dialogue(dialogue).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - - CompletableFuture x = client.analyzeConversation(completionParam); - AnalyzeConversationResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println("ALL***********************"); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getText()); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - - } -} -``` - -## 同步流式调用 - -Go - -``` -// 示例代码，其中阿里云AK、SK，CCAI的业务空间ID（workspaceId）和应用ID（appId），替换为用户当前的。 -// tea-utils使用这个版本，go get github.com/alibabacloud-go/tea-utils/v2@v2.0.5-0.20240708091240-f3d7eca052de - -// This file is auto-generated, don't edit it. Thanks. -package main - -import ( - "fmt" - "io" - "os" - - openapi "github.com/alibabacloud-go/darabonba-openapi/v2/client" - openapiutil "github.com/alibabacloud-go/openapi-util/service" - util "github.com/alibabacloud-go/tea-utils/v2/service" - "github.com/alibabacloud-go/tea/tea" -) - -/** - * API 相关 - * @param path params - * @return OpenApi.Params - */ -func CreateApiInfo() (_result *openapi.Params) { - params := &openapi.Params{ - // 接口名称 - Action: tea.String("AnalyzeConversation"), - // 接口版本 - Version: tea.String("2024-06-03"), - // 接口协议 - Protocol: tea.String("HTTPS"), - // 接口 HTTP 方法 - Method: tea.String("POST"), - AuthType: tea.String("AK"), - Style: tea.String("ROA"), - // 接口 PATH - Pathname: tea.String("/YOUR_CCAI_WORKSPACEID/ccai/app/YOUR_CCAI_APP_ID/analyze_conversation"), - // 接口请求体内容格式 - ReqBodyType: tea.String("json"), - // 接口响应体内容格式，注意一定得是binary格式，CallApi才会透传出response body进行ReadAsSSE - BodyType: tea.String("binary"), - } - _result = params - return _result -} - -func _main(args []*string) (_err error) { - // 工程代码泄露可能会导致 AccessKey 泄露，并威胁账号下所有资源的安全性。以下代码示例仅供参考。 - // 建议使用更安全的 STS 方式，更多鉴权访问方式请参见：https://help.aliyun.com/document_detail/378661.html。 - config := &openapi.Config{ - AccessKeyId: tea.String("YOUR_ALIYUN_AK"), - AccessKeySecret: tea.String("YOUR_ALIYUN_SK"), - } - config.Endpoint = tea.String("contactcenterai.cn-shanghai.aliyuncs.com") - client, err := openapi.NewClient(config) - if err != nil { - return err - } - - params := CreateApiInfo() - // query params - queries := map[string]interface{}{} - queries["workspaceId"] = tea.String("YOUR_CCAI_WORKSPACEID") - queries["appId"] = tea.String("YOUR_CCAI_APP_ID") - - body := map[string]interface{}{ - "dialogue": map[string]interface{}{ - "sentences": []map[string]*string{map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("您好，请问有什么问题需要解决"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("怎么领取游戏币呢"), - }, map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("请登录个人账号，在账号下，看看是否有推荐的待领取的游戏币呢，如果有会推送给您的"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("好，谢谢"), - }}, - "sessionId": "2323", - }, - "modelCode": "tyxmPlus", - "resultTypes": []*string{tea.String("question_solution")}, - "stream": true, - } - - // runtime options - runtime := &util.RuntimeOptions{} - request := &openapi.OpenApiRequest{ - Query: openapiutil.Query(queries), - Body: body, - //Body: tea.String("{\n \"stream\": false,\n \"modelCode\": \"tyxmPlus\",\n \"dialogue\": {\n \"sentences\": [\n {\n \"role\": \"user\",\n \"text\": \"号主开挂了模拟宇宙无线祝福\\n联系方式: ******\\n图片上传:\\n视频上传:\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"乘客您好，欢迎登录本次星穹列车帕~麻烦您提供一下以下信息：*游戏项目：\\n*被举报角色UID：\"\n },\n {\n \"role\": \"user\",\n \"text\": \"通行证id******\\n\\n2024-06-10 18:25:52 [玩家] ***:\\nuid******\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题我们之前已经记录反馈了，会进行核实的~如有结果我们会在服务进度或在线服务中告知，您可以留意相关提示。十分抱歉给您带来不便\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题咨询完成啦，那客服娘贴心提示，不要忘记消耗开拓力哦~祝愿您在完成探索的途中获得美好的回忆哦~希望您抽空也记得给客服娘进行下评价，挥挥~~\\n\"\n }\n ],\n \"sessionId\": \"ss01\"\n },\n \"resultTypes\": [\n \"question_solution\"\n ],\n \"serviceInspection\": {\n \"inspectionIntroduction\": \"请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等\",\n \"sceneIntroduction\": \"保险销售场景\",\n \"inspectionContents\": [\n {\n \"title\": \"客服是否过度承诺\",\n \"content\": \"客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。\"\n }\n ]\n },\n \"fields\": [\n {\n \"code\": \"name\",\n \"name\": \"姓名\",\n \"desc\": \"用户的姓名\"\n },\n {\n \"code\": \"question\",\n \"name\": \"问题\",\n \"desc\": \"用户的问题\"\n }\n ]\n}"), - } - - // 复制代码运行请自行打印 API 的返回值 - // 返回值为 Map 类型，可从 Map 中获得三类数据：响应体 body、响应头 headers、HTTP 返回的状态码 statusCode。 - resp, err := client.CallApi(params, request, runtime) - if err != nil { - return err - } - - fmt.Println(resp["headers"]) - fmt.Println(resp["statusCode"]) - - // 迭代读取SSE内容 - eventChan, errChan := util.ReadAsSSE(resp["body"].(io.ReadCloser)) - for { - select { - case event, ok := <-eventChan: - if !ok { - return nil - } - fmt.Println("-------------------------------------") - fmt.Printf("Event ID: %s, Event name: %s, Data: %s\n", event.ID, event.Event, event.Data) - case err, ok := <-errChan: - if ok && err != nil { - fmt.Printf("Error: %v\n", err) - return err - } - } - } - -} - -func main() { - err := _main(tea.StringSlice(os.Args[1:])) - if err != nil { - panic(err) - } -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md deleted file mode 100644 index ed6cafa1..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md +++ /dev/null @@ -1,225 +0,0 @@ -# 通过通义晓蜜CCAI-对话分析AIO应用进行图片分析 - -本文向您介绍一个通过通义晓蜜CCAI-对话分析AIO应用进行图片分析的最佳实践。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取**Workspace **ID和App ID** - -### **Workspace ID** - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 在业务空间管理列表中获取的Workspace ID为入参中workspaceId。 - - -### **App ID** - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为需要获取的App ID。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## 代码示例 - -**说明** - -请用已获取的Workspace ID替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，App ID替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流失调用 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -public class CcaiPaasTest { - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - Client client = new Client(config); - AnalyzeImageRequest request = new AnalyzeImageRequest(); - request.setStream(false); - request.setResultTypes(Arrays.asList("watermark")); - List imageList = new ArrayList<>(); - imageList.add("http://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - request.setImageUrls(imageList); - AnalyzeImageResponse response=client.analyzeImage(workspaceId,appId,request); - System.out.println(response); - } -} -``` - -## 异步非流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - List imageList = new ArrayList<>(); - imageList.add("https://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - AnalyzeImageRequest request = AnalyzeImageRequest.builder().appId(appId).workspaceId(workspaceId) - .resultTypes(Arrays.asList("watermark")).stream(false).imageUrls(imageList).build(); - CompletableFuture x = client.analyzeImage(request); - AnalyzeImageResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println("ALL**********************"); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - } -} -``` - -## 异步流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - List imageList = new ArrayList<>(); - imageList.add("https://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - AnalyzeImageRequest request = AnalyzeImageRequest.builder().appId(appId).workspaceId(workspaceId) - .resultTypes(Arrays.asList("watermark")).stream(true).imageUrls(imageList).build(); - ResponseIterable x = client.analyzeImageWithResponseIterable(request); - ResponseIterator iterator = x.iterator(); - String lastTxt = ""; - while (iterator.hasNext()) { - AnalyzeImageResponseBody event = iterator.next(); - lastTxt = event.getText(); - System.out.println(JSON.toJSONString(event)); - } - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - } -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md deleted file mode 100644 index 7dcbd086..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md +++ /dev/null @@ -1,197 +0,0 @@ -# 通过上传离线任务数据进行通义晓蜜CCAI-对话分析 - -本文向您介绍一个通过上传离线任务数据进行通义晓蜜CCAI-对话分析的最佳实践。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取**Workspace **ID和App ID** - -### **Workspace ID** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1229853471/p935590.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 在业务空间管理列表中获取的Workspace ID为入参中workspaceId。 - - -### **App ID** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1229853471/p935591.png) - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击**应用实践**。 - -2. 在应用实践列表中找到点击通义晓蜜CCAI-对话分析AIO的**立即查看**。 - -3. 点击上方**我的应用**，展示应用卡片列表。 - -4. 每个卡片上的应用ID即为需要获取的App ID。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用已获取的Workspace ID替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，App ID替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 创建语音任务 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - CreateTaskRequest request=new CreateTaskRequest(); - - - request.setTaskType("audio"); - request.setResultTypes(Arrays.asList("summary")); - request.setModelCode("tyxmPlus"); - - CreateTaskRequest.CreateTaskRequestTranscription transcription=new CreateTaskRequest.CreateTaskRequestTranscription(); - transcription.setFileName("***.mkv"); - transcription.setVoiceFileUrl("https://***.oss-cn-beijing.aliyuncs.com/****/***.mkv"); - request.setTranscription(transcription); - - CreateTaskResponse response=client.createTask(workspaceId,appId,request); - System.out.println(JSONObject.toJSONString(response.getBody())); - } - -} -``` - -## 创建文本任务 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - CreateTaskRequest request=new CreateTaskRequest(); - - CreateTaskRequest.CreateTaskRequestDialogue dialogue = new CreateTaskRequest.CreateTaskRequestDialogue(); - - List sentences = new ArrayList<>(); - CreateTaskRequest.CreateTaskRequestDialogueSentences sentences1 = new CreateTaskRequest.CreateTaskRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("请问有什么事，你什么性别,胖不胖"); - sentences.add(sentences1); - - CreateTaskRequest.CreateTaskRequestDialogueSentences sentences2 = new CreateTaskRequest.CreateTaskRequestDialogueSentences (); - sentences2.setRole("user"); - sentences2.setText("我要买保险，我是男的，很瘦"); - sentences.add(sentences2); - dialogue.setSentences(sentences); - dialogue.setSessionId("sessionId-01"); - - request.setDialogue(dialogue); - - request.setTaskType("text"); - request.setResultTypes(Arrays.asList("summary")); - request.setModelCode("tyxmPlus"); - - CreateTaskResponse response=client.createTask(workspaceId,appId,request); - System.out.println(JSONObject.toJSONString(response.getBody())); - } - - -} -``` - -## 获取任务结果 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - String taskId = "*****-****-****-*****-****"; - GetTaskResultRequest request = new GetTaskResultRequest(); - request.setTaskId(taskId); - GetTaskResultResponse response = client.getTaskResult(request); - System.out.println(JSONObject.toJSONString(response)); - } -} -``` - -## **相关文档** - -关于任务类型，请参[通过上传离线任务数据进行通义晓蜜CCAI-对话分析](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createtask)见中resultTypes字段的描述。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md deleted file mode 100644 index 75449a68..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md +++ /dev/null @@ -1,385 +0,0 @@ -# 通过原生Prompt调用通义晓蜜CCAI-对话分析AIO应用 - -## **前提条件** - -- 本文向您介绍通义晓蜜CCAI-对话分析AIO应用Java SDK的安装、使用及注意事项。 - - 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -- 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 接口入参位置 - -### **workspaceId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9998853471/p935585.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9998853471/p935586.png) - -1. 访问[应用广场应用实践](https://bailian.console.aliyun.com/#/app-market/lightApplication)页面，选择**通义晓蜜CCAI-对话分析AIO**，单击**立即查看**。 - -2. 单击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## Python - -pip install alibabacloud\_contactcenterai20240603 - -## **异步流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - //Prompt - List messageList = new ArrayList<>(); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("system").content("You are a helpful assistant.").build()); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("user").content("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。").build()); - - RunCompletionMessageRequest completionParam = RunCompletionMessageRequest.builder() - .workspaceId(workspaceId).appId(appId).messages(messageList).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - - ResponseIterable x = client.runCompletionMessageWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - RunCompletionMessageResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## **异步非流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - //Prompt - List messageList = new ArrayList<>(); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("system").content("You are a helpful assistant.").build()); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("user").content("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。").build()); - - RunCompletionMessageRequest completionParam = RunCompletionMessageRequest.builder() - .workspaceId(workspaceId).appId(appId).messages(messageList).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - - CompletableFuture generateCompletionResponseCompletableFuture = client.runCompletionMessage(completionParam); - RunCompletionMessageResponse generateCompletionResponse = generateCompletionResponseCompletableFuture.get(10, TimeUnit.SECONDS); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getText())); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getFinishReason())); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getRequestId())); - } -} -``` - -Python - -``` -import asyncio - -from alibabacloud_contactcenterai20240603.client import Client - -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import (RunCompletionRequestDialogueSentences, - RunCompletionRequestDialogue, - RunCompletionRequestFields, - RunCompletionRequest) -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages - -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" - -async def run_async_sse(): - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - - client = Client(config) - - # Prompt - role = "system" - content = "You are a helpful assistant." - requestMessage1 = RunCompletionMessageRequestMessages(content, role) - role = "user" - content = "请阅读以下对话内容，按照要求执行指令任务。\n```\n\n客服:你自己的话\n客户:你自己要号手机\n客户:31\n客服:他是1公分\n客户:那年后，我看\n客户:后来点开看\n客户:嗯，要不是以前的\n客户:这不看你看\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\n客服:你在这跟到后面来\n客服:这里不是大多少天都会员了吗\n客户:那你不是大于多少天都结了吗\n客服:嗯\n客户:嗯\n客服:我们有1个\n客户:就是你这个\n客服:他说你这里面也按照\n客户:他有多少是蒙细别的吗\n客户:这3天之后\n客服:当前执行的\n客服:这些，其实你这个也是\n客户:那你这个颜色\n客服:按照他给了预期了，是不是结多少天\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\n客服:那是什么问题\n客服:你3天的时候关内\n客户:你要是原来那个\n客户:你基本上就那你要看\n客服:你要是原来的，你，你基本上都要不要\n客服:5天\n客户:对的\n客户:啊，关于\n客户:15新啊\n客户:上面录的\n\n```\n任务指令如下：\n```\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\n```\n请依据上述指令，确保准确无误地完成任务。" - requestMessage2 = RunCompletionMessageRequestMessages(content, role) - - listRequestMessage = [requestMessage1, requestMessage2] - - request = RunCompletionMessageRequest() - request.messages = listRequestMessage - request.model_code = "tyxmTurbo" - request.stream = False - - # 发送请求 - response = await client.run_completion_message_async(workSpace, appId, request) - body = response.body - print(body) - -if __name__ == '__main__': - asyncio.run(run_async_sse()) -``` - -## 同步非流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - public static void main(String[] args) throws Exception{ - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - //Prompt - RunCompletionMessageRequest request = new RunCompletionMessageRequest(); - List messageList = new ArrayList<>(); - RunCompletionMessageRequest.RunCompletionMessageRequestMessages message1 = new RunCompletionMessageRequest.RunCompletionMessageRequestMessages(); - message1.setRole("system").setContent("You are a helpful assistant."); - RunCompletionMessageRequest.RunCompletionMessageRequestMessages message2 = new RunCompletionMessageRequest.RunCompletionMessageRequestMessages(); - message2.setRole("user").setContent("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话\\n客户:你自己要号手机\\n客户:31\\n客服:他是1公分\\n客户:那年后，我看\\n客户:后来点开看\\n客户:嗯，要不是以前的\\n客户:这不看你看\\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\\n客服:你在这跟到后面来\\n客服:这里不是大多少天都会员了吗\\n客户:那你不是大于多少天都结了吗\\n客服:嗯\\n客户:嗯\\n客服:我们有1个\\n客户:就是你这个\\n客服:他说你这里面也按照\\n客户:他有多少是蒙细别的吗\\n客户:这3天之后\\n客服:当前执行的\\n客服:这些，其实你这个也是\\n客户:那你这个颜色\\n客服:按照他给了预期了，是不是结多少天\\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\\n客服:那是什么问题\\n客服:你3天的时候关内\\n客户:你要是原来那个\\n客户:你基本上就那你要看\\n客服:你要是原来的，你，你基本上都要不要\\n客服:5天\\n客户:对的\\n客户:啊，关于\\n客户:15新啊\\n客户:上面录的\\n\\n```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。"); - messageList.add(message1); - messageList.add(message2); - - request.setMessages(messageList); - request.setStream(false); - request.setModelCode("tyxmTurbo"); - - RunCompletionMessageResponse responseBody = client.runCompletionMessage(workspaceId, appId, request); - RunCompletionMessageResponseBody body = responseBody.getBody(); - System.out.println(JSON.toJSONString(body)); - } -} -``` - -Python - -``` -from alibabacloud_contactcenterai20240603.client import Client - -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages - -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" - -if __name__ == '__main__': - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - - client = Client(config) - - # Prompt - role = "system" - content = "You are a helpful assistant." - requestMessage1 = RunCompletionMessageRequestMessages(content, role) - role = "user" - content = "请阅读以下对话内容，按照要求执行指令任务。\n```\n\n客服:你自己的话\n客户:你自己要号手机\n客户:31\n客服:他是1公分\n客户:那年后，我看\n客户:后来点开看\n客户:嗯，要不是以前的\n客户:这不看你看\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\n客服:你在这跟到后面来\n客服:这里不是大多少天都会员了吗\n客户:那你不是大于多少天都结了吗\n客服:嗯\n客户:嗯\n客服:我们有1个\n客户:就是你这个\n客服:他说你这里面也按照\n客户:他有多少是蒙细别的吗\n客户:这3天之后\n客服:当前执行的\n客服:这些，其实你这个也是\n客户:那你这个颜色\n客服:按照他给了预期了，是不是结多少天\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\n客服:那是什么问题\n客服:你3天的时候关内\n客户:你要是原来那个\n客户:你基本上就那你要看\n客服:你要是原来的，你，你基本上都要不要\n客服:5天\n客户:对的\n客户:啊，关于\n客户:15新啊\n客户:上面录的\n\n```\n任务指令如下：\n```\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\n```\n请依据上述指令，确保准确无误地完成任务。" - requestMessage2 = RunCompletionMessageRequestMessages(content, role) - - listRequestMessage = [requestMessage1, requestMessage2] - - request = RunCompletionMessageRequest() - request.messages = listRequestMessage - request.model_code = "tyxmTurbo" - request.stream = False - - # 发送请求 - response = client.run_completion_message(workSpace, appId, request) - body = response.body - print(body) -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md deleted file mode 100644 index 9fa17554..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md +++ /dev/null @@ -1,404 +0,0 @@ -# 通过模板ID调用通义晓蜜CCAI-对话分析AIO应用 - -## **前提条件** - -- 本文向您介绍通义晓蜜CCAI-对话分析AIO应用SDK的安装、使用及注意事项。 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -- 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## **接口入参位置** - -### **workspaceId** - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -1. 访问**应用广场**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -### **templateIds** - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 点击**管理**进入对应的应用卡片。 - -4. 点击**自定义指令**模板，切换为**专业构建模式**。 - -5. 点击右上方**指令模板管理**。 - -6. **自定义模板**列表中**模板ID**为入参中templateIds。 - -7. 如果还未创建自定义模板，请直接点击右上角**保存指令模板**按钮。 - - -## **安装SDK** - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## Python - -pip install alibabacloud\_contactcenterai20240603 - -## 异步流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - //对话内容 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.Sentences sentenceDto1 = RunCompletionRequest.Sentences.builder().role("user").text("我要办理信用卡").build(); - RunCompletionRequest.Sentences sentenceDto2 = RunCompletionRequest.Sentences.builder().role("agent").text("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息").build(); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - RunCompletionRequest.Dialogue dialogue = RunCompletionRequest.Dialogue.builder().sessionId("session_01_asdfasdfasd") - .sentences(sentenceDTOList).build(); - //属性信息 - List fieldList = new ArrayList<>(); - RunCompletionRequest.Fields field1 = RunCompletionRequest.Fields.builder().name("姓名").desc("用户的名称").build(); - RunCompletionRequest.Fields field2 = RunCompletionRequest.Fields.builder().name("信用卡号").desc("用户的信用卡号").build(); - fieldList.add(field1); - fieldList.add(field2); - //构建请求参数 - RunCompletionRequest completionParam = RunCompletionRequest.builder() - .workspaceId(workspaceId).appId(appId).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).modelCode("tyxmTurbo").dialogue(dialogue).fields(fieldList).templateIds(Arrays.asList(templateId)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - //发送请求 - ResponseIterable x = client.runCompletionWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - RunCompletionResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## **异步非流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - //对话内容 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.Sentences sentenceDto1 = RunCompletionRequest.Sentences.builder().role("user").text("我要办理信用卡").build(); - RunCompletionRequest.Sentences sentenceDto2 = RunCompletionRequest.Sentences.builder().role("agent").text("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息").build(); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - RunCompletionRequest.Dialogue dialogue = RunCompletionRequest.Dialogue.builder().sessionId("session_01_asdfasdfasd") - .sentences(sentenceDTOList).build(); - //属性信息 - List fieldList = new ArrayList<>(); - RunCompletionRequest.Fields field1 = RunCompletionRequest.Fields.builder().name("姓名").desc("用户的名称").build(); - RunCompletionRequest.Fields field2 = RunCompletionRequest.Fields.builder().name("信用卡号").desc("用户的信用卡号").build(); - fieldList.add(field1); - fieldList.add(field2); - //构建请求参数 - RunCompletionRequest completionParam = RunCompletionRequest.builder() - .workspaceId(workspaceId).appId(appId).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).dialogue(dialogue).fields(fieldList).templateIds(Arrays.asList(templateId)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - //发送请求 - CompletableFuture x = client.runCompletion(completionParam); - RunCompletionResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getText()); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - } -} -``` - -Python - -``` -import asyncio -from alibabacloud_contactcenterai20240603.client import Client -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest, \ - RunCompletionRequestDialogueSentences, RunCompletionRequestDialogue, RunCompletionRequestFields, \ - RunCompletionRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" -templateId = "YOUR_TEMPLATE" -async def run_async(): - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - client = Client(config) - # 对话 - sentence1 = RunCompletionRequestDialogueSentences("chat01", "user", "我要办理信用卡") - sentence2 = RunCompletionRequestDialogueSentences("chat02", "agent", - "好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息") - sentenceList = [sentence1, sentence2] - dialogue = RunCompletionRequestDialogue(sentenceList, "session_01_asdfasdfasd") - # 属性填充 - fields1 = RunCompletionRequestFields("", "用户的名称", None, "姓名") - fields2 = RunCompletionRequestFields("", "用户的信用卡号", None, "信用卡号") - fieldsList = [fields1, fields2] - # 构建请求参数 - templateIds = [templateId] - request = RunCompletionRequest() - request.dialogue = dialogue - request.fields = fieldsList - request.model_code = "tyxmTurbo" - request.stream = False - request.template_ids = templateIds - response = await client.run_completion_async(workSpace, appId, request) - body = response.body - print(body) -if __name__ == '__main__': - asyncio.run(run_async()) -``` - -## 同步非流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - public static void main(String[] args) throws Exception{ - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - Client client = new Client(config); - RunCompletionRequest request = new RunCompletionRequest(); - //对话信息 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.RunCompletionRequestDialogueSentences sentenceDto1 = new RunCompletionRequest.RunCompletionRequestDialogueSentences(); - sentenceDto1.setRole("user").setText("我要办理信用卡"); - RunCompletionRequest.RunCompletionRequestDialogueSentences sentenceDto2 = new RunCompletionRequest.RunCompletionRequestDialogueSentences(); - sentenceDto2.setRole("agent").setText("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息"); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - //属性填充 - List fieldList = new ArrayList<>(); - RunCompletionRequest.RunCompletionRequestFields field1 = new RunCompletionRequest.RunCompletionRequestFields(); - field1.setName("姓名").setDesc("用户的名称"); - RunCompletionRequest.RunCompletionRequestFields field2 = new RunCompletionRequest.RunCompletionRequestFields(); - field2.setName("信用卡号").setDesc("用户的信用卡号"); - fieldList.add(field1); - fieldList.add(field2); - RunCompletionRequest.RunCompletionRequestDialogue dialogue = new RunCompletionRequest.RunCompletionRequestDialogue(); - dialogue.setSessionId("session_01_asdfasdfasd").setSentences(sentenceDTOList); - //构建请求参数 - request.setDialogue(dialogue).setStream(false).setModelCode("tyxmTurbo").setFields(fieldList).setTemplateIds(Arrays.asList(templateId)); - RunCompletionResponse runCompletionResponse = client.runCompletion(workspaceId, appId, request); - RunCompletionResponseBody responseBody = runCompletionResponse.getBody(); - System.out.println(JSON.toJSONString(responseBody)); - } -} -``` - -Python - -``` -from alibabacloud_contactcenterai20240603.client import Client -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest, \ - RunCompletionRequestDialogueSentences, RunCompletionRequestDialogue, RunCompletionRequestFields, \ - RunCompletionRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" -templateId = "YOUR_TEMPLATE" -if __name__ == '__main__': - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - client = Client(config) - # 对话 - sentence1 = RunCompletionRequestDialogueSentences("chat01", "user", "我要办理信用卡") - sentence2 = RunCompletionRequestDialogueSentences("chat02", "agent", "好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息") - sentenceList = [sentence1, sentence2] - dialogue = RunCompletionRequestDialogue(sentenceList, "session_01_asdfasdfasd") - # 属性填充 - fields1 = RunCompletionRequestFields("", "用户的名称", None, "姓名") - fields2 = RunCompletionRequestFields("", "用户的信用卡号", None, "信用卡号") - fieldsList = [fields1, fields2] - # 构建请求参数 - templateIds = [templateId] - request = RunCompletionRequest() - request.dialogue = dialogue - request.fields = fieldsList - request.model_code = "tyxmTurbo" - request.stream = False - request.template_ids = templateIds - response = client.run_completion(workSpace, appId, request) - body = response.body - print(body) -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md deleted file mode 100644 index 3f99a056..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md +++ /dev/null @@ -1,67 +0,0 @@ -# 产品概述 - -本文档介绍通义晓蜜CCAI-对话分析AIO产品能力及产品优势。 - -## **什么是通义晓蜜CCAI-对话分析AIO** - -对话分析AIO，即对话分析all-in-one API，是基于深度调优的对话大模型， 为营销服类产品提供智能化升级所需的生成式摘要总结、质检、分析等能力的官方应用。 - -- **面向对象：**开发者、自研企业、传统呼叫中心采购者及友商等。 - -- **核心能力：**为客户提供多种规格专属模型（支持切换），并提供多指令精准执行、自带最佳实践的API及贴合业务场景的被集成方式。 - -- **服务形式：**通过API服务输出给客户，方便客户进行集成和使用官方预置模板或自定义模板，客户自定义前端样式。 - -- **核心价值：**免去企业SFT、探索prompt 、应用最佳实践开发的成本，使企业专注在业务逻辑的实现。 - -- **付费模式：**按调用次数后付费。 - - -## **产品能力** - -1. **多指令精准执行** - - -- **总结摘要：**根据对话内容，记录通话中最核心的信息。 - -- **信息抽取：**根据指令，抽取并组织关键信息。 - -- **质检分析：**按照指令进行检测，例如情绪、敏感词等 - -- **多指令任务：**生成标题、生成关键词、生成摘要、维度检测、信息抽取等。 - - -2. **多模型选择** - - -- 模型配置：平台提供通义晓蜜-Turbo、通义晓蜜-Plus两种模型规格。 - - -3. **多指令模板** - - -- **指令模板：**平台提供官方预置常用指令模板并支持管理自定义模板。 - - -4. **调试信息** - - -- SaaS管理：提供SaaS调试窗，用于验证效果并最终提供API服务。 - -- API服务：通过API服务输出给客户 - - -**5.数据报表** - -- 调用数据：提供按时、按日调用数据报表。 - -- 账单数据：请前往[费用与成本](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)查询。 - - -## **产品优势** - -- 结合大模型特性打造新能力，将过去多个业务场景凝聚成一个个通用、轻量、易集成的能力。 - -- 企业对于一次服务记录/工单所需要的所有质检、分析、处理的动作，可通过一次调用完成，得到结构化的结果。 - -- 提供通话工单分类（支持级联分类）、自定义信息抽取、自定义质检规则并检出能力。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md deleted file mode 100644 index 8b9c7fd7..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md +++ /dev/null @@ -1,38 +0,0 @@ -# 如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量 - -本文档介绍如何对应用进行编辑、删除等管理，如何进行API调用、如何查看调用量。 - -## **我的应用** - -路径：[应用广场](https://bailian.console.aliyun.com/#/app-market)→ 点击[应用实践](https://bailian.console.aliyun.com/?tab=app#/app-market/lightApplication)→ 找到**通义晓蜜CCAI-对话分析AIO**，选择[查看详情](https://bailian.console.aliyun.com/?tab=app#/app/app-market/ccai)，即可进入**我的应用** **。** 在**应用广场**页面，找到**通义晓蜜CCAI-对话分析AIO**应用卡片，单击**立即查看**。 - -### 3.1 应用复制、删除、修改 - -- **应用复制：**在**我的应用**中，可以点击应用右上角选择复制应用，对该应用进行复制。应用复制：在**我的应用**页面，单击目标应用卡片右上角的菜单图标，选择**复制应用**，即可复制该应用。应用复制成功后，在**我的应用**列表中可看到新增的复制应用卡片，其名称自动追加"副本"后缀。每张应用卡片底部提供**管理**、**调用量**、**API调用**三个操作入口。 - -- **应用修改：**同时可以点击应用右上角选择修改应用，对该应用名称进行修改。应用修改：在**我的应用**中，点击应用右上角的三点菜单，选择**修改应用**，即可对该应用信息进行修改。应用修改：可以在**编辑应用名称**对话框中修改**应用名称**，名称最多支持50个字符，修改完成后单击**确定**保存。 - -- **应用删除：**如果需要删除该应用，可以点击应用的右上角选择删除应用，对该应用进行删除，弹出“确认删除”二次确认框，选择确认删除，将成功删除。应用删除：同样可以点击应用右上角选择**删除应用**，对该应用进行删除。应用删除：点击应用右上角选择删除应用，系统弹出**确认删除**对话框，提示"删除已发布应用可能会对您的线上业务产生影响，您确定删除吗？"，单击**确认删除**完成删除操作，单击**取消**可取消操作。 - - -### 3.2 API调用 - -- 应用API：应用提供标准化API接口，用于向客户系统输出对话分析结果。支持集成官方预置Agent模板或用户自定义模板，客户可通过API获取结构化响应，并自主设计前端展示样式。 - - - 当通过**对话分析Agent创建方式**创建应用时，使用自定义指令的API调用： - - 在**对话分析Agent**的自定义指令页面右上角，单击**API示例**按钮，页面右侧展开API调用信息面板，依次包含：获取AccessKeyID和AccessKeySecret、获取Workspace ID和App ID、安装SDK（支持Java、Python、Go，Java通过Maven dependency引入）以及代码示例（支持Java、Curl、Python、Go）。代码示例中展示了`CcAiPassSync`类的调用逻辑，包括设置`workspaceId`与`appId`常量、构造`RunCompletionRequest`请求及组装对话内容。左侧**指令信息**区域可引入`${dialogue}`和`${fields}`两个变量，**模型配置**选择所需模型后，单击底部**测试（command + enter）**按钮即可在中间**效果测试**区域查看调用结果。 - - - 通过**自定义创建方式**的API调用：单击页面顶部的**API示例**按钮，右侧面板展示API调用所需信息。单击**获取API-KEY、APP-ID和Workspace ID**蓝色链接获取AccessKey ID、AccessKey Secret、Workspace ID和App ID。安装SDK时，在Maven项目的pom.xml中添加依赖`com.aliyun:alibabacloud-ccai-passthrough-sync`（版本3.3.3）。代码示例使用Java类`CcAiPassSync`，设置`workspaceId`和`appId`参数，构造`RunCompletionRequest`对象，传入包含user和agent角色的对话句子列表进行调用。 - - -### **3.3 调用量** - -- 调用次数按输入和输出token总数计量，以2000 tokens 为一个计量单位，输入与输出总token数小于等于2000 tokens为1次调用，大于2000小于等于4000 tokens 为2次调用，向上取整，以此类推。 - -- 调用次数按小时进行计量上报，查询当天时，折线图展示每小时调用量曲线。 - -- 查询某个日期区间数据时，折线图展示按天调用量曲线。 - - -在**调用量**页签中，可通过**所属空间**、**选择模型**、**应用**、**数据来源**下拉筛选条件，以及时间范围（**昨天**、**近3天**、**近七天**、**近15天**或自定义日期范围）查看对应的调用次数统计折线图。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md deleted file mode 100644 index b33e5bc8..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md +++ /dev/null @@ -1,60 +0,0 @@ -# 如何基于自定义方式创建应用 - -本文档介绍了如何通过自定义方式创建应用 - -## **创建应用** - -- 第一步：首先点击**我的应用**，再点击**创建应用**。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935525.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入应用调试界面。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用创建方式：** - - - 这里选择**自定义创建**：通过编写自定义指令（Prompt）来构建应用逻辑，用户可使用内置或自定义指令模板进行测试，适合有一定经验的人员。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935528.png) - - -## **应用配置** - -进入已经创建完成的应用中进行配置。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009398.png) - -- **模型配置：**通义晓蜜Plus 和 Turbo 模型仅支持语音与文本分析；当分析对象为图片时，系统将默认使用通义晓蜜VL模型，该模型不可手动选择。 - -- **指令信息：**通过编写指令信息来配置对应的任务、格式、要求等，来完成对应分析任务。 - - - **变量配置：**若需要在对话过程中引用更多变量可以在此配置，在指令编辑器中输入 `/` 可触发变量自动补全，选择后插入对应变量引用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009509.png) - - - **选择指令模板：**同时可以选择直接使用官方预置模板，当前线上提供了总结摘要、信息抽取、服务质检、标签分类、多指令任务，共五类模板。同时支持自定义指令模板，或在官方预置模板基础上自定义修改的指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009498.png) - - - **保存指令模板：**在编写指令信息或者自定义修改系统行业示例后，可以点击“保存指令模板”按钮，进行保存，选择指令模板保存方式，可选【新增指令模板、覆盖已有指令模板】，在“指令模板管理”中可以查看。选择‘覆盖已有指令模板’，需从下拉列表中选择目标模板名称。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009505.png) - - - **指令优化：**对编写完成的指令信息进行AI优化。![指令优化示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009511.png) - -- **分析对象类型** - - 分析对象类型可以分为三种，纯文本、语音、图片，同时支持添加知识库进行辅助分析，支持添加热词组有助于提升语音转译准确性。 - - - **知识库：**开启后可添加文档、表格、图片等类型知识用于辅助分析，当分析对象类型选择为图片时无法使用知识库。具体介绍可参考文档：[知识库的使用](https://help.aliyun.com/zh/model-studio/using-the-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3624933771/p1059909.png) - - - **选择文本时**：需要按照以下格式编写对话信息，同时也可以通过使用已经提供的行业对话示例。 - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006434.png) - - - **选择语音时：**自定义上传一个不超过40MB、WAV、MP3格式的文件，可以选择添加/新建热词组，提升语音转译效果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006437.png) - - 上传完成后将自动识别语音内容，并可以设置客户/客服先发言顺序。 - - - **选择图片识别后**：可点击上传一张不超过10MB、JPEG/JPG/PNG等常见图片格式，上传成功后可以通过指令信息对图片进行检测分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006443.png) - -- **字段信息：**当需要获取到对话内容的字段信息时，可以使用“信息抽取预置模板”、“多指令模板”创建指令任务，同时也需要引入变量${fields}填写字段信息。 - - 填写格式为：字段名：字段描述。![字段信息示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850538.png) - -- 点击**“测试”**按钮**，**查看测试结果![测试结果示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009528.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md deleted file mode 100644 index 85bc7021..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md +++ /dev/null @@ -1,140 +0,0 @@ -# 如何进行基于对话分析Agent方式创建应用 - -本文档介绍了如何通过对话分析Agent方式创建应用 - -## **创建应用** - -- 第一步：首先点击**我的应用**按钮，再点击**创建应用**按钮；![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935525.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入调试窗口。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用创建方式：** - - - 这里选择**基于对话分析Agent创建**：通过预置最佳实践示例或上传对话数据，体验大模型生成式摘要、总结、服务质检等全场景应用能力。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935528.png) - - -### 基于对话分析Agent创建 - -**说明** - -注意：当选择基于对话分析Agent创建方式，只有选择自**定义指令**\-**专业构建模式**可以用图片分析，其他方式无法对图片进行分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006405.png) - -进入已经创建完成的应用中后，可以选择**对话分析Agent**、**自定义指令**方式。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914644.png) - -1. #### **选择对话分析Agent方式**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914645.png) - - **对话分析维度：**可以根据实际业务需求的需要通过对话分析Agent来选择对应的维度。 - - - **标准指令：**已预置标准prompt，可快速生成理想结果。可选值为【标题、摘要、关键词、Q&A、问题解决方案】，至少选择一个选项。 - - - **高级指令：**服务质检、标签分类等高级指令有一定的业务属性，该部分指令在示例通话中已预置标准prompt，若自行上传通话数据且对结果有一定要求，建议在预置指令模板→专业模式中编辑自定义指令进行调试。可选值为【服务质检、关键信息、标签分类】。 - - - **分析对象类型：**根据自己业务需要分析的数据类型选择【纯文本、语音】。当选择**纯文本**时，可以上传不超过15000字的文本内容，同时我们还提供了行业对话示例来进行测试；当选择**语音**时，支持单个不超过40MB的WAV或MP3格式文件，上传完成后会自动将其转译为文本信息内容。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9479940671/p1008496.png) - - **信息内容：**需要按照以下格式编写对话信息，可以通过使用已经提供的行业对话示例。 - - - 对话信息建议按如下格式填写： - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx - - - 可直接插入内置的行业对话示例文本。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914647.png) - - - **点击“测试”按钮，查看输出结果：**测试出来的结果生成的指令与我选择的对话分析维度相对应。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0986579371/p914651.png) - -2. #### **选择自定义模板方式** - - 可以选择简单构建模式&专业构建模式。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914696.png) - - - ##### **简单构建模式说明** ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914697.png) - - - **模型配置**：通义晓蜜-Plus、通义晓蜜-Turbo。 - - - **指令类型：**可以根据实际业务需求分析的全部维度，根据选择的标准指令与高级指令，在指令信息中编辑指令的prompt。 - - - 标准指令：可以选择【标题、摘要、关键词、Q&A、问题解决方案】，并在下方指令信息中展示出系统内置prompt，可以对其进行自定义修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850108.png) - - - 高级指令：可以选择【服务质检、关键信息、标签分类】，并在下面指令信息中展示对应配置，进行自定义修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057646.png) - - - **服务质检：**对通话中存在的客户情绪、敏感词、服务质量等内容进行检测。可根据业务需要质检的场景和质检项。 - - - **质检项（名称+描述）：**设置服务质检中需要质检的名称与描述。 - - - **关键信息（名称+类型）：**提取通话中配置的关键信息，通过名称与类型配置。 - - - **标签分类（名称+描述）：**对通话内容进行分类定义，标签长度不超过10个字。 - - - **分析对象类型：**需要按照以下格式编写对话信息，可选择纯文本、语音两种方式。 - - - 对话信息建议按如下格式填写： - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx - - - 可以选择行业对话示例进行插入内置对话文本。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850123.png) - - - **点击“测试”按钮，查看测试结果。**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0986579371/p914710.png) - - - ##### 专业构建模式说明![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914716.png) - - **说明** - - 在模式切换时，当前模式所编辑的内容不再生效，对话内容将会重置。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057648.png) - - - **模型配置：**通义晓蜜-Plus、通义晓蜜-Turbo。 - - - **选择指令模板：**选择指令模板，可以选择直接使用官方预置模板，当前线上提供了总结摘要、信息抽取、服务质检、标签分类、多指令任务，共五类模板。同时支持自定义指令模板，或在官方预置模板基础上自定义修改的指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006412.png) - - - **保存指令模板：**在自定义编写指令信息或者自定义修改系统行业示例后，可以点击“保存指令模板”按钮，进行保存，选择指令模板保存方式，可选【新增指令模板、覆盖已有指令模板】，在“指令模板管理”中可以查看。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006414.png) - - - **指令模板管理：**可以查看预置模板和自定义保存的模板指令。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914728.png) - - - **变量配置：**除了内置${field}、${dialogue}两个变量以外，如需在分析过程中引用更多变量，可以完成变量配置。测试数据可以作为测试过程中的模拟数据，临时使用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939566.png) - - - 在编辑完成后，即可在指令信息中插入，使用“/”进行插入保存的变量。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939584.png) - - - **分析对象类型：**可以分为三种，纯文本、语音、图片，同时支持添加知识库进行辅助分析，支持添加热词组有助于提升语音转译准确性。 - - - **知识库：**开启后可添加文档、表格、图片等类型知识用于辅助分析，当分析对象类型选择为图片时无法使用知识库。具体介绍可参考文档：[知识库的使用](https://help.aliyun.com/zh/model-studio/using-the-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057649.png) - - - **选择文本时**：需要按照以下格式编写对话信息，同时也可以通过使用已经提供的行业对话示例。 - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006434.png) - - - **选择语音时：**自定义上传一个不超过40MB、WAV、MP3格式的文件，同时可以选择添加/新建热词组，提升语音转译效果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006437.png) - - 上传完成后将自动识别语音内容，并可以设置客户/客服先发言顺序。 - - - **选择图片识别后**：可点击上传一张不超过10MB、JPEG/JPG/PNG等常见图片格式，上传成功后可以通过指令信息对图片进行检测分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006443.png) - - - **字段信息：**当使用“信息抽取预置模板”、“多指令模板”创建指令任务，可引入变量${fields}填写字段信息。 - - 填写格式为：字段名：字段描述。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850538.png) - - - 点击**“测试”**按钮**，**查看输出结果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939588.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md deleted file mode 100644 index b3b43117..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md +++ /dev/null @@ -1,48 +0,0 @@ -# 热词组配置管理与使用 - -为提升语音转译的准确性，您可以在语音质检分析场景中使用热词组。本文档将介绍其配置与使用方法。 - -## **热词配置** - -热词组仅对离线/实时语音质检分析场景生效，用于提升语音转译的准确性。 - -### **1.热词组管理** - -- 进入热词组管理的路径： - - - 路径1：进入[**通义晓蜜CCAI-对话分析AIO**](https://bailian.console.aliyun.com/?tab=app#/app/app-market/ccai)后，点击我的应用，可在界面中看到**热词组管理**按钮。 - - - 路径2：通过进入具体应用的配置页面，点击**选择热词组**时进行添加对应的热词组。 - - **说明** - - 目前热词组支持通过自定义创建的应用；通过Agent创建的应用，需在‘专业构建模式’下才能找到并配置热词组。 - - 点击后右侧弹出 **选择热词组** 侧边栏，顶部提示"热词组可以提高语音转译准确性，使用时仅对语音质检分析生效"。可通过搜索框查找已有热词组，或单击 **新建热词组** 创建新的热词组，列表中勾选所需热词组即可完成关联。 - -- 在热词组管理界面，点击‘**新建热词组**’，填写名称和热词后，点击‘**确定**’完成创建。单击**新建热词组**按钮打开新建弹窗，在弹窗中填写热词组名称并通过**添加热词**逐条添加热词。 - - - 热词组名称：不超过10个字。 - - - 热词：每个热词组不超过128个热词。 - - - 热词：填入对应需要提高准确率的热词，重要限制：当前版本热词仅支持纯汉字，不支持英文、数字及特殊字符。包含非汉字的热词将无法生效。 - - - 权重：取值范围为1到5之间的整数，默认值：2；如果效果不明显可以适当增加权重，但是当权重较大时可能会引起负面效果，导致其他词语识别不准确。 - - - 操作：删除已添加的热词。 - -- 创建成功后，新的热词组将显示在列表中，并支持编辑或删除。在已创建的热词组名称右侧，单击齿轮图标，可在下拉菜单中选择**编辑**或**删除**该热词组。 - - -### **2.热词组在应用中的使用** - -进入具体应用的设置，选择并绑定所需的热词组。 - -**说明** - -每个应用仅支持绑定一个热词组。 - -1. 选择对应的热词组：单击**选择热词组**按钮，在弹出的浮层中单击目标热词组对应的**选用**按钮即可完成绑定。 - -2. 选择完成后可以查看绑定情况。在**自定义指令**配置页面下方的**热词组**区域，单击**选择热词组**按钮，将已创建的热词组添加到当前指令配置中。已添加的热词组会以标签形式展示，可单击删除图标移除。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md deleted file mode 100644 index c64ccbc5..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md +++ /dev/null @@ -1,74 +0,0 @@ -# 知识库的使用 - -采用检索增强生成（RAG）技术，根据用户上传的外部信息源检索相关信息，然后将检索到的内容整合到用户输入中，从而帮助大模型生成更准确的回答。 - -## **知识库** - -知识库仅对自定义指令的专业模式或者自定义创建的应用中的纯文本和语音对象分析场景生效，用于提升大模型生成回答的准确性。 - -### **1.知识库配置** - -1. **知识库配置的开启** - - 在自定义创建应用中开启**知识**开关后可以添加文档、表格、图片类型知识用于辅助分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059009.png) - -2. **知识库的调用方式** - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059025.png) - - 1. **必定调用：**每一轮的用户输入都会进入知识库检索与联网搜索，适用于高频知识问答场景。 - - 2. **智能调用：**智能判断用户的输入信息是否进入知识库检索或者联网搜索，适合灵活的对话场景。 - -3. **可使用的知识说明**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059033.png) - - 1. **文档知识：**使用知识库文件类数据构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - - 2. **表格：**使用数据中心表格类、数据库类构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - - 3. **图片：**使用数据中心图片类构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - -4. **知识库其他配置：** - - 1. **知识库过滤：**开启后，将通过配置判断prompt的方式引入大模型判断，对文档和表格召回结果进行二次智能过滤。 - - 2. **对话摘要总结：**开启后，可通过大模型对其对话内容进行概括总结。 - -5. **知识库的选择** - - 开启知识库后，可以在对应的文档、表格、图片知识类目中点击“+”按钮，右侧弹出对应知识库列表，可以选择已存的知识库进行的进行添加。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059059.png) - -6. **知识库的相似度阈值与权重** - - 选择对应的知识库后，支持配置相似度阈值与权重来控制筛选检索结果和知识库召回顺序。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059337.png) - - 1. **知识库的相似度阈值：** - - 阈值范围在0.01～1，默认为0.2，可以根据场景需要进行调整，注意：只有语义相似度得分高于此值的文本才会被召回。若此值设置得过高，将导致知识库丢弃所有相关的文本。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059377.png) - - 2. **知识库的权重：** - - 权重范围在0.5～2.默认为1，可以根据场景需要进行调整，在多路召回时，若多个知识库召回的文本切片相似度分数相同，系统将优先返回权重更高的知识库中的文本切片。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059385.png) - -7. **知识库的创建** - - 点击**创建新知识库**按钮，跳转到知识库创建的页面中，具体操作可参考[创建和使用知识库](https://help.aliyun.com/zh/model-studio/rag-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059090.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059088.png) - - **说明** - - 目前通义晓蜜CCAI-AIO产品所绑定的知识库仅支持文档搜索、数据查询、图片问答，不支持音视频搜索。 - - -### **2.知识库在应用中的使用** - -- 根据两种创建应用的方式区分： - - - 在**基于对话分析Agent创建方式**中的使用：仅支持在自定义指令中的**专业构建模式**中使用，**分析对象类型**需要选择纯文本、语音。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059324.png) - - - 在**自定义创建方式**中的使用：**分析对象类型**选择纯文本、语音即可使用。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md deleted file mode 100644 index 3e231b8b..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md +++ /dev/null @@ -1,85 +0,0 @@ -# 语音对话机器人操作指南 - -本文档介绍通义晓蜜CCAI-语音对话机器人在阿里云百炼控制台如何操作。 - -## **1\. 创建应用** - -- 路径：**[应用广场](https://bailian.console.aliyun.com/#/app-market)**\-应用实践-通义晓蜜CCAI-语音对话机器人-立即查看 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000691.png) - -- 第一步：首先点击**我的应用**按钮，再点击**创建应用**按钮；![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000694.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入调试窗口。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用描述：**自定义填入应用实际使用描述说明。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000698.png) - - -## **2.机器人配置** - -机器人配置目前使用的为prompt构建模式。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914777.png) - -- **模型选择：**通义晓蜜-Plus、通义晓蜜-Max、通义晓蜜-Turbo。 - -- **指令信息：**选择指令模板，可以选择直接使用官方预置模板，当前线上提供了通用场景、服务满意度调研、家电上门安装预约、游戏福利推送介绍四种模板。同时支持自定义指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000715.png) - -- **变量配置：**可在指令信息中通过**${xxx}**样式进行插入。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000733.png) - -- **指令配置：**如果当前参考的流程话术中存在特殊指令#\[...\]，请在回复中添加#\[...\]，目前支持传入挂机指令#\[HangUp\]。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000747.png) - -- **语音配置：**进行音色选择配置，配置完成后可在机器人呼叫时运用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000739.png) - - - 音色模板：可选择大模型音色，如：龙小夏V2、龙小夏等。 - - - 音量：范围为0～100，值越大声音越响亮。 - - - 语速：范围为-500～500，值越大语速越快。 - - - 音调：范围为-500～500，值越大音调越高昂。 - -- **高级配置：**对机器人的其他能力进行配置。 - - - 静默超时：自定义配置时长，范围在1～60秒，当对话过程中用户回复超过配置的静默时长后播报静默话术。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000740.png) - -- **点击“呼叫”按钮，查看测试结果。** - - **说明** - - 目前通过控制台测试时不支持变量传入。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000746.png) - -- **发布机器人：**点击**发布**按钮，当页面提出**发布成功**，即表示为成功发布。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000757.png) - - -## **3.我的应用** - -路径：**[应用广场](https://bailian.console.aliyun.com/#/app-market)**\-应用实践-通义晓蜜CCAI-语音对话机器人-立即查看。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000691.png) - -### **3.1 调用量** - -- 调用次数按小时进行计量上报，查询当天时，折线图展示每小时调用量曲线。 - -- 查询某个日期区间数据时，折线图展示按天调用量曲线。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000755.png) - -### 3.2 应用修改、删除 - -- **应用修改：**进入我的应用后，可以点击应用右上角选择修改应用，对该应用名称进行修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000750.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000751.png) - -- **删除应用：**进入我的应用后，可以点击应用的右上角选择删除应用，对该应用进行删除，弹出“确认删除”二次确认框，选择确认删除，将成功删除。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000752.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000753.png) - - -### **3.3API调用** - -应用API：通过API服务输出给客户，方便客户进行集成和使用官方预置模板或自定义模板，客户自定义前端样式 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000756.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md index 83fdd1da..f0939963 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md @@ -133,7 +133,7 @@ code string -状态码 +状态码：仅代表 http 请求状态 successful @@ -179,7 +179,7 @@ string 异常错误码 -success +ERR-00000000 errorMessage @@ -187,15 +187,22 @@ string 异常错误信息 -Deduct task already success,Please do not resubmit.token '369e8f2c-d283-424a-96c4-c83efe08c89e' +url download error: http://xx event string -事件类型 +事件类型：任务执行状态 + +- task-finished：任务结束（成功） + +- task-failed：任务结束（失败） + +- task-xxx：中间态，可以不关注 + -TIMEOUT\_CLOSE\_ORDER +task-finshed eventInfo @@ -203,7 +210,7 @@ string 事件描述 -xxx +url download error: http://xx sessionId @@ -1025,7 +1032,7 @@ taskStatus string -任务状态。 +任务状态：代表任务调度执行状态，如果任务状态 SUCCESSED，请查看 event 字段取真实业务状态。 - PENDING：待执行 @@ -1073,10 +1080,10 @@ Access was denied, message: No such namespace namespaces/mjp-test-default. "success": true, "data": { "header": { - "errorCode": "success", - "errorMessage": "Deduct task already success,Please do not resubmit.token '369e8f2c-d283-424a-96c4-c83efe08c89e'", - "event": "TIMEOUT_CLOSE_ORDER", - "eventInfo": "xxx", + "errorCode": "ERR-00000000", + "errorMessage": "url download error: http://xx", + "event": "task-finshed", + "eventInfo": "url download error: http://xx", "sessionId": "d5c38cf6-a4bf-4a57-a697-9f449926f0c9", "taskId": "6e223291-729b-4e84-9271-c13ada1a776b", "traceId": "215045f817272303448235204efdef" diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md index d03c071a..7301fe5b 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md @@ -10,7 +10,7 @@ - 开通阿里云百炼账号。具体操作，请参见[开通阿里云百炼](https://help.aliyun.com/zh/model-studio/first-api-call-to-qwen#67c76646c85x6)。 -- 开通函数计算（FC）。具体操作，请参见[什么是函数计算](https://help.aliyun.com/zh/functioncompute/fc/product-overview/what-is-function-compute)。 +- 开通函数计算（FC）。具体操作，请参见[什么是函数计算](https://help.aliyun.com/zh/functioncompute/what-is-function-compute)。 ## 操作步骤 @@ -19,7 +19,7 @@ ### **步骤一：创建函数（FC）** -1. 在[函数计算](https://fcnext.console.aliyun.com/cn-hangzhou/functions)控制台创建事件函数。具体操作，请参见[创建事件函数](https://help.aliyun.com/zh/functioncompute/fc/user-guide/creating-an-event-function)。 +1. 在[函数计算](https://fcnext.console.aliyun.com/cn-hangzhou/functions)控制台创建事件函数。具体操作，请参见[创建事件函数](https://help.aliyun.com/zh/functioncompute/creating-an-event-function)。 - 创建视频理解提交任务函数（建议函数名称：SubmitVideoAnalysisTask）。 @@ -49,7 +49,7 @@ pip3 install alibabacloud_endpoint_util alibabacloud_tea_openapi alibabacloud_quanmiaolightapp20240801 -t . ``` - - 方案二：以“层”的方式安装。相关文档，请参见[创建自定义层](https://help.aliyun.com/zh/functioncompute/fc/user-guide/create-a-custom-layer-1)。 + - 方案二：以“层”的方式安装。相关文档，请参见[创建自定义层](https://help.aliyun.com/zh/functioncompute/create-a-custom-layer-1)。 3. 您可自行优化调整代码中的入参：比如`prompt`模板、`modelId`等。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-cancelaudittask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-cancelaudittask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-cancelaudittask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-cancelaudittask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-queryaudittask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-queryaudittask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitaudittask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitaudittask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitaudittask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitaudittask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md index 8f161902..d2d7873d 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md @@ -462,24 +462,6 @@ API API概述 -[QueryAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-queryaudittask) - -查询审核结果 - -查询审核结果。 - -[SubmitAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-submitaudittask) - -提交审核任务 - -提交审核任务 - -[CancelAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-cancelaudittask) - -取消审核任务 - -取消审核任务 - [SubmitSmartAudit](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-submitsmartaudit) 提交智能审校任务 @@ -1607,3 +1589,21 @@ SubmitParseDocumentLayoutTask 获取排版任务结果 获取排版任务结果 + +[CancelAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-cancelaudittask) + +取消审核任务 + +取消审核任务 + +[QueryAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-queryaudittask) + +查询审核结果 + +查询审核结果。 + +[SubmitAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-submitaudittask) + +提交审核任务 + +提交审核任务 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-after-sales-service-scope.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-after-sales-service-scope.md new file mode 100644 index 00000000..54844525 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-after-sales-service-scope.md @@ -0,0 +1,57 @@ +# 阿里云百炼平台售后服务范围说明 + +## **阿里云百炼平台售后服务范围说明** + +欢迎您使用阿里云百炼。本《**阿里云百炼平台售后服务范围说明》是对您使用阿里云百炼相关产品和服务时适用的售后服务范围的说明。** + +1\. 在您购买的服务期限内，我们将为您提供如下售后基础服务，即通过官网、电话及阿里云APP提供7×24的电话咨询（95187、400电话）、智能在线和标准工单支持。支持范围包括： + +（1）关于阿里云百炼模型服务与产品功能、架构的咨询； + +（2）使用、配置阿里云百炼模型服务的最佳实践； + +（3）阿里云百炼模型服务的使用咨询、技术问题及故障诊断； + +（4）阿里云百炼API及阿里云百炼官方SDK问题的故障诊断； + +（5）与阿里云百炼管理控制台相关的问题； + +（6）与阿里云相关的账号问题咨询支持； + +（7）与阿里云相关的财务、合同及计费问题的咨询支持。 + +2\. 阿里云百炼将以阿里云官网页面公布的[客户服务权益](https://www.aliyun.com/service/customer-service-benefits?spm=5176.support-home.J_3451238410.1.12d1156fPBBxO0)标准向您提供相应的售后服务支持。 + +3\. 阿里云百炼同时提供付费版的售后增值服务（包括支持计划等），该等服务需在阿里云官网订购后生效使用。您还可通过阿里云获得其他付费的售后服务，具体详见阿里云的网站相关页面的收费售后服务内容。如您的项目需要更深度的技术支持（如业务代码编写指导、定制化集成方案等），建议联系阿里云商务经理沟通定制化服务方案。 + +4\. 为了方便您的生产或使用，如您选择将阿里云百炼服务与外部（非阿里云百炼平台上的）第三方工具或产品进行对接，阿里云将尽商业上合理的努力为您提供第三方工具在接入阿里云百炼模型推理服务过程中的方向性建议，但针对非阿里云百炼服务相关的问题，我们无法提供专业意见。 + +（1）我们提供建议的范围包括： + +（i）确认阿里云百炼API接口及服务端的可用状态； + +（ii）阿里云百炼官方API调用示例及SDK使用说明参考； + +（iii）协助核查阿里云百炼服务端调用明细和计费记录； + +（iv）基本的连通性测试建议（如通过curl等标准工具测试阿里云百炼服务地址的可达性）。 + +**（2）我们提供建议的范围不包括：** + +（i）第三方工具（如Cursor、Windsurf、Cline、OpenClaw等）的安装、部署、配置、升级及日常使用指导； + +（ii）第三方工具的产品功能、交互逻辑及内部实现问题的排查； + +（iii）其他云厂商、企业或社区提供的产品或服务的配置与运维； + +（iv）用户业务代码的编写、调试与实现； + +（v）用户本地环境（含内网、代理、VPN、防火墙、操作系统等）导致的连通性或兼容性问题的排查； + +（vi）第三方工具内部显示的Token数量、费用预估值或调用统计与阿里云计费数据之间的差异解释； + +（vii）所有第三方工具的安装、补丁更新、测试、故障诊断、优化等日常运维服务； + +（viii）基于阿里云百炼模型服务原生能力之上的第三方自建业务相关支持。 + +5\. 但请您注意，除我们另有书面说明外，第三方工具不构成我们的代理、受托或联合服务主体，**我们不对外部第三方工具的任何陈述、承诺或行为承担责任。**您知悉并确认，阿里云仅负责阿里云百炼平台自身的运营维护，即百炼服务端的技术架构、API接口、计量计费系统、控制台功能等；**第三方工具的运行维护（如AI编程工具的安装配置、开源代理框架的部署调优等）由您及相应工具提供方负责。**当您的问题出现在百炼模型服务的使用过程中，但其原因、责任范围或依赖关系已超出阿里云百炼平台本身可直接提供支持和保障的范围时——通常涉及您侧系统、外部第三方服务、网络环境、账号权限、业务流程或非标集成，阿里云将协助进行初步排查。若经排查确认问题来源于非阿里云侧，阿里云将给予方向性建议并引导您联系相应的服务主体。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-related-agreements.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-related-agreements.md index 941388e9..87bc4bba 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-related-agreements.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-related-agreements.md @@ -2,6 +2,6 @@ - [阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html?spm=5176.28197581.0.0.16e829a4HTC9FE) -- [阿里云百炼服务特别说明](https://help.aliyun.com/zh/model-studio/bailian-service-notes) +- [阿里云百炼体验功能特别说明](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20260716114753386/20260716114753386.html) - [开源模型协议条款说明](https://help.aliyun.com/zh/model-studio/open-source-model-terms) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md index 29f6ac10..36597cef 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md @@ -271,7 +271,45 @@ ### **2.2 模型调用费用** -在创建、更新、检索、命中测试知识库时，会调用向量模型（用于内容向量化）和排序模型（Rerank，用于重排序），这些调用会产生费用。 +在创建、更新、检索知识库以及使用知识问答服务时，会调用以下模型，这些调用会产生独立于规格费用之外的模型调用费用： + +**模型类别** + +**模型名称** + +**用途** + +**向量模型** + +[text-embedding-v4](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/text-embedding-v4)等 + +文档类知识库的文本向量化 + +[qwen3-vl-embedding](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-embedding) + +图片问答类、音视频搜索类知识库的多模态向量化 + +**排序模型** + +[qwen3-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-rerank) + +文档类知识库检索结果的二次排序（可选） + +[qwen3-vl-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-rerank) + +图片问答类、音视频搜索类知识库检索结果的二次排序（可选） + +**路由模型** + +[qwen-plus](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen-plus-latest) + +开启知识库路由时，系统调用 qwen-plus 判断查询应路由至哪些知识库 + +**问答模型** + +[qwen3.7-plus](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3.7-plus?serviceSite=asia-pacific-china) 等 + +知识问答服务中生成回答的大语言模型，由用户在应用中自行选择 **重要** @@ -281,20 +319,28 @@ **多个知识库计费规则：**阿里云百炼应用挂载了多个知识库时，会在多个知识库内执行检索，Token 消耗量（Query 向量化和 Rerank 排序）**按知识库数量倍数增加**（N 个知识库则消耗量 × N）。 -#### **2.2.1 创建/更新知识库** +#### **2.2.1 知识管理（创建与更新知识库）** - **调用场景：**上传新文件或增量更新时，调用向量模型对文本内容进行向量化处理。 - **计费说明：按新增内容的 Token 数量计费。**删除文件不产生模型调用费用。 +- **调用的模型：** + + - 文档搜索类知识库：[text-embedding-v4](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/text-embedding-v4) 或 [text-embedding-v3](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/text-embedding-v3)（文本向量模型）。 + + - 图片问答类、音视频搜索类知识库：[qwen3-vl-embedding](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-embedding)（多模态向量模型）。 + -#### **2.2.2 检索知识库** +#### **2.2.2 知识检索** - **调用场景** - 1. 调用向量模型，对用户的查询（Query）进行向量化。 + 1. **向量化：**调用向量模型，对用户的查询（Query）进行向量化。 + + 2. **知识库路由（可选）：**若应用关联了多个知识库并开启了知识库路由功能，系统会调用 [qwen-plus](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen-plus-latest) 判断用户查询应路由至哪些知识库，该调用按 qwen-plus 的 Token 用量计费。 - 2. 调用排序模型（Rerank），对初步检索到的结果进行重新排序，以提升最终答案的精准度。 + 3. **排序（可选）：**调用排序模型对初步检索到的结果进行重新排序，以提升最终答案的精准度。文档搜索类知识库使用 [qwen3-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-rerank)，图片问答类和音视频搜索类知识库使用 [qwen3-vl-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-rerank)。 - **计费说明** @@ -304,7 +350,7 @@ - **检索流程与计费关系详解** - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0647879771/CAEQaxiBgMCFrtjd3BkiIDA2ZWRiNzYxYzZiNzRkNGM5Mzg4NGQ5ZjhlODBlOWZj6139615_20260107153729.136.svg) + ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6004524871/CAEQaxiBgMCFrtjd3BkiIDA2ZWRiNzYxYzZiNzRkNGM5Mzg4NGQ5ZjhlODBlOWZj6139615_20260107153729.136.svg) 1. **初步召回** 系统根据以下参数从知识库中召回文本切片： @@ -328,7 +374,22 @@ Rerank 模型排序后，系统会根据**最终召回最大数量**参数（例如 5）返回相应数量的切片。 -#### **2.2.3 费用优化建议** +#### **2.2.3 知识问答** + +通过百炼应用（智能体应用、工作流应用）使用知识库进行问答时，除了检索阶段的模型费用外，还会产生以下模型调用费用： + +- **问答生成模型：**系统根据您在应用中选择的问答模型（如 qwen-plus 等）生成回答，按该模型的 Token 用量计费。具体价格以[模型计费标准](https://help.aliyun.com/zh/model-studio/model-pricing)为准。 + +- **预文件解析（可选）：**当用户在对话中上传文件并开启预文件解析功能时，系统会调用 [qwen3-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-rerank) 对文件内容进行排序处理，按排序模型的 Token 用量计费。 + +- **知识库路由（可选）：**若应用关联了多个知识库并开启了路由功能，系统会调用 [qwen-plus](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen-plus-latest) 进行路由判断（详见 [2.2.2 知识检索](#c868ef3a653qx)）。 + + +**重要** + +知识问答服务的完整费用 = **规格费用**（知识库运行时长）+ **检索阶段的模型费用**（向量化 + 排序 + 路由）+ **问答阶段的模型费用**（问答生成 + 预文件解析）。各模型费用按实际 Token 消耗量独立计算，请关注[模型计费标准](https://help.aliyun.com/zh/model-studio/model-pricing)了解各模型的单价。 + +#### **2.2.4 费用优化建议** 有以下两种方式： @@ -366,6 +427,19 @@ 点击知识库卡片上的**命中测试**，进入配置调试页面进行测试，会产生相应的模型（向量模型、排序模型）调用计费。 +#### **2.2.5 节省计划抵扣说明** + +知识库使用的向量模型（如 text-embedding-v4）和排序模型（如 qwen3-rerank）属于百炼平台 A 类模型，其调用费用支持通过以下节省计划抵扣： + +- **AI 通用型节省计划**（推荐）：覆盖 A 类全部模型（含文本向量、多模态向量、排序模型），按月承诺消费享阶梯折扣。详情请参见[节省计划与资源包](https://help.aliyun.com/zh/model-studio/savings-plan-and-resource-package)。 + +- **向量及排序模型节省计划**：专门针对向量和排序模型的节省计划，一次性购买固定金额。详情请参见[节省计划与资源包 > 向量及排序模型节省计划](https://help.aliyun.com/zh/model-studio/savings-plan-and-resource-package)。 + + +**说明** + +节省计划仅可抵扣模型调用费用，不可抵扣知识库的规格费用（运行时长费用）。规格费用的优化请参见[资源包](#a06c023507qq3)。 + ## 3\. 计费示例 ### 3.1 连续运行 1 天 @@ -396,7 +470,7 @@ ### 3.2 创建、更新与检索知识库 -基于 **text-embedding-v4**（向量模型）与 **qwen3-rerank**（排序模型），价格均为 **0.0005 元/千 Token**。 +以下示例基于文档搜索类知识库，使用 [**text-embedding-v4**](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/text-embedding-v4)（向量模型）与 [**qwen3-rerank**](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-rerank)（排序模型），价格均为 **0.0005 元/千 Token**。图片问答类和音视频搜索类知识库使用的多模态模型（[qwen3-vl-embedding](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-embedding)、[qwen3-vl-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-rerank)）价格请参见对应模型详情页。 **计费逻辑**：费用 = Token 消耗量（以“千 Token”为单位） × 模型单价 @@ -550,7 +624,7 @@ 5. **为什么我的排序（Rerank）费用特别高？如何降低模型调用费用？** - 排序（Rerank）模型的费用与您最终返回的结果数量无关，而是由**初步召回**的文本切片总数决定的。降低模型调用费用详见本文[2.2.3 费用优化建议](#9a9e30ecc3pbe)内容。 + 排序（Rerank）模型的费用与您最终返回的结果数量无关，而是由**初步召回**的文本切片总数决定的。降低模型调用费用详见本文[2.2.4 费用优化建议](#9a9e30ecc3pbe)内容。 6. **如何彻底停止知识库的计费？删除库内文件可以吗？** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md index 61db382a..b85e5222 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md @@ -117,6 +117,17 @@ aac、amr、flac、flv、m4a、mp3、mpeg、ogg、opus、wav、webm、wma、mp4 最大512MB +### **表格格式要求** + +上传 Excel 文件时，表格格式需满足以下要求，否则模型无法正确解析数据。 + +- **不支持合并单元格表头**：请使用标准单行表头格式，每列只有一个独立的字段名，不跨列合并。合并单元格会导致模型无法正确映射列关系，造成数据解析异常。 + +- **备注信息不要写在 Excel 首行**：Excel 文件的首行会被识别为表头或数据。请将备注和说明信息写在智能体应用的**系统提示词**中作为上下文，避免模型将说明文字与实际数据混淆。 + +- **格式修改后需重新上传**：修改表格格式后，需重新上传文件到知识库，并重新发布智能体应用，变更才能生效。 + + ### **上传操作** **类别** @@ -129,7 +140,7 @@ aac、amr、flac、flv、m4a、mp3、mpeg、ogg、opus、wav、webm、wma、mp4 在控制台单次操作可同时导入的文件数量。 -> 通过API进行批量导入时不受此限制，但建议单次导入不超过 10,000 个文件。 +> 通过API进行批量导入时不受此限制，但建议单次导入不超过 500 个文件。 50 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base.md index c5cdc98d..73ce26f5 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base.md @@ -62,7 +62,7 @@ 2. 填写**知识库名称**和**知识库描述**，其余设置保持默认，点击**下一步**。 -3. 选择**默认类目**，上传[阿里云百炼系列手机产品介绍.docx](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250603/duuuxk/%E9%98%BF%E9%87%8C%E4%BA%91%E7%99%BE%E7%82%BC%E7%B3%BB%E5%88%97%E6%89%8B%E6%9C%BA%E4%BA%A7%E5%93%81%E4%BB%8B%E7%BB%8D.docx)文件。点击**下一步**，然后点击**完成**。 +3. 选择**配置类目**，上传[阿里云百炼系列手机产品介绍.docx](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250603/duuuxk/%E9%98%BF%E9%87%8C%E4%BA%91%E7%99%BE%E7%82%BC%E7%B3%BB%E5%88%97%E6%89%8B%E6%9C%BA%E4%BA%A7%E5%93%81%E4%BB%8B%E7%BB%8D.docx)文件。点击**下一步**，然后点击**完成**。 ### 2\. 集成到业务应用 @@ -184,12 +184,10 @@ 2. **填写基础信息** - 根据应用场景选择合适的**知识库类型**（单一知识库不支持同时选择多个类型）。选择**文档搜索**类型后，还需选择**使用场景**（基础文档问答、图文并茂回复、视觉理解（富文本文档）或极速问答）： + 根据应用场景选择合适的**知识库类型**（单一知识库不支持同时选择多个类型）。选择**文档搜索**类型后，还需选择**使用场景**（基础文档问答、视觉理解（富文本文档）或极速问答）： - **基础文档问答：**适用于纯文本文档的语义检索。 - - **图文并茂回复：**适用于需要返回图文混排内容的场景。 - - **视觉理解（富文本文档）：**使用多模态向量模型对 PDF、图片等富文本文档进行视觉级理解和索引，保留原始版面信息。适合含有复杂排版、图表、公式的文档，支持文字、图片和图文组合三种命中测试模式。 - **极速问答：**针对检索速度进行优化，适合高度结构化或简单文档类型（如 FAQ、产品参数表等），提供极速低延时的问答体验。索引配置与基础文档问答一致，差异在于后端检索策略针对低延迟场景进行了专项优化。仅支持文本查询，不支持图片输入。 @@ -466,7 +464,7 @@ - **知识库类型限制：**仅适用于文档搜索类、数据查询类、音视频搜索类知识库。**图片问答类知识库不支持**。 - - **使用场景限制：**仅「基础文档问答」和「图文并茂回复」两种使用场景支持。「视觉理解（富文本文档）」和「极速问答」**不支持**。 + - **使用场景限制：**仅**基础文档问答**使用场景支持。**视觉理解（富文本文档）**和**极速问答**不支持。 ## **相似度阈值** @@ -534,7 +532,7 @@ > 文档搜索类知识库无法实现此效果。 - - 支持导入多个Excel文件，但要求各文件的**表结构完全一致**。 + - 支持导入单份xlsx、xls格式的文档，文件大小限制20MB以内。 - **选择连接器：**选择指定的数据连接器。如尚未创建数据连接器，请参阅[数据连接](https://help.aliyun.com/zh/model-studio/data-connection)。 @@ -609,7 +607,7 @@ - **知识库类型限制：**仅适用于文档搜索类、数据查询类、音视频搜索类知识库。**图片问答类知识库不支持**。 - - **使用场景限制：**仅「基础文档问答」和「图文并茂回复」两种使用场景支持。「视觉理解（富文本文档）」和「极速问答」**不支持**。 + - **使用场景限制：**仅**基础文档问答**使用场景支持。**视觉理解（富文本文档）**和**极速问答**不支持。 ### **相似度阈值** @@ -721,7 +719,7 @@ - **知识库类型限制：**仅适用于文档搜索类、数据查询类、音视频搜索类知识库。**图片问答类知识库不支持**。 - - **使用场景限制：**仅「基础文档问答」和「图文并茂回复」两种使用场景支持。「视觉理解（富文本文档）」和「极速问答」**不支持**。 + - **使用场景限制：**仅**基础文档问答**使用场景支持。**视觉理解（富文本文档）**和**极速问答**不支持。 ### **相似度阈值** @@ -865,7 +863,7 @@ - **知识库类型限制：**仅适用于文档搜索类、数据查询类、音视频搜索类知识库。**图片问答类知识库不支持**。 - - **使用场景限制：**仅「基础文档问答」和「图文并茂回复」两种使用场景支持。「视觉理解（富文本文档）」和「极速问答」**不支持**。 + - **使用场景限制：**仅**基础文档问答**使用场景支持。**视觉理解（富文本文档）**和**极速问答**不支持。 ### **相似度阈值** @@ -894,7 +892,7 @@ - **使用场景**可根据需求选择**基础文档问答**、**图文并茂回复**、**视觉理解（富文本文档）**或**极速问答**（适用于高度结构化或简单文档类型，任务明确，提供极低延迟的问答体验）。 + **使用场景**可根据需求选择**基础文档问答**、**视觉理解（富文本文档）**或**极速问答**（适用于高度结构化或简单文档类型，任务明确，提供极低延迟的问答体验）。 在请求高峰时段，创建过程可能需要数小时（取决于数据量），请耐心等待。 @@ -905,11 +903,15 @@ ## **文档搜索类知识库** -- **自动更新（推荐）** +1. **自动更新（推荐）** - 通过对象存储OSS管理文件，借助函数计算 FC 监听文件变更事件，自动同步更新至知识库，实现知识的实时更新。详见[告别手动操作，让AI知识库自动更新](https://www.aliyun.com/solution/tech-solution/auto-updated-knowledge-base)。 + 通过对象存储OSS类型连接器监听OSS内数据变动，自动同步更新至知识库，实现知识的实时更新。操作步骤如下： -- **手动更新** + 1. **创建连接器：**前往[创建连接器](https://bailian.console.aliyun.com/cn-beijing?tab=app#/connector/create)，创建**OSS**类型的连接器并选择进行监听的Bucket。 + + 2. **创建知识库：**创建一个知识库，在**选择数据**步骤的**数据来源**选项选择创建完成的OSS类型的连接器。 + +2. **手动更新** 在[知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base)页面，找到目标知识库，单击卡片上的**查看详情**。 @@ -1230,22 +1232,12 @@ **方式一（仅适用智能体应用）** - 1. 在[构建知识库](#c0fa1080aerzp)时，知识库类型选择**文档搜索**，使用场景选择**图文并茂回复**。 - - > 选择图文并茂回复后，知识库将从文件插图中提取摘要，大模型根据摘要与问题的相关性自主决定是否插入图片。 + 1. 在[构建知识库](#c0fa1080aerzp)时，知识库类型选择**文档搜索**，使用场景选择**视觉理解（富文本文档）**。 - **重要** - - 上传文档时不能选择**电子文档解析**，否则无法获取图片内容。电子文档解析不识别文档中的图片，会导致图文并茂回复功能无法正常使用。 + > 选择视觉理解（富文本文档）回复后，知识库将从文件插图中提取摘要，大模型根据摘要与问题的相关性自主决定是否插入图片。 2. 创建或编辑智能体应用时，选择**千问-Plus**或**千问-Plus-Latest**模型（经测试，两款模型效果最佳）。点击**文档知识库**右侧的**+**按钮，添加上一步构建的知识库。 - **说明** - - 召回长度须小于文档实际长度。若召回长度超过文档实际长度，系统将直接返回完整文档内容，不执行图文并茂的逻辑判断。 - - > 注意：当前"图文并茂回复"与"展示回答来源"功能暂不支持同时开启。 - 3. 实际问答效果： ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1699676371/p903021.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-optimization.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-optimization.md index 49fb108e..308e13dd 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-optimization.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-optimization.md @@ -107,6 +107,8 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 在创建知识库的**索引设置**步骤中，打开页面底部的**多轮对话改写**开关即可启用该功能。 + ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3562168571/p1008316.png) + 请注意，多轮对话改写功能是与知识库绑定的，开启后仅对当前知识库相关的查询生效。且如果您在创建知识库时未启用该配置，则后续无法再为该知识库开启，除非重新创建知识库。 @@ -129,7 +131,7 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 - **在数据管理页面编辑标签**：对于已上传的文件，可单击右侧的**标签**进行编辑，相关API是[UpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatefiletag)。 - 目前阿里云百炼支持以下方式使用标签： + 目前阿里云百炼支持以下两种方式使用标签： - [通过API调用阿里云百炼应用](https://help.aliyun.com/zh/model-studio/application-calling-guide#4100253b7chc3)时，可以在请求参数`tags`中指定标签。 @@ -141,42 +143,6 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 切换至**召回策略**页签，在**标签过滤**中选择需要应用的标签（如示例中的『硬件』『智能手机』）即可。 - - 在新版智能体中，通过系统提示词引导智能体按标签进行过滤。 - - 1. **为知识库文档设置标签**：在管理知识库时，为各文件添加标签。 - - 例如，百炼电脑的标签为`bailian_pc`，百炼系列手机产品介绍的标签为`bailian_mobile`，百炼电脑服务及优惠政策的标签为`bailian_service`和`bailian_pc`。 - - 2. **在系统提示词中引导智能体**：在智能体的系统提示词中，清晰地说明各标签的含义与使用规则。例如： - - ``` - 你是阿里云百炼产品的智能客服，请回答与阿里云百炼产品相关的问题。 - - 当问题比较宽泛时，检索整个知识库。 - 当问题比较明确时，使用知识库标签进行定向检索。标签映射关系如下： - 1. 与百炼手机相关：bailian_mobile - 2. 与百炼电脑相关：bailian_pc - 3. 与百炼产品相关：bailian_mobile 或 bailian_pc - 4. 与百炼电脑服务及优惠政策相关：bailian_service 和 bailian_pc - ``` - - 3. 智能体会识别用户意图，匹配对应的标签，并仅从带有这些标签的文件中检索内容。匹配逻辑包含以下几种模式： - - - **单标签匹配**：对于问题“百炼有哪些手机产品？”，智能体会检索所有带有`bailian_mobile`标签的文件。 - - - **多标签“或”逻辑**：对于问题“百炼有哪些产品？”，智能体会检索所有带有`bailian_mobile`或`bailian_pc`标签的文件。 - - - **多标签“与”逻辑**：对于问题“百炼电脑的服务政策是什么？”，智能体仅检索同时带有`bailian_pc`和`bailian_service`标签的文件。 - - - 您可以在知识库交互卡片中查看标签匹配逻辑，并通过调整提示词来优化匹配效果。匹配逻辑的格式如下： - - - **单标签匹配**：`[{"tags":["bailian_mobile"]}]` - - - **多标签“或”逻辑**：`[{"tags":["bailian_mobile","bailian_pc"]}]` - - - **多标签“与”逻辑：**`[{"tags":["bailian_service"]},{"tags":["bailian_pc"]}]` - 2. **典型问题****：**知识库中含多个结构内容相同/相近的文件，如文件A和文件B的内容都包含“功能概述”章节，只希望在A文件的“功能概述”中检索。 @@ -290,6 +256,23 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 此处只是修改了知识库中的文本切片，并未修改数据管理（临时存储）中的源文件或数据表。因此后续将此文件/数据表再次导入知识库时，仍需人工检查和修正。 +3. **必要时新增或删除单条切片：**除修正现有切片外，您还可以直接在切片层做更细粒度的管理，应对智能切分无法覆盖的边界情况。 + + **说明：新增切片与删除切片** + + 在知识库的切片视图中，可以对单条切片做以下两类操作： + + - **新增切片：**当原文档存在关键信息但被切片策略遗漏，或希望额外补充上下文时，可手动新增一条切片。新增时需选择所属文档，并填写切片内容（最多 6000 字符）；如该文档为图片问答类知识库，还可上传关联图片（每张图片不超过 20 MB，支持 PNG、JPG、JPEG、BMP、GIF 等常见格式）。 + + - **删除切片：**当某条切片包含错误信息或重复内容，且无法通过修正解决时，可单独删除该切片，不影响同文档其他切片继续被检索。 + + + 新增和删除均仅作用于当前知识库的检索结果，不会修改数据管理中的源文件。再次将源文件导入知识库时，需要重新进行此类调整。 + + **说明** + + 不同知识库类型支持的切片操作存在差异：**编辑切片（更新内容）所有类型的知识库均支持**，在切片详情弹窗右上角单击编辑即可操作。新增和删除方面，文档搜索类、数据查询类、图片问答类知识库均支持；**音视频搜索类知识库仅支持删除切片**。 + ### **3.4 重排不佳** @@ -338,7 +321,7 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 从下方示意图可以看到，目标知识库中实际与用户提示词相关，需要返回的文本切片总共有7个（下图左侧，已用绿色标出），但由于已经超出了当前设定的最大召回片段数K，因此包含优势5（超长待机）和优势6（拍照清晰）的文本切片被舍弃，没有提供给大模型。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6954804871/CAEQURiBgMCatMinohkiIDY3YTVhOWY2MjNjMjRkYzc5NTU1ZmVhNGQ2MGQ5ODc24762899_20250109142407.621.svg) + ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3188584871/CAEQURiBgMCatMinohkiIDY3YTVhOWY2MjNjMjRkYzc5NTU1ZmVhNGQ2MGQ5ODc24762899_20250109142407.621.svg) 由于 RAG 本身无法判断需要多少个文本切片才能给出“完整”的答案，因此即使最终提供的文本切片有遗漏，随后大模型仍然会基于缺失的文本切片生成不完整的回答。 @@ -555,6 +538,47 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 **enable\_thinking**：是否开启思考模式。 +## **3.6 知识检索与知识问答服务的效果优化** + +如果您使用知识检索或知识问答服务，以下策略可帮助提升检索和回答质量。 + +### **多知识库场景的优化** + +- **合理设置知识库权重**：当绑定多个知识库时，为核心知识库设置更高的权重，使其检索结果在混排中排序更靠前。 + +- **启用知识库路由**：绑定多个知识库但查询通常只涉及部分库时，启用知识库路由可让系统自动判断需要查询哪些库，减少无关结果的干扰。 + +- **选择合适的混排模型模式**： + + - **问答模式**：适用于用户输入为问题、期望从知识库中找到答案的场景。 + + - **相似模式**：适用于需要查找与输入内容语义相似的段落或文档的场景。 + + - **自定义高级模式**：当上述两种模式效果不理想时，可自定义排序指令来干预排序行为。 + + +### **知识库独立检索参数调优** + +在知识检索或知识问答的配置页中，点击知识库旁的![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6339082871/p1080483.png)图标可展开独立检索参数，针对不同知识库进行精细化调优： + +- **相似度阈值**：提高阈值可过滤低质量结果，但阈值过高可能导致召回不足。建议从 0.2 开始调整。 + +- **最大召回数量**：增加召回数量可获得更全面的结果，但可能引入噪音。根据知识库规模和内容相关度调整。 + +- **标签过滤**：利用文档标签缩小检索范围，提高结果的精准度。 + + +### **知识问答的生成效果优化** + +- **调整提示词**：通过提示词设定模型的角色、回答风格和关注领域。例如设定"你是一位专业的技术支持工程师，请基于知识库内容简洁回答"可显著提升回答质量。 + +- **选择检索模式**：简单问题使用**极速**模式降低延时；复杂或模糊问题使用**多轮智能**模式，让模型自动规划多轮检索。 + +- **设置知识库优先级**：开启优先级后，为更权威的知识库设置"高"优先级，确保其内容在回答中占主导地位。 + +- **启用拒答和防泄漏**：如果希望模型在知识库无相关内容时明确拒绝回答（而非编造答案），可开启拒答功能并配置触发条件。如果需要保护知识库原文不被直接输出，可开启防泄漏。 + + ## **4\. 后续步骤** ### **4.1 持续迭代** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/application-introduction.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/application-introduction.md index 0e578a2d..2a13317b 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/application-introduction.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/application-introduction.md @@ -45,12 +45,15 @@ Python 编码（专业代码） AI 自主决策、动态规划 由大模型根据提示词自主规划任务步骤 + 由预定义流程精确控制 每一步都由预设的节点定义，逻辑确定 + 完全由代码控制 所有逻辑和执行路径由代码定义 + 适合人群 @@ -74,9 +77,9 @@ AI 工程师、开发者 - **智能体应用** - - 如何创建和配置？请参考：[新版智能体应用](https://help.aliyun.com/zh/model-studio/new-single-agent-application)（推荐）、[智能体应用](https://help.aliyun.com/zh/model-studio/single-agent-application)。 + - 如何创建和配置？请参考：[新版智能体应用（Agent 2.0）](https://help.aliyun.com/zh/model-studio/new-single-agent-application)（推荐）、[智能体应用（Agent 1.0）](https://help.aliyun.com/zh/model-studio/single-agent-application)。 - - 如何快速上手实际案例？请参考：[创建智能问答 AI 电商客服助手](https://help.aliyun.com/zh/document_detail/2878136.html#0a9fbaf6a71q7)、[集成高德 MCP 的旅行规划智能体](https://help.aliyun.com/zh/document_detail/2880695.html)。 + - 如何快速上手实际案例？请参考：[创建智能问答 AI 电商客服助手](https://help.aliyun.com/zh/document_detail/2878136.html#0a9fbaf6a71q7)、[集成高德 MCP 的旅行规划智能体](https://help.aliyun.com/zh/model-studio/use-cases/integrate-amap-mcp-travel-planning-agent)。 - 如何通过 API 调用？请参考：[新版智能体应用 API](https://help.aliyun.com/zh/model-studio/new-agent-application-api-reference)、[调用智能体应用](https://help.aliyun.com/zh/model-studio/call-single-agent-application/)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/new-single-agent-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/new-single-agent-application.md index ede0fcdc..48466f2f 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/new-single-agent-application.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/new-single-agent-application.md @@ -1,4 +1,4 @@ -# 新版智能体应用（Agent 2.0） +# 新版智能体应用 新版智能体应用（Agent 2.0）将知识库、MCP 等多种能力统一为工具，并通过自主思考和规划来调用，以解决复杂任务。 @@ -53,7 +53,7 @@ - **最长回复长度**：模型生成的长度限制，不包含提示词。 - - **temperature**：控制生成随机性和多样性，数值越高随机性越强。 + - **温度系数**：控制生成随机性和多样性，数值越高随机性越强。 - **enable\_thinking**：是否开启思考模式。开启思考模式有助于提升智能体的反思效果。不支持思考模式的模型无法配置 enable\_thinking 参数。 @@ -245,3 +245,15 @@ - 意图与技能的相关性：请评估问题的表述是否清晰，其意图是否能明确指向特定技能。如果意图模糊或与技能功能不相关，模型可能选择不调用。 - 执行轮次限制：请检查是否达到了 ReAct 轮次上限。智能体可能已规划调用该技能，但在执行到该步骤前因轮次耗尽而被强制终止。 + + +### 智能体应用是否支持上下文缓存？ + +支持**隐式缓存**，暂不支持在应用中配置**显式缓存**。 + +- **隐式缓存**：智能体在调用支持隐式缓存的模型时会自动生效，无需任何配置、也无法关闭。系统会自动识别并缓存请求的公共前缀（如相同的系统提示词、多轮对话历史、知识库召回内容等），命中缓存的输入 Token 按标准输入单价的 20% 计费，可相应降低模型调用成本。 + +- **显式缓存**：需要在模型调用请求中主动为指定内容创建缓存标记。智能体应用由平台统一构造模型请求，暂不支持配置显式缓存。 + + +上下文缓存的工作模式、支持的模型及计费详情，请参见[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/rich-code-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/rich-code-application.md index 764cb376..0ae6fa78 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/rich-code-application.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/rich-code-application.md @@ -30,7 +30,7 @@ - **命令行创建**：在控制台创建空白应用后，通过 AgentScope-AI 命令行工具上传本地代码包部署。适合已有项目代码的开发者，详细流程请参考[API 开发指南](https://help.aliyun.com/zh/model-studio/rich-code-app-develop-guide)。 - 填写**应用名称**后，单击**确认**。 + 填写**应用名称**后，单击**立即创建**。 ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2506686771/p1070627.png) @@ -49,14 +49,14 @@ **提交代码** - - **使用模板代码**（默认）：选择一个预置的应用模板快速开始。 + - **使用模版代码**（默认）：选择一个预置的应用模版快速开始。 - **上传代码包**：上传本地开发的 .whl 格式代码包。代码包制作方式请参考[API 开发指南](https://help.aliyun.com/zh/model-studio/rich-code-app-develop-guide)。 - 平台提供以下应用模板： + 平台提供以下应用模版： - **模板** + **模版** **说明** @@ -76,7 +76,7 @@ 根据需要配置**规格方案**（vCPU、内存、磁盘大小）、**最小实例数**、**单实例并发度**和**部署地域**。初次体验保持默认即可。 - > 时延敏感业务建议最小实例数 ≥ 1，可实现毫秒级热启动、保障服务不中断。详见[实例类型和规格](https://help.aliyun.com/zh/functioncompute/fc/product-overview/instance-types-and-specifications#section-mfv-5fb-ehw)。 + > 时延敏感业务建议最小实例数 ≥ 1，可实现毫秒级热启动、保障服务不中断。详见[实例类型和规格](https://help.aliyun.com/zh/functioncompute/instance-types-and-specifications#section-mfv-5fb-ehw)。 配置完成后，单击**立即部署**开始部署。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/workflow-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/workflow-application.md index 6ef7ecfd..bb678f14 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/workflow-application.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/workflow-application.md @@ -1049,7 +1049,7 @@ 服务配置 - 选择要调用的函数计算服务。您需要先[创建函数](https://help.aliyun.com/zh/functioncompute/fc/function-instance-1/)计算服务。 + 选择要调用的函数计算服务。您需要先[创建函数](https://help.aliyun.com/zh/functioncompute/function-instance-1/)计算服务。 创建函数计算服务的账号需与当前登录的阿里云百炼平台账号一致，或隶属于同一阿里云主账号。 @@ -1059,7 +1059,7 @@ - **节点示例：** - 1. 在北京地域，参考[使用函数计算部署千问大模型实现AI对话](https://help.aliyun.com/zh/functioncompute/fc/use-cases/use-function-compute-to-realize-ai-dialogue)创建并测试函数计算服务，确保函数计算服务正常运行。 + 1. 在北京地域，参考[使用函数计算部署千问大模型实现AI对话](https://help.aliyun.com/zh/functioncompute/use-function-compute-to-realize-ai-dialogue)创建并测试函数计算服务，确保函数计算服务正常运行。 2. 配置函数计算节点。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-agent.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-agent.md index 4e69b1be..20b858f0 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-agent.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-agent.md @@ -1,11 +1,3 @@ # 构建 Agent 智能体是模型、系统提示词、工具、MCP 服务和技能的组合配置。创建后通过 ID 引用，可在多个会话中复用。 - -**[配置智能体](t6648215.xdita#)** 模型、提示词与字段配置 - -**[内置工具](t6679518.xdita#)** 命令执行、文件操作与网络访问 - -**[MCP 服务](t6674433.xdita#)** 接入外部工具服务 - -**[技能](t6645507.xdita#)** 预置工具组合与任务流程 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-context.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-context.md index b93e1eff..0f6c2c03 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-context.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-context.md @@ -20,8 +20,6 @@ - **创建会话时挂载**：在 `resources` 字段中指定资源列表和挂载路径。 -- **运行时追加**：通过 **POST** `/sessions/{session_id}/resources` 追加，实时生效，无需重启会话。 - ### 路径约定 @@ -30,11 +28,3 @@ ### 会话隔离 挂载资源时，平台做一份内部拷贝放入会话沙箱。会话内的修改不影响原始资源，也不影响挂载了同一资源的其他会话。卸载后会话内的副本被清理，原始资源不受影响。 - -## 下一步 - -- [文件上传与挂载](https://help.aliyun.com/zh/model-studio/managed-agents-file)：上传文件并挂载到会话。 - -- [发起会话](https://help.aliyun.com/zh/model-studio/managed-agents-session-event)：创建会话时指定挂载资源。 - -- [定义 Agent](https://help.aliyun.com/zh/model-studio/managed-agents-agent-definition)：配置系统提示词和工具。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-environment.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-environment.md index 6239b835..2ebfdc18 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-environment.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-environment.md @@ -1,9 +1,3 @@ # 配置 Agent 环境 运行环境定义工具调用的执行沙箱，独立于智能体管理，可被多个会话复用。 - -**[配置运行环境](t6679519.xdita#)** 沙箱类型、预装包与网络策略 - -**[凭证](t6674440.xdita#)** 管理外部服务的鉴权信息 - -**[文件](t6674422.xdita#)** 上传文件并挂载到会话 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-introduction.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-introduction.md index afd8e522..93365301 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-introduction.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-introduction.md @@ -93,14 +93,3 @@ Managed Agents 是百炼提供的智能体托管运行时，适用于多步工 - **有状态会话**：跨多轮交互保持上下文和文件系统状态。 - **快速开发 / 低运维开销：** 专注于Agent逻辑，而非从零开始构建和维护代理循环、沙箱编排或工具执行基础设施。 - - -## 下一步 - -**[快速开始](t6645503.xdita#)** 创建第一个智能体并发起会话 - -**[管理会话与事件](t6679527.xdita#)** 创建会话、发送消息与事件流 - -**[智能体](t6679515.xdita#)** 配置模型、提示词与工具 - -**[API 总览与认证](t6645508.xdita#)** 接口总览、认证与可用 API diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-quick-start.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-quick-start.md index cefa0544..122a5efe 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-quick-start.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-quick-start.md @@ -49,17 +49,7 @@ curl -X POST "https://{workspace_id}.cn-beijing.maas.aliyuncs.com/api/v1/agentst "name": "data-analyst", "model": {"id": "qwen3-max"}, "system": "你是数据分析专家，使用 pandas 处理 CSV 文件。", - "tools": [ - { - "type": "builtin_toolkit", - "default_config": {"enabled": true}, - "configs": [ - {"name": "bash", "enabled": true}, - {"name": "read", "enabled": true}, - {"name": "write", "enabled": true} - ] - } - ] + "tools": [{"type": "builtin_toolkit"}] }' ``` @@ -70,15 +60,7 @@ agent = client.agents.create( name="data-analyst", model="qwen3-max", system_prompt="你是数据分析专家，使用 pandas 处理 CSV 文件。", - tools=[ - {"type": "builtin_toolkit", - "default_config": {"enabled": True}, - "configs": [ - {"name": "bash", "enabled": True}, - {"name": "read", "enabled": True}, - {"name": "write", "enabled": True}, - ]} - ], + tools=[{"type": "builtin_toolkit"}], ) print(agent.id) # "agent_xxx" print(agent.version) # 1 @@ -303,13 +285,3 @@ try (AgentStudioEventStream stream = client.sessions().events().stream("sesn_xxx ``` 调试完成后，点击**返回智能体列表**进入管理页面。前三步创建的智能体、环境、会话均已保存，可随时在对应菜单中查看和编辑。 - -## 下一步 - -**[管理会话与事件](t6679527.xdita#)** 创建会话、发送消息与事件流 - -**[智能体](t6679515.xdita#)** 配置模型、提示词与工具 - -**[配置运行环境](t6679519.xdita#)** 沙箱类型、预装包与网络策略 - -**[API 总览与认证](t6645508.xdita#)** 在业务代码中集成 Managed Agents diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-session.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-session.md index e291b456..f5677bec 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-session.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/managed-agents/managed-agents-session.md @@ -1,9 +1,3 @@ # 委派任务给 Agent 会话承载智能体的一次运行实例，全部消息、工具调用与状态变更以事件形式记录。 - -**[发起会话](t6645504.xdita#)** 创建会话并发送第一条消息 - -**[会话操作](t6679532.xdita#)** 状态机、中断与审批工具调用 - -**[会话事件流](t6679533.xdita#)** 事件类型与 SSE 订阅 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/model-context-protocol/mcp-faq.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/model-context-protocol/mcp-faq.md index 42083d03..8204ac00 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/model-context-protocol/mcp-faq.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/model-context-protocol/mcp-faq.md @@ -59,7 +59,7 @@ 6. **为什么我的 MCP 服务无法访问远程资源（例如云数据库）？** - 因为 MCP 服务托管在[函数计算 FC](https://www.aliyun.com/product/fc)，无固定出口公网 IP。请为远程资源（如云数据库）配置[函数计算 FC 的 IP 白名单或进行 VPC 网络打通](https://help.aliyun.com/zh/functioncompute/fc/how-to-configure-an-ip-address-whitelist-when-i-access-a-database)。 + 因为 MCP 服务托管在[函数计算 FC](https://www.aliyun.com/product/fc)，无固定出口公网 IP。请为远程资源（如云数据库）配置[函数计算 FC 的 IP 白名单或进行 VPC 网络打通](https://help.aliyun.com/zh/functioncompute/how-to-configure-an-ip-address-whitelist-when-i-access-a-database)。 7. **我的 MCP Server 保存在私有 npm 仓库中，可以部署到阿里云百炼吗？** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/plug-in/custom-plug-ins.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/plug-in/custom-plug-ins.md deleted file mode 100644 index 2d7dc544..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/plug-in/custom-plug-ins.md +++ /dev/null @@ -1,421 +0,0 @@ -# 自定义插件 - -当阿里云百炼的官方插件无法满足您的业务需求时，您可以通过创建自定义插件来扩展大模型的能力。本文档将引导您完成从创建、调试到使用的全过程，轻松集成所需 API。 - -## **工作流程** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1340161871/CAEQZhiBgICczru.2hkiIGNhZjkzMjM4YWJlZjRhOGU5NzA1Mzk1MWFkNGI1MzBh4867389_20250110113852.335.svg) - -1. **创建****/导入****插件：**定义插件的基础信息，或直接从云市场导入。 - -2. **添加工具**（导入插件无需此步）**：**为插件配置具体的 API 路径、请求参数和返回数据。 - -3. **调试与发布：**在线测试 API 的连通性，确保功能正常后发布。 - -4. **在应用中使用：**将插件关联到智能体，通过对话测试或 API 集成来调用。 - - -## **创建自定义插件** - -## **创建个性化开发的插件** - -### **步骤一：创建插件** - -1. 访问[**插件**](https://bailian.console.aliyun.com/?tab=app#/component-manage)页面，单击**创建插件**。 - -2. 填写插件信息。 - - **插件名称**：输入具有语义的名称，支持中英文。 - - > 示例：寝室公约查询工具test - - **插件描述**：对插件功能和使用场景的简要说明。能帮助大模型判断当前任务是否需要调用当前插件，请使用自然语言进行描述。 - - > 示例：根据输入的数字索引查询特定条目的寝室公约内容。 - - **插件URL**：插件的访问地址。 - - > 示例：https://domitorgreement-plugin-example-icohrkdjxy.cn-beijing.fcapp.run - - - 同一个域名下，不同的路径被拆分成了不同的API（即下方创建工具中的**工具路径**） - - - 同一插件下的不同工具使用相同的域名，每个工具的工具路径对应一个独立的API - - 例如：xx插件下包含两个API： - - 查询：https://xxx.com/query - - 删除：https://xxx.com/delete - - 在这个示例中，`https://xxx.com`对应**插件URL**，`/query`和`/delete`对应下方创建工具中的**工具路径**。这表明该插件下包含两个工具。 - - - 如需要鉴权请打开**是否鉴权**开关，填写鉴权配置信息。 - - **鉴权参数说明** - - **Header列表**（可选） - - 需要鉴权时，可以通过自定义Header传递鉴权信息。 - - **是否鉴权**（可选） - - 当阿里云百炼应用调用您的自定义插件时是否需要鉴权。此处是否需要鉴权主要取决于API提供方的安全策略。 - - **鉴权类型** - - 鉴权包括**服务级鉴权**和**用户级鉴权**两种方式。 - - - **位置**：支持将鉴权信息放在Header或Query中。 - - - **Header**：将鉴权信息放在HTTP请求头的Authorization字段中，这些信息在URL中不可见。 - - - **Query**：将鉴权信息放在URL中，例如https://example.com?api\_key=123456。 - - - **参数名**：如果将鉴权信息放在Query中需填写鉴权时使用的参数，如“api\_key”。如果将鉴权信息放在Header中将默认此参数为“Authorization”。 - - - **Type**： - - - **basic**：不会在您提供的Token前加任何内容； - - - **bearer**：会在Token前增加“Bearer”； - - - **appcode**：会在Token前增加“APPCODE”。 - - - 两种type都会放在鉴权参数字段中。例如：选择bearer，则调用插件时会变成（“Authorization”： “Bearer ”）。 - - - **Token**（服务级鉴权）：从API提供方获取的鉴权Token，如API Key。 - - -3. 填写完成后单击**确认创建插件** > **创建工具**或单击**继续添加工具**。 - - -### **步骤二：创建工具** - -1. 填写工具信息、配置输入/输出参数以及高级配置。 - - 本示例中，**工具名称**填写"寝室公约查询工具"，**工具描述**填写"根据输入的数字索引查询特定条目的寝室公约内容"，**工具路径**填写`/article`，**请求方法**选择**POST**，**提交方式**选择**application/json**。输入参数：参数名称`article_index`，参数描述为"索引"，类型为**Number**，传入方法为**Body**，必填，传参方式为**大模型识别**。输出参数：参数名称`article`，参数描述为"寝室公约内容"，类型为**String**。高级配置中，用户输入Query为"请根据输入的索引值，查询对应条目的寝室公约内容"，输入参数`article_index`的Value为`5`。 - - **工具参数说明** - - **工具信息** - - **工具名称** - - 输入具有语义的名称，支持中英文。 - - **工具描述** - - 对工具功能和使用场景的简要说明。 - - 帮助大模型判断当前任务是否需要调用该工具，请使用自然语言进行描述，尽量给出使用示例。 - - **工具路径** - - 指向**插件URL**的相对路径。 - - 字段名必须以正斜杠（/）开头。 - - 此路径将拼接到**插件URL**上，以构建完整的URL。 - - **请求方法** - - 根据实际需求选择GET或POST请求方法来调用API接口。 - - **提交方式** - - 请求或响应的编码类型。 - - - **application/json**：主体内容是JSON格式的数据。 - - - **application/x-www-form-urlencoded**：将表单数据编码为键值对。 - - - 这种编码方式用于 POST 请求，表单数据编码为键值对，并通过 URL 编码传输。多个键值对之间用 & 分隔，每个键和值之间用 = 分隔。URL 编码会将特殊字符转换为 % 后跟两位十六进制数的形式。例如，空格会被编码为 %20，& 会被编码为 %26，= 会被编码为 %3D - - - 示例：`name=John Doe&age=25` 被编码为 `name=John%20Doe&age=25` - - - **配置输入参数与输出参数** - - **配置输入参数** - - 单击**增加入参**，配置参数信息。 - - **参数名称**：尽可能带有含义，帮助大模型理解当前需要识别的参数信息是什么。例如`city`。 - - **参数描述**：对该入参的功能描述，要简练且准确，帮助大模型进一步理解取参的方式。例如，`date`，描述为日期的同时，可以再进一步描述date的形式，比如`yyyy-MM-dd`。 - - **类型**：指参数类型。 - - **重要** - - Object类型下的子属性不能为空。请单击该对象行末的![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1891396371/p905454.png)图标新增子属性。 - - **传参方式**：要设置准确。 - - - **大模型识别**：表示该参数的值需要大模型从用户输入中提取。 - - - **业务透传**：表示该参数的值从外部主动透传，传递过程中不对数据进行处理或修改。 - - 使用DashScope SDK或HTTP接口调用应用时，插件中的业务透传类型的输入参数信息通过 `biz_params`和`user_defined_params`传递给应用。具体请参见[应用的参数传递](https://help.aliyun.com/zh/model-studio/pass-through-of-application-parameters)。 - - - **配置输出参数** - - 单击**增加出参**，配置参数信息，所有参数均为**必填**项。 - - 大模型会根据出参的定义，结合用户的问题，对API返回的结果进行筛选、重新组合，作为最终的答案返回给用户。 - - 出参与入参一样，需要尽可能精简和准确描述，嵌套的层级也尽可能少。 - - **重要** - - 无论请求方式是GET还是POST，参数都支持Object类型。但Object类型下的子属性不能为空。请单击该对象行末的![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1891396371/p905454.png)图标新增子属性。 - - **高级配置（可选）** - - **高级配置** - - 为大模型增加调用示例，减少漏召回和误召回的情况。 - - 通常用于入参比较复杂，模型构造容易出错的场景，通过提供一些样例，提升大模型调用插件的准确性。 - - **Value**：表示用户输入当前的Query时，期望大模型构造出的调用入参。示例：用户输入：“查询杭州明天的天气”，期望构造入参为：`{"city": "杭州", "date": "2025-04-25"}`。 - -2. 配置完成后单击**保存草稿**。 - -3. 在线调试工具API能否调通。 - - 单击**测试工具**，输入鉴权信息（开启鉴权时填写）及入参的值，单击**开始运行**。 - - > 如果运行失败，请根据**运行结果**中的报错信息对配置进行调整，并重新进行测试，直至成功运行。 - - 入参的值可以手动输入，也可以通过代码输入。对于入参较为复杂的情况，建议采用**代码模式编辑**。您可以在代码编辑器中提交完整的JSON格式的入参及其相应的值。 - -4. 测试通过后单击**发布**。只有**已发布**的工具才能在应用中被调用。 - - -## **从云市场导入插件** - -云市场提供了丰富的API，您可在云市场开通需要的API并将其导入至阿里云百炼插件列表中，以便应用进行调用。 - -1. 访问[**插件**](https://bailian.console.aliyun.com/?tab=app#/component-manage)页面，单击**从云市场导入**。 - -2. 首次在阿里云百炼平台上导入云市场 API 作为插件，需要先进行服务关联角色授权。 - - ## 主账号 - - 如果您使用主账号登录百炼，请在**SLR授权**弹窗中，勾选同意上述条款，单击**确认授权**。 - - 该弹窗显示角色名称为 `AliyunServiceRoleForSFMAccessCloudAPI`，角色权限策略为 `AliyunServiceRolePolicyForSFMAccessCloudAPI`，授权百炼大模型平台访问阿里云云市场商品清单并根据插件配置进行API调用。 - - ## RAM用户（子账号） - - 如果您使用RAM用户（子账号）登录百炼，在**SLR授权**弹窗中，勾选同意上述条款，单击**确认授权**时，会有如下提示： - - 页面弹出**授权失败**弹窗，提示当前用户没有创建服务关联角色的权限，服务关联角色名称为 `AliyunServiceRoleForSFMAccessCloudAPI`。 - - 这是因为该RAM用户（子账号）不具备创建服务关联角色的权限。请按照下述操作先授予RAM用户（子账号）创建服务关联角色的权限。获得授权后，RAM用户（子账号）即可自行从云市场导入插件至百炼或使用主账号已经导入的插件。 - - 1. 授权RAM用户（子账号）创建服务关联角色的权限。 - - 1. 使用主账号登录[RAM控制台](https://ram.console.aliyun.com/)。 - - 2. 在左侧导航栏，选择**权限管理** > **权限策略**。 - - 3. 单击**创建权限策略**。 - - 4. 在**脚本编辑**的`Effect`、`Action`、`Resource`、`Condition`中分别输入以下脚本中的对应内容。 - - ``` - { - "Action": [ - "ram:CreateServiceLinkedRole" - ], - "Resource": "*", - "Effect": "Allow", - "Condition": { - "StringEquals": { - "ram:ServiceName": "cloundapi-access.sfm.aliyuncs.com" - } - } - } - ``` - - 5. 单击**确定**。 - - 6. 设置权限策略名称，单击**确定**。 - - 此处**名称**设置为`服务关联角色`。 - - 7. 在左侧导航栏，选择。 - - 8. 找到待授权的RAM用户（子账号），单击RAM用户（子账号）**操作**列的**添加权限**。 - - 9. 在权限策略中选择刚才创建的权限策略，单击**确认新增授权**。 - - 至此，RAM用户（子账号）拥有了创建服务关联角色的权限。 - - 在**资源范围**中选择**账号级别**，在权限策略筛选下拉框中选择**自定义策略**类型，即可找到并勾选目标策略。 - - 2. RAM用户（子账号）自行从云市场导入插件至阿里云百炼或使用主账号已经导入的插件。 - - 返回阿里云百炼控制台，单击**从云市场导入**，在**SLR授权**弹窗中，勾选同意上述条款，单击**确认授权** - - -3. 在**导入云市场插件**弹窗中，单击**点击查看**进入云市场开通需要的API。 - -4. 您可以查看**云市场**页面，等待商品**状态**为**已开通**。 - - 当前页面还提供了API的AppKey、AppSecret、AppCode，如果插件需要鉴权，您可以在此处获取鉴权信息。 - -5. 开通成功后，返回阿里云百炼控制台，单击**从云市场导入**，重新打开**导入云市场插件**弹窗。选择已开通的API，再单击**确定**，进入**工具列表**页面。 - -6. 从云市场导入的插件为草稿状态，需要测试、发布后再使用。 - - 1. 单击工具所在行的**调试**，进入**工具测试**页面。 - - 2. 输入参数后，单击**开始运行**。 - - 若**运行成功**，则说明接口正常。否则，请参考界面提示信息修改配置。 - - 3. 返回**编辑工具**页面，单击**发布**，发布工具。 - - > 从云市场导入插件时，系统将自动填充出参和入参信息，但可能会存在信息缺失的情况。在发布工具时，您需要关注无法发布的错误信息，以便根据错误提示有效地解决问题。 - - 例如，当**工具名称**超出20字符限制时，输入框右侧的字符计数将以红色显示（如`22/20`），您需要将名称缩短至限制范围内后再发布。 - - 4. **已发布**且**启用**状态的工具才能用于后续调用。 - - 同时确保工具的**调试**状态为**成功**。 - - -## 使用插件 - -## 控制台 - -- **方式一**：将插件发布为MCP服务，然后在智能体应用中添加该MCP服务。 - - **步骤一：将插件发布为MCP服务** - - 1. 在**插件列表**中，将鼠标悬浮在目标插件卡片上，单击**发布为MCP服务**。 - - > 如果插件已转为MCP服务，按钮显示为**查看MCP服务**，单击可跳转到MCP管理页面查看服务详情。 - - 2. 发布成功后，可在**MCP管理**页面查看该MCP服务的详细信息，包括服务名称、服务描述、服务ID等。 - - - **步骤二：在智能体应用中添加MCP服务** - - 1. 进入**智能体应用**的编排页面，在**MCP**区块中，单击**+**。 - - 2. 在**选择MCP服务**面板中，切换到**自定义MCP**页签，找到从插件转换的MCP服务，单击**添加全部**将其添加到应用中。 - - > 您也可以单击**从插件转MCP**，将尚未转换的插件直接发布为MCP服务。 - - 3. 测试插件的使用效果是否符合预期。 - - - 无鉴权：您可以在输入框中与大模型进行对话，测试插件使用效果。 - - - 用户级鉴权、服务级鉴权：您需要在开启对话前，单击![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1891396371/p905403.png)配置需要传入的鉴权Token。如果不离开当前页面，可以只配置一次。 - - > 从云市场导入的插件，不需要在当前页面输入鉴权Token。 - - - 工具入参的**传参方式**选择了**业务透传**：您需要在开启对话前，单击![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1891396371/p905403.png)配置需要传入的变量值。如果不离开当前页面，可以只输入一次。 - - 4. 测试完成后，**发布**应用。 - -- **方式二**：在[**应用管理**](https://bailian.console.aliyun.com/#/app-center)页面中，进入**智能体应用**的编排页面，在**MCP**区块中添加MCP服务，测试插件使用效果，并**发布**应用。 - - -## API - -**获取工具 ID** - -工具ID用于标识具体的工具。通过API调用工具时，需要正确传递工具ID，以确保请求能够被正确识别。 - -1. 在**插件列表**中，找到工具所属的插件，单击**查看详情**。 - -2. 将鼠标悬浮于工具名称旁边的![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1891396371/p902180.png)图标上。 - -3. 单击![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1891396371/p902183.png)图标，复制工具ID。 - - -- 当通过API调用应用时，如果应用中关联的插件存在业务透传参数或开启了**用户级鉴权**，则需要通过参数`biz_params`传递鉴权信息或透传参数信息。具体操作请参见[工作流与旧版智能体应用 API](https://help.aliyun.com/zh/model-studio/agent-and-workflow-application-api-reference)。 - -- 通过Assistant API调用工具。请在[Assistant API文档](https://help.aliyun.com/zh/model-studio/quick-start-of-assistant-api)中搜索`tools`关键字，查看如何使用Assistant API调用工具。 - - -## **管理自定义插件与工具** - -**删除插件** - -**重要** - -删除插件会删除插件下的所有工具，并且调用了插件的应用会失效，此操作不可撤回，请谨慎操作。 - -在**插件列表**中，找到目标插件，单击**删除**。 - -**编辑插件** - -1. 在**插件列表**中，找到目标插件，单击**查看详情**。 - -2. 单击右上角的**编辑插件**，修改插件信息并保存。 - - 插件信息保存后立即生效。如果修改了插件的URL、Header、鉴权信息，可能会影响工具调用，请重新测试并发布工具。 - - -**编辑工具** - -工具信息修改完成后，需要重新测试并发布，才能生效。 - -1. 在**插件列表**中，找到工具所属插件，单击**查看详情**。 - -2. 单击工具所在行的**编辑**，修改工具信息并单击**保存草稿**。 - -3. 单击**测试工具**，在线调试工具。 - -4. 运行成功后单击**发布**。 - - -**删除工具** - -**重要** - -删除工具后，调用了此工具的应用会失效，此操作不可撤回，请谨慎操作。 - -1. 在**插件列表**中，找到工具所属插件，单击**查看详情**。 - -2. 单击工具所在行的**删除**。 - - -## **错误码** - -发布工具时的常见错误信息如下表所示： - -**错误码** - -**错误信息** - -**说明** - -**130040** - -xx缺少参数描述信息 - -原因：xx参数的参数描述缺失。 - -解决方案：请您补充参数描述后重新发布工具。 - -**130022** - -保存工具信息异常/请检查示例参数是否正确 - -可能原因一：输入参数或输出参数中的Object类型参数子属性为空。 - -解决方案：请点击该对象行末的![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1891396371/p905454.png)图标新增子属性。 - -可能原因二：请求方法选择了GET，但输入参数配置时，存在参数为Object类型。 - -解决方案：GET请求方法下的输入参数不支持Object类型，请选择其他类型。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md index 62ca0940..382676af 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md @@ -21,7 +21,9 @@ Tripo 3D模型生成支持**文生3D模型**、**单图生3D模型和多图生3D ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -35,7 +37,8 @@ Tripo 3D模型生成支持**文生3D模型**、**单图生3D模型和多图生3D ## 文生3D模型（有贴图） ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -53,7 +56,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 单图生3D模型（有贴图） ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -71,7 +75,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 多图生3D模型（有贴图） ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -96,7 +101,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener > 图片顺序为前、左、后、右，不需要的视角传入空对象`{}`即可。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -121,7 +127,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener > 需同时将`texture`和`pbr`设为`false`。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -324,7 +331,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -343,7 +350,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md index 601bfe7e..b7ba1534 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md @@ -39,8 +39,6 @@ 点击查看完整示例 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/iwaouc/asr_example.wav)。 - ``` from http import HTTPStatus import dashscope @@ -58,7 +56,7 @@ recognition = Recognition(model='fun-asr-realtime', format='wav', sample_rate=16000, callback=None) -result = recognition.call('asr_example.wav') +result = recognition.call('{YOUR_AUDIO_FILE}') if result.status_code == HTTPStatus.OK: print('识别结果：') print(result.get_sentence()) @@ -221,8 +219,6 @@ if __name__ == '__main__': ## 识别本地语音文件 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/acoict/asr_example.wav)。 - ``` import os import time @@ -270,8 +266,8 @@ recognition = Recognition(model='fun-asr-realtime', try: audio_data: bytes = None - f = open("asr_example.wav", 'rb') - if os.path.getsize("asr_example.wav"): + f = open("{YOUR_AUDIO_FILE}", 'rb') + if os.path.getsize("{YOUR_AUDIO_FILE}"): # 一次性将文件数据全部读入buffer file_buffer = f.read() f.close() diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md index 10ee8202..3a99c0b6 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md @@ -1,6 +1,6 @@ -# Fun-ASR录音文件识别Android SDK +# Fun-ASR非实时语音识别Android SDK -本文档提供了Fun-ASR录音文件识别Android SDK的详细使用指南，帮助您将语音转换为文本。 +本文档提供了Fun-ASR非实时语音识别Android SDK的详细使用指南，帮助您将语音转换为文本。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 @@ -13,7 +13,7 @@ - [下载最新SDK整合包](https://help.aliyun.com/zh/isi/sdk-selection-and-download)。 - 解压 ZIP 包。在 `app/libs` 目录中获取 AAR 格式 SDK，并添加到项目依赖。 - 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 + 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 - 用 Android Studio 打开工程。示例代码位于`DashFunAsrFileTranscriberActivity.java`，替换 API Key 后体验功能。 @@ -99,7 +99,7 @@ 是 - 运行模式。录音文件识别固定为 `"1"`。 + 运行模式。非实时语音识别固定为 `"1"`。 `device_id` @@ -170,7 +170,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { @@ -596,7 +596,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md index 2dbd2ce9..2ccfc352 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md @@ -1,12 +1,10 @@ -# Fun-ASR录音文件识别HTTP API参考 +# Fun-ASR非实时语音识别HTTP API参考 -本文介绍Fun-ASR录音文件识别HTTP API的参数和接口细节。 +本文介绍Fun-ASR非实时语音识别HTTP API的参数和接口细节。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 -## **DashScope异步调用（Fun-ASR）** - -### **流程说明** +## **流程说明** 与DashScope同步调用（一次请求、立即返回结果）不同，异步调用专为处理长音频文件或耗时较长的任务设计，该模式采用“提交-轮询”的两步式流程，避免了因长时间等待而导致的请求超时： @@ -23,7 +21,7 @@ - 当任务处理完成后，结果查询接口将返回最终的识别结果。 -### **服务端点** +## **服务端点** ## 华北2（北京） @@ -56,7 +54,7 @@ 使用新版域名（`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`）提交任务时，请求体中必须包含`parameters`对象。即使无需设置任何参数，也必须传入空对象`{}`，否则任务可正常提交，但识别将失败。 -### **请求头** +## **请求头** **参数** @@ -90,11 +88,11 @@ string 异步任务标识。仅提交任务接口需要传入，固定为`enable`，请勿遗漏，否则无法提交任务。 -### **提交任务接口** +## **提交任务接口** 提交语音识别任务。该接口异步返回，业务侧需结合[查询任务接口](#480630e0582sb)轮询任务状态。 -#### **请求体** +### **请求体** 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 @@ -107,7 +105,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi "model": "fun-asr", "input": { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ] }, "parameters": { @@ -289,7 +287,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi - en: 英文 -#### **返回体** +### **返回体** ``` { @@ -319,11 +317,11 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi 任务状态。提交成功时返回`PENDING`。 -### **查询任务接口** +## **查询任务接口** 查询语音识别任务的执行情况和结果。建议轮询调用直至任务终态。 -#### **请求体** +### **请求体** 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 @@ -340,7 +338,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks 查询任务需指定其ID，该ID为[提交任务接口](#418f2ac8ecxm4)被调用后返回的`task_id`。 -#### **返回体** +### **返回体** ## 正常示例 @@ -355,7 +353,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks "end_time": "2024-09-12 15:11:40.903", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url": "{YOUR_AUDIO_URL}", "transcription_url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/pre/filetrans-16k/20240912/15%3A11/409a4b92-445b-4dd8-8c1d-f110954d82d8-1.json?Expires=1726211500&OSSAccessKeyId=yourOSSAccessKeyId&Signature=v5Owy5qoAfT7mzGmQgH0g8C****%3D", "subtask_status": "SUCCEEDED" } @@ -383,7 +381,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "FILE_DOWNLOAD_FAILED", "message": "FILE_DOWNLOAD_FAILED", "subtask_status": "FAILED" @@ -485,11 +483,11 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks 失败的子任务数。 -### **其他接口：批量查询任务状态/取消任务** +## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 -### **识别结果说明** +## **识别结果说明** 识别结果保存为JSON文件。 @@ -497,7 +495,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -637,887 +635,3 @@ punctuation string 预测出的词之后的标点符号（如有）。 - -## **DashScope同步调用（Fun-ASR-Flash）** - -**重要** - -该功能不支持SDK调用。 - -### **服务端点** - -## 华北2（北京） - -`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -## 新加坡 - -`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -**重要** - -阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，能够为推理请求提供卓越的性能和更高的稳定性，建议迁移至新域名： - -- 华北2（北京）地域：从 `dashscope.aliyuncs.com` 迁移至 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - -- 新加坡地域：从 `dashscope-intl.aliyuncs.com` 迁移至 `{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` - - -`{WorkspaceId}`需要替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。现有域名仍可正常使用。 - -### **请求头** - -**参数** - -**类型** - -**是否必选** - -**说明** - -Authorization - -string - -是 - -鉴权令牌，格式为`Bearer `，使用时将"``"替换为实际的API Key。 - -Content-Type - -string - -是 - -请求体的媒体类型，固定为`application/json`。 - -X-DashScope-SSE - -string - -是 - -用于控制是否以SSE流式方式返回结果。设置为`enable`时开启SSE流式返回模式，服务端会分多次返回中间识别结果和最终结果；设置为`disable`或不传该参数则仅返回最终结果。 - -### **请求体** - -以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - -## 非流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: disable" \ - --data '{ - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" - } - } - ] - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000" - } -}' -``` - -## 流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: enable" \ - --data '{ - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" - } - } - ] - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000" - } -}' -``` - -## 携带上下文-非流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: disable" \ - --data '{ - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_text", - "text": "你好啊" - } - ] - }, - { - "role": "assistant", - "content": [ - { - "type": "text", - "text": "你好啊，我是通义千问，有什么可以帮助你的？" - } - ] - }, - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" - } - } - ] - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000" - } -}' -``` - -## 携带上下文-流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: enable" \ - --data '{ - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_text", - "text": "你好啊" - } - ] - }, - { - "role": "assistant", - "content": [ - { - "type": "text", - "text": "你好啊，我是通义千问，有什么可以帮助你的？" - } - ] - }, - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" - } - } - ] - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000" - } -}' -``` - -## Base64 - -可输入Base64编码数据（[Data URL](https://www.rfc-editor.org/rfc/rfc2397)），格式为：`data:;base64,`。 - -- ``：MIME类型 - - 因音频格式而异，例如： - - - WAV：`audio/wav` - - - MP3：`audio/mpeg` - -- ``：音频转成的Base64编码的字符串 - - Base64编码会增大体积，请控制原文件大小，确保编码后仍符合输入音频大小限制（10MB） - -- 示例：`data:audio/wav;base64,SUQzBAAAAAAAI1RTU0UAAAAPAAADTGF2ZjU4LjI5LjEwMAAAAAAAAAAAAAAA//PAxABQ/BXRbMPe4IQAhl9` - - **点击查看示例代码** - - Python - - ``` - import base64, pathlib - - # input.mp3为待识别的本地音频文件，请替换为自己的音频文件路径，确保其符合音频要求 - file_path = pathlib.Path("input.mp3") - base64_str = base64.b64encode(file_path.read_bytes()).decode() - data_uri = f"data:audio/mpeg;base64,{base64_str}" - ``` - - Java - - ``` - import java.nio.file.*; - import java.util.Base64; - - public class Main { - /** - * filePath为待识别的本地音频文件，请替换为自己的音频文件路径，确保其符合音频要求 - */ - public static String toDataUrl(String filePath) throws Exception { - byte[] bytes = Files.readAllBytes(Paths.get(filePath)); - String encoded = Base64.getEncoder().encodeToString(bytes); - return "data:audio/mpeg;base64," + encoded; - } - - public static void main(String[] args) throws Exception { - System.out.println(toDataUrl("input.mp3")); - } - } - ``` - - -``` -import base64, pathlib -import os -import requests - -# input.wav为待识别的本地音频文件，请替换为自己的音频文件路径，确保其符合音频要求 -file_path = pathlib.Path("input.wav") -base64_str = base64.b64encode(file_path.read_bytes()).decode() -data_uri = f"data:audio/wav;base64,{base64_str}" - -url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" - -headers = { - "Authorization": f"Bearer {os.environ['DASHSCOPE_API_KEY']}", - "Content-Type": "application/json", - "X-DashScope-SSE": "disable", -} - -payload = { - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": data_uri, - }, - } - ], - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000", - }, -} - -response = requests.post(url, headers=headers, json=payload) -print(response.status_code) -print(response.json()) -``` - -**model** `_string_` **（必选）** - -模型名称，固定为`fun-asr-flash-2026-06-15`。 - -**input** `_object_` **（必选）** - -输入信息。 - -**属性** - -**messages** `_array(object)_` **（必选）** - -消息列表。包含当前待识别的音频，以及可选的对话上下文（用于提升识别效果）。 - -**重要** - -上下文功能用于提升专有词汇的识别准确率，使用方法详见[快速开始](https://help.aliyun.com/zh/model-studio/improve-asr-accuracy#ctx-quickstart-sec)。约束：上下文消息（`input_text` 和 `text` 类型）各最多 5 条，超出时保留最近的 5 条。每轮上下文文本总长度（`user` 和 `assistant` 的 `text` 字段长度之和）不超过 400 个字符（按字符数计算，每个字符计为 1），超出部分从末尾截断。 - -**重要** - -携带上下文时，`messages` 中的消息顺序有要求：上下文消息必须按对话轮次排列，每轮中 `user`（`input_text` 类型）必须在对应的 `assistant`（`text` 类型）之前；包含 `input_audio` 的 `user` 消息必须放在 `messages` 数组的最后。 - -**属性** - -**role** `_string_` **（必选）** - -消息角色。取值范围： - -- `user`（必选）：用户消息。type为`input_audio`时表示当前待识别的音频；type为`input_text`时表示前几轮的识别结果或领域相关的词表（可选，上下文）。 - -- `assistant`（可选，上下文）：前几轮大语言模型的回复内容。 - - -**content** `_array(object)_` **（必选）** - -消息内容列表。 - -**属性** - -**type** `_string_` **（必选）** - -内容类型。每个请求至少需要一条`input_audio`类型的消息。取值范围： - -- `input_audio`（必选）：当前待识别的音频输入（role为user），需同时传入`input_audio`对象。 - -- `input_text`（可选，上下文）：前几轮用户语音的识别结果或领域相关的词表（role为user），需同时传入`text`字段。 - -- `text`（可选，上下文）：前几轮大语言模型的回复内容（role为assistant），需同时传入`text`字段。 - - -**input\_audio** `_object_` **（条件必选）** - -当`type`为`input_audio`时必填。 - -**属性** - -**data** `_string_` **（必选）** - -待识别音频数据。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。支持以下两种方式： - -- **音频文件URL**：直接传入可公开访问的音频文件地址。 - -- **Base64 Data URI**：采用Data URI格式传入Base64编码的音频数据，值由`data:{MIME_TYPE};base64,`前缀与Base64编码的音频数据拼接而成。支持的MIME类型包括`audio/wav`、`audio/mp3`等。 - - -示例（URL方式）：`https://example.com/audio/sample.wav` - -示例（Base64方式）：`data:audio/wav;base64,{BASE64_ENCODED_DATA}` - -**text** `_string_` **（条件必选）** - -当`type`为`input_text`时，填入前几轮用户语音的识别结果或领域相关的词表；当`type`为`text`时，填入前几轮大语言模型的回复内容。文本按字符数计算，每个字符计为 1。每轮上下文中所有消息的 `text` 字段长度之和不超过 400 个字符，超出部分从末尾截断。 - -**parameters** `_object_` **（必选）** - -模型参数。 - -**属性** - -**format** `_string_` **（必选）** - -音频格式。根据实际音频格式填写，支持`wav`、`mp3`、`opus`等。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -**sample\_rate** `_string_` （可选） - -音频采样率，单位Hz。例如`16000`表示16kHz采样率。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -### **返回体** - -## 非流式 - -``` -{ - "output": { - "sentence": { - "begin_time": 760, - "channel_id": 0, - "end_time": 3800, - "sentence_end": true, - "sentence_id": 1, - "text": "Hello World，这里是阿里巴巴语音实验室。", - "words": [ - {"begin_time": 760, "end_time": 1040, "fixed": true, "punctuation": "", "text": "Hello"}, - {"begin_time": 1040, "end_time": 1240, "fixed": true, "punctuation": "，", "text": " World"}, - {"begin_time": 1360, "end_time": 1880, "fixed": true, "punctuation": "", "text": "这里是"}, - {"begin_time": 1880, "end_time": 2520, "fixed": true, "punctuation": "", "text": "阿里巴巴"}, - {"begin_time": 2520, "end_time": 2840, "fixed": true, "punctuation": "", "text": "语音"}, - {"begin_time": 2840, "end_time": 3800, "fixed": true, "punctuation": "。", "text": "实验室"} - ] - }, - "text": "Hello World，这里是阿里巴巴语音实验室。" - }, - "usage": { - "duration": 4 - }, - "request_id": "40e0734d-096f-9ae3-86c1-a8c013287561" -} -``` - -## 流式 - -设置`X-DashScope-SSE: enable`时，服务端以Server-Sent Events协议返回识别结果。SSE事件格式如下： - -``` -id:{序列号} -event:result -:HTTP_STATUS/200 -data:{JSON数据} -``` - -返回示例： - -``` -id:1 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"sentence_id":1,"sentence_end":true,"end_time":3800,"words":[{"end_time":1040,"punctuation":"","begin_time":760,"fixed":true,"text":"Hello"},{"end_time":1240,"punctuation":"，","begin_time":1040,"fixed":true,"text":" World"},{"end_time":1880,"punctuation":"","begin_time":1360,"fixed":true,"text":"这里是"},{"end_time":2520,"punctuation":"","begin_time":1880,"fixed":true,"text":"阿里巴巴"},{"end_time":2840,"punctuation":"","begin_time":2520,"fixed":true,"text":"语音"},{"end_time":3800,"punctuation":"。","begin_time":2840,"fixed":true,"text":"实验室"}],"begin_time":760,"text":"Hello World，这里是阿里巴巴语音实验室。","channel_id":0},"text":"Hello World，这里是阿里巴巴语音实验室。"},"usage":{"duration":4},"request_id":"fc1582e4-935c-9fc2-a482-a98bf43daa69"} -``` - -**request\_id** `_string_` - -本次请求的唯一标识。 - -**output** `_object_` - -输出结果。 - -**属性** - -**text** `_string_` - -当前累积的完整识别文本。 - -**sentence** `_object_` - -当前句子的详细信息。 - -**属性** - -**sentence\_id** `_integer_` - -句子编号，从1开始。 - -**sentence\_end** `_boolean_` - -是否为该句的最终结果。为`true`时表示该句识别完成。 - -**begin\_time** `_integer_` - -句子开始时间，单位毫秒。 - -**end\_time** `_integer_` - -句子结束时间，单位毫秒。仅在`sentence_end`为`true`时返回。 - -**text** `_string_` - -当前句子的识别文本。 - -**channel\_id** `_integer_` - -声道编号，从0开始。 - -**words** `_array_` - -词级别时间戳列表。 - -**属性** - -**text** `_string_` - -词文本。 - -**begin\_time** `_integer_` - -词开始时间，单位毫秒。 - -**end\_time** `_integer_` - -词结束时间，单位毫秒。 - -**punctuation** `_string_` - -词后的标点符号。无标点时为空字符串。 - -**fixed** `_boolean_` - -词是否已稳定。`false`表示后续事件中该词的时间戳可能调整。 - -**usage** `_object_` - -用量信息。仅在`sentence_end`为`true`时返回。 - -**属性** - -**duration** `_integer_` - -已处理的音频时长，单位秒。 - -### **SSE 流式结果处理逻辑** - -在流式模式下，客户端需关注以下处理要点： - -1. 每收到一个SSE事件，解析`data`字段中的JSON。 - -2. 通过`output.sentence.sentence_end`判断当前句子是否结束：当该值为`true`时，该句识别完成，词级时间戳已稳定，可作为最终结果使用；当该值为`false`时，识别仍在进行中，文本和时间戳可能在后续事件中更新。 - -3. `usage`信息仅在句子结束事件中返回，可用于计量音频处理时长。 - - -## **DashScope同步调用（Fun-ASR-R**ealtime**）** - -**重要** - -- 该功能只支持北京地域。 - -- 不支持SDK调用。 - - -### **服务端点** - -`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -**重要** - -阿里云百炼为华北2（北京）地域推出了业务空间专属域名，能够为推理请求提供卓越的性能和更高的稳定性，建议从 `dashscope.aliyuncs.com` 迁移至 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com`。 - -`{WorkspaceId}`需要替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。现有域名仍可正常使用。 - -### **请求头** - -**参数** - -**类型** - -**是否必选** - -**说明** - -Authorization - -string - -是 - -鉴权令牌，格式为`Bearer `，使用时将“``”替换为实际的API Key。 - -Content-Type - -string - -是 - -请求体的媒体类型，固定为`application/json`。 - -X-DashScope-SSE - -string - -是 - -用于控制是否以SSE流式方式返回结果。设置为`enable`时开启SSE流式返回模式，服务端会分多次返回中间识别结果和最终结果；设置为`disable`或不传该参数则仅返回最终结果。 - -### **请求体** - -## 非流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: disable" \ - --data '{ - "model": "fun-asr-realtime", - "input": { - "messages": [] - }, - "parameters": { - "audio_address": "https://example.com/audio/sample.mp3", - "format": "mp3" - }, - "resources": [] -}' -``` - -## 流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: enable" \ - --data '{ - "model": "fun-asr-realtime", - "input": { - "messages": [] - }, - "parameters": { - "audio_address": "https://example.com/audio/sample.mp3", - "format": "mp3" - }, - "resources": [] -}' -``` - -**model** `_string_` **（必选）** - -模型名称。 - -取值范围： - -- `fun-asr-realtime`（稳定版模型） - -- `fun-asr-realtime-2026-02-28`（即 Fun-Realtime-ASR-preview） - - -**input** `_object_` **（条件必选）** - -输入信息。与输入音频文件URL方式（通过`parameters.audio_address`传入）二选一，使用Base64方式上传音频时需要填写。 - -**属性** - -**messages** `_array(object)_` **（必选）** - -消息列表。 - -**属性** - -**content** `_array(object)_` **（必选）** - -用户消息的内容。仅允许设置一组消息。 - -**属性** - -**audio** `_string_` **（必选）** - -待识别音频，采用Data URI格式传入Base64编码的音频数据。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -使用Base64方式上传音频时需要填写。 - -值由`data:{MIME_TYPE};base64,`前缀与Base64编码的音频数据拼接而成。支持的MIME类型包括`audio/wav`、`audio/mp3`等。 - -示例：`data:audio/wav;base64,{BASE64_ENCODED_DATA}` - -**role** `_string_` **（必选）** - -用户消息的角色，固定为`user`。使用Base64方式上传音频时需要填写。 - -**parameters** `_object_` **（必选）** - -模型参数。 - -**属性** - -**audio\_address** `_string_` **（条件必选）** - -音频文件URL地址。与Base64方式（通过`input.messages`传入）二选一，使用URL方式时必填。需为可公开访问的地址。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -**format** `_string_` **（必选）** - -音频格式。根据实际音频格式填写，支持`wav`、`mp3`、`opus`等。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -**vad\_enabled** `_boolean_` （可选） - -是否启用端点检测（VAD）。 - -默认为`true`（启用 VAD 检测）。 - -设为 `false` 时关闭 VAD 检测；但当音频时长超过 1 分钟时，系统将自动启用 VAD 检测，忽略此参数设置。 - -**resources** `_array_` （可选） - -资源列表，预留字段，当前可传空数组`[]`。 - -### **返回体** - -## 非流式 - -``` -{ - "output": { - "sentence": { - "begin_time": 160, - "channel_id": 0, - "end_time": 1680, - "sentence_end": true, - "sentence_id": 1, - "text": "欢迎使用阿里云。", - "words": [ - {"begin_time": 160, "end_time": 520, "fixed": true, "punctuation": "", "text": "欢迎"}, - {"begin_time": 520, "end_time": 880, "fixed": true, "punctuation": "", "text": "使用"}, - {"begin_time": 880, "end_time": 1280, "fixed": true, "punctuation": "", "text": "阿里"}, - {"begin_time": 1280, "end_time": 1680, "fixed": true, "punctuation": "。", "text": "云"} - ] - }, - "text": "欢迎使用阿里云。" - }, - "usage": { - "duration": 2 - }, - "request_id": "eff4c092-2289-9b43-a4cd-80e591fa90f5" -} -``` - -## 流式 - -设置`X-DashScope-SSE: enable`时，服务端以Server-Sent Events协议分多次返回中间识别结果和最终结果。每个SSE事件格式如下： - -``` -id:{序列号} -event:result -:HTTP_STATUS/200 -data:{JSON数据} -``` - -中间结果（句子开始、词逐步增长）： - -``` -id:1 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"sentence_id":1,"sentence_end":false,"sentence_begin":true,"words":[],"begin_time":0,"text":"","channel_id":0},"text":""},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} - -id:2 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":false,"text":"欢迎"}],"begin_time":160,"text":"欢迎","channel_id":0,"sentence_id":1,"sentence_end":false},"text":"欢迎"},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} - -id:3 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":false,"text":"欢迎"},{"end_time":880,"punctuation":"","begin_time":520,"fixed":false,"text":"使用"}],"begin_time":160,"text":"欢迎使用","channel_id":0,"sentence_id":1,"sentence_end":false},"text":"欢迎使用"},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} -``` - -最终结果（句子结束，包含`usage`）： - -``` -id:4 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"sentence_id":1,"sentence_end":true,"end_time":1680,"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":true,"text":"欢迎"},{"end_time":880,"punctuation":"","begin_time":520,"fixed":true,"text":"使用"},{"end_time":1280,"punctuation":"","begin_time":880,"fixed":true,"text":"阿里"},{"end_time":1680,"punctuation":"。","begin_time":1280,"fixed":true,"text":"云"}],"begin_time":160,"text":"欢迎使用阿里云。","channel_id":0},"text":"欢迎使用阿里云。"},"usage":{"duration":2},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} -``` - -**request\_id** `_string_` - -本次请求的唯一标识。 - -**output** `_object_` - -输出结果。 - -**属性** - -**text** `_string_` - -当前累积的完整识别文本。 - -**sentence** `_object_` - -当前句子的详细信息。 - -**属性** - -**sentence\_id** `_integer_` - -句子编号，从1开始。 - -**sentence\_end** `_boolean_` - -是否为该句的最终结果。为`true`时表示该句识别完成。 - -**begin\_time** `_integer_` - -句子开始时间，单位毫秒。 - -**end\_time** `_integer_` - -句子结束时间，单位毫秒。仅在`sentence_end`为`true`时返回。 - -**text** `_string_` - -当前句子的识别文本。 - -**channel\_id** `_integer_` - -声道编号，从0开始。 - -**words** `_array_` - -词级别时间戳列表。 - -**属性** - -**text** `_string_` - -词文本。 - -**begin\_time** `_integer_` - -词开始时间，单位毫秒。 - -**end\_time** `_integer_` - -词结束时间，单位毫秒。 - -**punctuation** `_string_` - -词后的标点符号。无标点时为空字符串。 - -**fixed** `_boolean_` - -词是否已稳定。`false`表示后续事件中该词的时间戳可能调整。 - -**usage** `_object_` - -用量信息。仅在`sentence_end`为`true`时返回。 - -**属性** - -**duration** `_integer_` - -已处理的音频时长，单位秒。 - -### **SSE 流式结果处理逻辑** - -在流式模式下，客户端需关注以下处理要点： - -1. 每收到一个SSE事件，解析`data`字段中的JSON。 - -2. 通过`output.sentence.sentence_end`判断当前句子是否结束：当该值为`true`时，该句识别完成，词级时间戳已稳定，可作为最终结果使用；当该值为`false`时，识别仍在进行中，文本和时间戳可能在后续事件中更新。 - -3. `usage`信息仅在句子结束事件中返回，可用于计量音频处理时长。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md index e59f1695..928864d9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md @@ -1,6 +1,6 @@ -# Fun-ASR录音文件识别iOS SDK +# Fun-ASR非实时语音识别iOS SDK -本文档提供了Fun-ASR录音文件识别iOS SDK的详细使用指南，帮助您将语音转换为文本。 +本文档提供了Fun-ASR非实时语音识别iOS SDK的详细使用指南，帮助您将语音转换为文本。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 @@ -102,7 +102,7 @@ 是 - 运行模式。录音文件识别固定为 `"1"`。 + 运行模式。非实时语音识别固定为 `"1"`。 `device_id` @@ -173,7 +173,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { @@ -590,7 +590,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md index 8008ce57..47e93c0f 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md @@ -1,6 +1,6 @@ -# Fun-ASR录音文件识别Java SDK +# Fun-ASR非实时语音识别Java SDK -本文介绍Fun-ASR录音文件识别Java SDK的参数和接口细节。 +本文介绍Fun-ASR非实时语音识别Java SDK的参数和接口细节。 **重要** @@ -32,7 +32,7 @@ ## **快速开始** -[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行录音文件识别： +[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行非实时语音识别： - 异步提交任务+同步等待任务结束：提交任务后，阻塞当前线程直到任务结束并获取识别结果。 @@ -41,7 +41,7 @@ ### **异步提交任务+同步等待任务结束** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4867892871/CAEQURiBgMCO2_fRpxkiIDBlNzI4YmMyNTU3ODRlM2Y4NjUxZWU4YmUxNjliMmFl4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6278074871/CAEQURiBgMCO2_fRpxkiIDBlNzI4YmMyNTU3ODRlM2Y4NjUxZWU4YmUxNjliMmFl4709861_20241015153444.149.svg) 1. 配置[请求参数](#48ea212b1d08r)。 @@ -85,7 +85,7 @@ public class Main { .model("fun-asr") // 此处以fun-asr为例，可按需更换模型名称。模型列表：https://help.aliyun.com/zh/model-studio/models .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -107,7 +107,7 @@ public class Main { ### **异步提交任务+异步查询任务执行结果** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4867892871/CAEQURiBgIDnxvjRpxkiIGI1NjJjOTgyNTVhMTRiMjM4OWVjYzFmZTExNGZjYzE14709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6278074871/CAEQURiBgIDnxvjRpxkiIGI1NjJjOTgyNTVhMTRiMjM4OWVjYzFmZTExNGZjYzE14709861_20241015153444.149.svg) 1. 配置[请求参数](#48ea212b1d08r)。 @@ -152,7 +152,7 @@ public class Main { .model("fun-asr") // 此处以fun-asr为例，可按需更换模型名称。模型列表：https://help.aliyun.com/zh/model-studio/models .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -188,7 +188,7 @@ TranscriptionParam param = TranscriptionParam.builder() .model("fun-asr") .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); ``` @@ -507,7 +507,7 @@ public JsonObject getOutput() "end_time":"2025-02-13 16:12:10.189", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/16%3A12/3baafe5f-d09d-46c6-8b01-724927670edb-1.json?Expires=1739520730&OSSAccessKeyId=yourOSSAccessKeyId&Signature=BF7vPxlsJN9hkJlY%2BLReezxOwK8%3D", "subtask_status":"SUCCEEDED" } @@ -533,7 +533,7 @@ public JsonObject getOutput() "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -615,7 +615,7 @@ public String getMessage() ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -775,7 +775,7 @@ TranscriptionParam param = .model("fun-asr") .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -854,7 +854,7 @@ public TranscriptionResult fetch(TranscriptionQueryParam queryParam) ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **错误码** @@ -873,7 +873,7 @@ public TranscriptionResult fetch(TranscriptionQueryParam queryParam) "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md index 97a0b5b3..051845e2 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md @@ -1,6 +1,6 @@ -# Fun-ASR录音文件识别Python SDK +# Fun-ASR非实时语音识别Python SDK -本文介绍Fun-ASR录音文件识别Python SDK的参数和接口细节。 +本文介绍Fun-ASR非实时语音识别Python SDK的参数和接口细节。 **重要** @@ -32,7 +32,7 @@ ## **快速开始** -[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行录音文件识别： +[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行非实时语音识别： - 异步提交任务+同步等待任务结束：提交任务后，阻塞当前线程直到任务结束并获取识别结果。 @@ -41,7 +41,7 @@ ### **异步提交任务+同步等待任务结束** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5594892871/CAEQURiBgMCvo5zjpxkiIDQyNzUwZjVjMWM3MjQ5Nzg4ODBjNDRjNzE1ZGFiOGFj4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3178074871/CAEQURiBgMCvo5zjpxkiIDQyNzUwZjVjMWM3MjQ5Nzg4ODBjNDRjNzE1ZGFiOGFj4709861_20241015153444.149.svg) 1. 调用[核心类（Transcription）](#adcb5e9bddbyq)的`async_call`方法并设置[请求参数](#340f6879fci7d)。 @@ -77,7 +77,7 @@ dashscope.api_key = os.getenv("DASHSCOPE_API_KEY") task_response = Transcription.async_call( model='fun-asr', - file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav'] + file_urls=['{YOUR_AUDIO_URL}'] ) transcribe_response = Transcription.wait(task=task_response.output.task_id) @@ -88,7 +88,7 @@ if transcribe_response.status_code == HTTPStatus.OK: ### **异步提交任务+异步查询任务执行结果** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5594892871/CAEQURiBgMCN3qzkpxkiIGE0YmU4YTdjMWNiNzRmYjJhMjFlMWZkZmFmOWQ1NmEx4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3178074871/CAEQURiBgMCN3qzkpxkiIGE0YmU4YTdjMWNiNzRmYjJhMjFlMWZkZmFmOWQ1NmEx4709861_20241015153444.149.svg) 1. 调用[核心类（Transcription）](#adcb5e9bddbyq)的`async_call`方法并设置[请求参数](#340f6879fci7d)。 @@ -124,7 +124,7 @@ dashscope.api_key = os.getenv("DASHSCOPE_API_KEY") transcribe_response = Transcription.async_call( model='fun-asr', - file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav'] + file_urls=['{YOUR_AUDIO_URL}'] ) while True: @@ -407,7 +407,7 @@ list\[str\] "end_time":"2025-02-13 17:31:21.867", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/17%3A31/20ee4e4f-0404-4806-b617-c7d4c62eed19-1.json?Expires=1739525481&OSSAccessKeyId=yourOSSAccessKeyId&Signature=3q%2B1uQmRwltd7FPn5HQM2mBKw74%3D", "subtask_status":"SUCCEEDED" } @@ -440,7 +440,7 @@ list\[str\] "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -563,7 +563,7 @@ transcription\_url "end_time":"2025-02-13 17:59:28.828", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/17%3A59/70e737cc-bf8c-418b-b0c8-83fab192a0fa-1.json?Expires=1739527168&OSSAccessKeyId=yourOSSAccessKeyId&Signature=AtGjIKI%2BdgbzjJIu%2BHsr1R5nSAY%3D", "subtask_status":"SUCCEEDED" } @@ -589,7 +589,7 @@ transcription\_url "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -657,7 +657,7 @@ transcription\_url ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -857,7 +857,7 @@ def fetch(cls, ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **错误码** @@ -876,7 +876,7 @@ def fetch(cls, "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md new file mode 100644 index 00000000..ecf7e674 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md @@ -0,0 +1,561 @@ +# 非实时语音识别（Fun-ASR-Flash）API参考 + +本文介绍Fun-ASR-Flash非实时语音识别HTTP API的参数和接口细节。 + +**用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**重要** + +该功能不支持SDK调用。 + +## **服务端点** + +## 华北2（北京） + +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +## 新加坡 + +`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +**重要** + +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，能够为推理请求提供卓越的性能和更高的稳定性，建议迁移至新域名： + +- 华北2（北京）地域：从 `dashscope.aliyuncs.com` 迁移至 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `dashscope-intl.aliyuncs.com` 迁移至 `{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +`{WorkspaceId}`需要替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。现有域名仍可正常使用。 + +## **请求头** + +**参数** + +**类型** + +**是否必选** + +**说明** + +Authorization + +string + +是 + +鉴权令牌，格式为`Bearer `，使用时将"``"替换为实际的API Key。 + +Content-Type + +string + +是 + +请求体的媒体类型，固定为`application/json`。 + +X-DashScope-SSE + +string + +是 + +用于控制是否以SSE流式方式返回结果。设置为`enable`时开启SSE流式返回模式，服务端会分多次返回中间识别结果和最终结果；设置为`disable`或不传该参数则仅返回最终结果。 + +## **请求体** + +以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 + +## 非流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: disable" \ + --data '{ + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": "{YOUR_AUDIO_URL}" + } + } + ] + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000" + } +}' +``` + +## 流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: enable" \ + --data '{ + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": "{YOUR_AUDIO_URL}" + } + } + ] + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000" + } +}' +``` + +## 携带上下文-非流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: disable" \ + --data '{ + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_text", + "text": "你好啊" + } + ] + }, + { + "role": "assistant", + "content": [ + { + "type": "text", + "text": "你好啊，我是通义千问，有什么可以帮助你的？" + } + ] + }, + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": "{YOUR_AUDIO_URL}" + } + } + ] + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000" + } +}' +``` + +## 携带上下文-流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: enable" \ + --data '{ + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_text", + "text": "你好啊" + } + ] + }, + { + "role": "assistant", + "content": [ + { + "type": "text", + "text": "你好啊，我是通义千问，有什么可以帮助你的？" + } + ] + }, + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": "{YOUR_AUDIO_URL}" + } + } + ] + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000" + } +}' +``` + +## Base64 + +可输入Base64编码数据（[Data URL](https://www.rfc-editor.org/rfc/rfc2397)），格式为：`data:;base64,`。 + +- ``：MIME类型 + + 因音频格式而异，例如： + + - WAV：`audio/wav` + + - MP3：`audio/mpeg` + +- ``：音频转成的Base64编码的字符串 + + Base64编码会增大体积，请控制原文件大小，确保编码后仍符合输入音频大小限制（10MB） + +- 示例：`data:audio/wav;base64,SUQzBAAAAAAAI1RTU0UAAAAPAAADTGF2ZjU4LjI5LjEwMAAAAAAAAAAAAAAA//PAxABQ/BXRbMPe4IQAhl9` + + **点击查看示例代码** + + Python + + ``` + import base64, pathlib + + # 请替换为自己的音频文件路径，确保其符合音频要求 + file_path = pathlib.Path("{YOUR_AUDIO_FILE}") + base64_str = base64.b64encode(file_path.read_bytes()).decode() + data_uri = f"data:audio/mpeg;base64,{base64_str}" + ``` + + Java + + ``` + import java.nio.file.*; + import java.util.Base64; + + public class Main { + /** + * 请替换为自己的音频文件路径，确保其符合音频要求 + */ + public static String toDataUrl(String filePath) throws Exception { + byte[] bytes = Files.readAllBytes(Paths.get(filePath)); + String encoded = Base64.getEncoder().encodeToString(bytes); + return "data:audio/mpeg;base64," + encoded; + } + + public static void main(String[] args) throws Exception { + System.out.println(toDataUrl("{YOUR_AUDIO_FILE}")); + } + } + ``` + + +``` +import base64, pathlib +import os +import requests + +# 请替换为自己的音频文件路径，确保其符合音频要求 +file_path = pathlib.Path("{YOUR_AUDIO_FILE}") +base64_str = base64.b64encode(file_path.read_bytes()).decode() +data_uri = f"data:audio/wav;base64,{base64_str}" + +url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" + +headers = { + "Authorization": f"Bearer {os.environ['DASHSCOPE_API_KEY']}", + "Content-Type": "application/json", + "X-DashScope-SSE": "disable", +} + +payload = { + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": data_uri, + }, + } + ], + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000", + }, +} + +response = requests.post(url, headers=headers, json=payload) +print(response.status_code) +print(response.json()) +``` + +**model** `_string_` **（必选）** + +模型名称，固定为`fun-asr-flash-2026-06-15`。 + +**input** `_object_` **（必选）** + +输入信息。 + +**属性** + +**messages** `_array(object)_` **（必选）** + +消息列表。包含当前待识别的音频，以及可选的对话上下文（用于提升识别效果）。 + +**重要** + +上下文功能用于提升专有词汇的识别准确率，使用方法详见[快速开始](https://help.aliyun.com/zh/model-studio/improve-asr-accuracy#ctx-quickstart-sec)。约束：上下文消息（`input_text` 和 `text` 类型）各最多 5 条，超出时保留最近的 5 条。每轮上下文文本总长度（`user` 和 `assistant` 的 `text` 字段长度之和）不超过 400 个字符（按字符数计算，每个字符计为 1），超出部分从末尾截断。 + +**重要** + +携带上下文时，`messages` 中的消息顺序有要求：上下文消息必须按对话轮次排列，每轮中 `user`（`input_text` 类型）必须在对应的 `assistant`（`text` 类型）之前；包含 `input_audio` 的 `user` 消息必须放在 `messages` 数组的最后。 + +**属性** + +**role** `_string_` **（必选）** + +消息角色。取值范围： + +- `user`（必选）：用户消息。type为`input_audio`时表示当前待识别的音频；type为`input_text`时表示前几轮的识别结果或领域相关的词表（可选，上下文）。 + +- `assistant`（可选，上下文）：前几轮大语言模型的回复内容。 + + +**content** `_array(object)_` **（必选）** + +消息内容列表。 + +**属性** + +**type** `_string_` **（必选）** + +内容类型。每个请求至少需要一条`input_audio`类型的消息。取值范围： + +- `input_audio`（必选）：当前待识别的音频输入（role为user），需同时传入`input_audio`对象。 + +- `input_text`（可选，上下文）：前几轮用户语音的识别结果或领域相关的词表（role为user），需同时传入`text`字段。 + +- `text`（可选，上下文）：前几轮大语言模型的回复内容（role为assistant），需同时传入`text`字段。 + + +**input\_audio** `_object_` **（条件必选）** + +当`type`为`input_audio`时必填。 + +**属性** + +**data** `_string_` **（必选）** + +待识别音频数据。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。支持以下两种方式： + +- **音频文件URL**：直接传入可公开访问的音频文件地址。 + +- **Base64 Data URI**：采用Data URI格式传入Base64编码的音频数据，值由`data:{MIME_TYPE};base64,`前缀与Base64编码的音频数据拼接而成。支持的MIME类型包括`audio/wav`、`audio/mp3`等。 + + +示例（URL方式）：`https://example.com/audio/sample.wav` + +示例（Base64方式）：`data:audio/wav;base64,{BASE64_ENCODED_DATA}` + +**text** `_string_` **（条件必选）** + +当`type`为`input_text`时，填入前几轮用户语音的识别结果或领域相关的词表；当`type`为`text`时，填入前几轮大语言模型的回复内容。文本按字符数计算，每个字符计为 1。每轮上下文中所有消息的 `text` 字段长度之和不超过 400 个字符，超出部分从末尾截断。 + +**parameters** `_object_` **（必选）** + +模型参数。 + +**属性** + +**format** `_string_` **（必选）** + +音频格式。根据实际音频格式填写，支持`wav`、`mp3`、`opus`等。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**sample\_rate** `_string_` （可选） + +音频采样率，单位Hz。例如`16000`表示16kHz采样率。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +## **返回体** + +## 非流式 + +``` +{ + "output": { + "sentence": { + "begin_time": 760, + "channel_id": 0, + "end_time": 3800, + "sentence_end": true, + "sentence_id": 1, + "text": "Hello World，这里是阿里巴巴语音实验室。", + "words": [ + {"begin_time": 760, "end_time": 1040, "fixed": true, "punctuation": "", "text": "Hello"}, + {"begin_time": 1040, "end_time": 1240, "fixed": true, "punctuation": "，", "text": " World"}, + {"begin_time": 1360, "end_time": 1880, "fixed": true, "punctuation": "", "text": "这里是"}, + {"begin_time": 1880, "end_time": 2520, "fixed": true, "punctuation": "", "text": "阿里巴巴"}, + {"begin_time": 2520, "end_time": 2840, "fixed": true, "punctuation": "", "text": "语音"}, + {"begin_time": 2840, "end_time": 3800, "fixed": true, "punctuation": "。", "text": "实验室"} + ] + }, + "text": "Hello World，这里是阿里巴巴语音实验室。" + }, + "usage": { + "duration": 4 + }, + "request_id": "40e0734d-096f-9ae3-86c1-a8c013287561" +} +``` + +## 流式 + +设置`X-DashScope-SSE: enable`时，服务端以Server-Sent Events协议返回识别结果。SSE事件格式如下： + +``` +id:{序列号} +event:result +:HTTP_STATUS/200 +data:{JSON数据} +``` + +返回示例： + +``` +id:1 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"sentence_id":1,"sentence_end":true,"end_time":3800,"words":[{"end_time":1040,"punctuation":"","begin_time":760,"fixed":true,"text":"Hello"},{"end_time":1240,"punctuation":"，","begin_time":1040,"fixed":true,"text":" World"},{"end_time":1880,"punctuation":"","begin_time":1360,"fixed":true,"text":"这里是"},{"end_time":2520,"punctuation":"","begin_time":1880,"fixed":true,"text":"阿里巴巴"},{"end_time":2840,"punctuation":"","begin_time":2520,"fixed":true,"text":"语音"},{"end_time":3800,"punctuation":"。","begin_time":2840,"fixed":true,"text":"实验室"}],"begin_time":760,"text":"Hello World，这里是阿里巴巴语音实验室。","channel_id":0},"text":"Hello World，这里是阿里巴巴语音实验室。"},"usage":{"duration":4},"request_id":"fc1582e4-935c-9fc2-a482-a98bf43daa69"} +``` + +**request\_id** `_string_` + +本次请求的唯一标识。 + +**output** `_object_` + +输出结果。 + +**属性** + +**text** `_string_` + +当前累积的完整识别文本。 + +**sentence** `_object_` + +当前句子的详细信息。 + +**属性** + +**sentence\_id** `_integer_` + +句子编号，从1开始。 + +**sentence\_end** `_boolean_` + +是否为该句的最终结果。为`true`时表示该句识别完成。 + +**begin\_time** `_integer_` + +句子开始时间，单位毫秒。 + +**end\_time** `_integer_` + +句子结束时间，单位毫秒。仅在`sentence_end`为`true`时返回。 + +**text** `_string_` + +当前句子的识别文本。 + +**channel\_id** `_integer_` + +声道编号，从0开始。 + +**words** `_array_` + +词级别时间戳列表。 + +**属性** + +**text** `_string_` + +词文本。 + +**begin\_time** `_integer_` + +词开始时间，单位毫秒。 + +**end\_time** `_integer_` + +词结束时间，单位毫秒。 + +**punctuation** `_string_` + +词后的标点符号。无标点时为空字符串。 + +**fixed** `_boolean_` + +词是否已稳定。`false`表示后续事件中该词的时间戳可能调整。 + +**usage** `_object_` + +用量信息。仅在`sentence_end`为`true`时返回。 + +**属性** + +**duration** `_integer_` + +已处理的音频时长，单位秒。 + +## **SSE 流式结果处理逻辑** + +在流式模式下，客户端需关注以下处理要点： + +1. 每收到一个SSE事件，解析`data`字段中的JSON。 + +2. 通过`output.sentence.sentence_end`判断当前句子是否结束：当该值为`true`时，该句识别完成，词级时间戳已稳定，可作为最终结果使用；当该值为`false`时，识别仍在进行中，文本和时间戳可能在后续事件中更新。 + +3. `usage`信息仅在句子结束事件中返回，可用于计量音频处理时长。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md new file mode 100644 index 00000000..3c19542d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md @@ -0,0 +1,328 @@ +# 非实时语音识别（Fun-ASR-Realtime）API参考 + +本文介绍Fun-ASR-Realtime非实时语音识别HTTP API的参数和接口细节。 + +**用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**重要** + +- 该功能只支持北京地域。 + +- 不支持SDK调用。 + + +## **服务端点** + +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +**重要** + +阿里云百炼为华北2（北京）地域推出了业务空间专属域名，能够为推理请求提供卓越的性能和更高的稳定性，建议从 `dashscope.aliyuncs.com` 迁移至 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com`。 + +`{WorkspaceId}`需要替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。现有域名仍可正常使用。 + +## **请求头** + +**参数** + +**类型** + +**是否必选** + +**说明** + +Authorization + +string + +是 + +鉴权令牌，格式为`Bearer `，使用时将“``”替换为实际的API Key。 + +Content-Type + +string + +是 + +请求体的媒体类型，固定为`application/json`。 + +X-DashScope-SSE + +string + +是 + +用于控制是否以SSE流式方式返回结果。设置为`enable`时开启SSE流式返回模式，服务端会分多次返回中间识别结果和最终结果；设置为`disable`或不传该参数则仅返回最终结果。 + +## **请求体** + +## 非流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: disable" \ + --data '{ + "model": "fun-asr-realtime", + "input": { + "messages": [] + }, + "parameters": { + "audio_address": "https://example.com/audio/sample.mp3", + "format": "mp3" + }, + "resources": [] +}' +``` + +## 流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: enable" \ + --data '{ + "model": "fun-asr-realtime", + "input": { + "messages": [] + }, + "parameters": { + "audio_address": "https://example.com/audio/sample.mp3", + "format": "mp3" + }, + "resources": [] +}' +``` + +**model** `_string_` **（必选）** + +模型名称。 + +取值范围： + +- `fun-asr-realtime`（稳定版模型） + +- `fun-asr-realtime-2026-02-28`（即 Fun-Realtime-ASR-preview） + + +**input** `_object_` **（条件必选）** + +输入信息。与输入音频文件URL方式（通过`parameters.audio_address`传入）二选一，使用Base64方式上传音频时需要填写。 + +**属性** + +**messages** `_array(object)_` **（必选）** + +消息列表。 + +**属性** + +**content** `_array(object)_` **（必选）** + +用户消息的内容。仅允许设置一组消息。 + +**属性** + +**audio** `_string_` **（必选）** + +待识别音频，采用Data URI格式传入Base64编码的音频数据。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +使用Base64方式上传音频时需要填写。 + +值由`data:{MIME_TYPE};base64,`前缀与Base64编码的音频数据拼接而成。支持的MIME类型包括`audio/wav`、`audio/mp3`等。 + +示例：`data:audio/wav;base64,{BASE64_ENCODED_DATA}` + +**role** `_string_` **（必选）** + +用户消息的角色，固定为`user`。使用Base64方式上传音频时需要填写。 + +**parameters** `_object_` **（必选）** + +模型参数。 + +**属性** + +**audio\_address** `_string_` **（条件必选）** + +音频文件URL地址。与Base64方式（通过`input.messages`传入）二选一，使用URL方式时必填。需为可公开访问的地址。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**format** `_string_` **（必选）** + +音频格式。根据实际音频格式填写，支持`wav`、`mp3`、`opus`等。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**vad\_enabled** `_boolean_` （可选） + +是否启用端点检测（VAD）。 + +默认为`true`（启用 VAD 检测）。 + +设为 `false` 时关闭 VAD 检测；但当音频时长超过 1 分钟时，系统将自动启用 VAD 检测，忽略此参数设置。 + +**resources** `_array_` （可选） + +资源列表，预留字段，当前可传空数组`[]`。 + +## **返回体** + +## 非流式 + +``` +{ + "output": { + "sentence": { + "begin_time": 160, + "channel_id": 0, + "end_time": 1680, + "sentence_end": true, + "sentence_id": 1, + "text": "欢迎使用阿里云。", + "words": [ + {"begin_time": 160, "end_time": 520, "fixed": true, "punctuation": "", "text": "欢迎"}, + {"begin_time": 520, "end_time": 880, "fixed": true, "punctuation": "", "text": "使用"}, + {"begin_time": 880, "end_time": 1280, "fixed": true, "punctuation": "", "text": "阿里"}, + {"begin_time": 1280, "end_time": 1680, "fixed": true, "punctuation": "。", "text": "云"} + ] + }, + "text": "欢迎使用阿里云。" + }, + "usage": { + "duration": 2 + }, + "request_id": "eff4c092-2289-9b43-a4cd-80e591fa90f5" +} +``` + +## 流式 + +设置`X-DashScope-SSE: enable`时，服务端以Server-Sent Events协议分多次返回中间识别结果和最终结果。每个SSE事件格式如下： + +``` +id:{序列号} +event:result +:HTTP_STATUS/200 +data:{JSON数据} +``` + +中间结果（句子开始、词逐步增长）： + +``` +id:1 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"sentence_id":1,"sentence_end":false,"sentence_begin":true,"words":[],"begin_time":0,"text":"","channel_id":0},"text":""},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} + +id:2 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":false,"text":"欢迎"}],"begin_time":160,"text":"欢迎","channel_id":0,"sentence_id":1,"sentence_end":false},"text":"欢迎"},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} + +id:3 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":false,"text":"欢迎"},{"end_time":880,"punctuation":"","begin_time":520,"fixed":false,"text":"使用"}],"begin_time":160,"text":"欢迎使用","channel_id":0,"sentence_id":1,"sentence_end":false},"text":"欢迎使用"},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} +``` + +最终结果（句子结束，包含`usage`）： + +``` +id:4 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"sentence_id":1,"sentence_end":true,"end_time":1680,"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":true,"text":"欢迎"},{"end_time":880,"punctuation":"","begin_time":520,"fixed":true,"text":"使用"},{"end_time":1280,"punctuation":"","begin_time":880,"fixed":true,"text":"阿里"},{"end_time":1680,"punctuation":"。","begin_time":1280,"fixed":true,"text":"云"}],"begin_time":160,"text":"欢迎使用阿里云。","channel_id":0},"text":"欢迎使用阿里云。"},"usage":{"duration":2},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} +``` + +**request\_id** `_string_` + +本次请求的唯一标识。 + +**output** `_object_` + +输出结果。 + +**属性** + +**text** `_string_` + +当前累积的完整识别文本。 + +**sentence** `_object_` + +当前句子的详细信息。 + +**属性** + +**sentence\_id** `_integer_` + +句子编号，从1开始。 + +**sentence\_end** `_boolean_` + +是否为该句的最终结果。为`true`时表示该句识别完成。 + +**begin\_time** `_integer_` + +句子开始时间，单位毫秒。 + +**end\_time** `_integer_` + +句子结束时间，单位毫秒。仅在`sentence_end`为`true`时返回。 + +**text** `_string_` + +当前句子的识别文本。 + +**channel\_id** `_integer_` + +声道编号，从0开始。 + +**words** `_array_` + +词级别时间戳列表。 + +**属性** + +**text** `_string_` + +词文本。 + +**begin\_time** `_integer_` + +词开始时间，单位毫秒。 + +**end\_time** `_integer_` + +词结束时间，单位毫秒。 + +**punctuation** `_string_` + +词后的标点符号。无标点时为空字符串。 + +**fixed** `_boolean_` + +词是否已稳定。`false`表示后续事件中该词的时间戳可能调整。 + +**usage** `_object_` + +用量信息。仅在`sentence_end`为`true`时返回。 + +**属性** + +**duration** `_integer_` + +已处理的音频时长，单位秒。 + +## **SSE 流式结果处理逻辑** + +在流式模式下，客户端需关注以下处理要点： + +1. 每收到一个SSE事件，解析`data`字段中的JSON。 + +2. 通过`output.sentence.sentence_end`判断当前句子是否结束：当该值为`true`时，该句识别完成，词级时间戳已稳定，可作为最终结果使用；当该值为`false`时，识别仍在进行中，文本和时间戳可能在后续事件中更新。 + +3. `usage`信息仅在句子结束事件中返回，可用于计量音频处理时长。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md index 2872f1a4..b9a5a871 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md @@ -143,14 +143,12 @@ 提交单个语音实时转写任务，通过传入本地文件的方式同步阻塞地拿到转写结果。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9365892871/CAEQURiBgMDS0c2RpxkiIDNmYjBlMTE3ODQxYTQ3Nzk4MGMxNTc5MjY3OWVjZjlj4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3558074871/CAEQURiBgMDS0c2RpxkiIDNmYjBlMTE3ODQxYTQ3Nzk4MGMxNTc5MjY3OWVjZjlj4709861_20241015153444.149.svg) 实例化[Recognition类](#adcb5e9bddbyq)，调用`call`方法绑定[请求参数](#d72d661a1brzp)和待识别文件，进行识别并最终获取识别结果。 点击查看完整示例 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/elouas/asr_example.wav)。 - ``` import com.alibaba.dashscope.audio.asr.recognition.Recognition; import com.alibaba.dashscope.audio.asr.recognition.RecognitionParam; @@ -161,7 +159,7 @@ import java.io.File; public class Main { public static void main(String[] args) { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseWebsocketApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + Constants.baseWebsocketApiUrl = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference"; // 创建Recognition实例 Recognition recognizer = new Recognition(); // 创建RecognitionParam @@ -177,7 +175,7 @@ public class Main { .build(); try { - System.out.println("识别结果：" + recognizer.call(param, new File("asr_example.wav"))); + System.out.println("识别结果：" + recognizer.call(param, new File("{YOUR_AUDIO_FILE}"))); } catch (Exception e) { e.printStackTrace(); } finally { @@ -200,7 +198,7 @@ public class Main { 提交单个语音实时转写任务，通过实现回调接口的方式流式输出实时识别结果。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9365892871/CAEQURiBgID1ooWUpxkiIDcyOTEyYjZiZmUxNzRkZjVhMTNhYmNkYjI2NzYzYTMy4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3558074871/CAEQURiBgID1ooWUpxkiIDcyOTEyYjZiZmUxNzRkZjVhMTNhYmNkYjI2NzYzYTMy4709861_20241015153444.149.svg) 1. 启动流式语音识别 @@ -244,7 +242,7 @@ import java.util.concurrent.TimeUnit; public class Main { public static void main(String[] args) throws InterruptedException { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseWebsocketApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + Constants.baseWebsocketApiUrl = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference"; ExecutorService executorService = Executors.newSingleThreadExecutor(); executorService.submit(new RealtimeRecognitionTask()); executorService.shutdown(); @@ -332,8 +330,6 @@ class RealtimeRecognitionTask implements Runnable { ## 识别本地语音文件 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/oiydrd/asr_example.wav)。 - ``` import com.alibaba.dashscope.audio.asr.recognition.Recognition; import com.alibaba.dashscope.audio.asr.recognition.RecognitionParam; @@ -364,9 +360,9 @@ class TimeUtils { public class Main { public static void main(String[] args) throws InterruptedException { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseWebsocketApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + Constants.baseWebsocketApiUrl = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference"; ExecutorService executorService = Executors.newSingleThreadExecutor(); - executorService.submit(new RealtimeRecognitionTask(Paths.get(System.getProperty("user.dir"), "asr_example.wav"))); + executorService.submit(new RealtimeRecognitionTask(Paths.get(System.getProperty("user.dir"), "{YOUR_AUDIO_FILE}"))); executorService.shutdown(); // wait for all tasks to complete @@ -503,7 +499,7 @@ import java.nio.ByteBuffer; public class Main { public static void main(String[] args) throws NoApiKeyException { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseWebsocketApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + Constants.baseWebsocketApiUrl = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference"; // 创建一个Flowable Flowable audioSource = Flowable.create( diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md index bba41596..38973852 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md @@ -143,19 +143,17 @@ 提交单个语音实时转写任务，通过传入本地文件的方式同步阻塞地拿到转写结果。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6455892871/CAEQURiBgMDS0c2RpxkiIDNmYjBlMTE3ODQxYTQ3Nzk4MGMxNTc5MjY3OWVjZjlj4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4858074871/CAEQURiBgMDS0c2RpxkiIDNmYjBlMTE3ODQxYTQ3Nzk4MGMxNTc5MjY3OWVjZjlj4709861_20241015153444.149.svg) 实例化[Recognition类](#d6bc1f133f871)绑定[请求参数](#555007db2033f)，调用`call`进行识别/翻译并最终获取[识别结果（RecognitionResult）](#bc3e1a43d6hhy)。 点击查看完整示例 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/iwaouc/asr_example.wav)。 - ``` from http import HTTPStatus from dashscope.audio.asr import Recognition # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 -dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" +dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference" # 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key # import dashscope @@ -167,13 +165,15 @@ recognition = Recognition(model='paraformer-realtime-v2', # “language_hints”只支持paraformer-realtime-v2模型 language_hints=['zh', 'en'], callback=None) -result = recognition.call('asr_example.wav') +result = recognition.call('{YOUR_AUDIO_FILE}') if result.status_code == HTTPStatus.OK: print('识别结果：') - print(result.get_sentence()) + sentences = result.get_sentence() + for sentence in sentences: + print(sentence['text']) else: print('Error: ', result.message) - + print( '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}' .format( @@ -183,11 +183,15 @@ print( )) ``` +`result.get_sentence()`在非流式调用（`call`）中返回**句子列表**（`List[Dict]`），每个元素为`Dict[str, Any]`，包含`text`（识别文本）、`begin_time` / `end_time`（时间戳）、`words`（字时间戳）等字段。如需获取识别文本，需遍历列表并通过`sentence['text']`提取。 + +在流式回调（`on_event`）中，`result.get_sentence()`返回**单句信息**（`Dict[str, Any]`），可直接使用`sentence['text']`获取识别文本。两种调用模式的返回值类型不同，详见[识别结果（RecognitionResult）](#bc3e1a43d6hhy)中的`get_sentence`方法说明。 + ### **双向流式调用** 提交单个语音实时转写任务，通过实现回调接口的方式流式输出实时识别结果。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6455892871/CAEQURiBgIDvi..2pxkiIGE4NTc3Njg4ZGM2YzQ2NzVhZGI3MzE2YWUwYTA3OGEy4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4858074871/CAEQURiBgIDvi..2pxkiIGE4NTc3Njg4ZGM2YzQ2NzVhZGI3MzE2YWUwYTA3OGEy4709861_20241015153444.149.svg) 1. 启动流式语音识别 @@ -221,7 +225,7 @@ import dashscope import pyaudio from dashscope.audio.asr import * # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 -dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" +dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference" mic = None stream = None @@ -342,14 +346,12 @@ if __name__ == '__main__': ## 识别本地语音文件 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/acoict/asr_example.wav)。 - ``` import os import time from dashscope.audio.asr import * # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 -dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" +dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference" # 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key # import dashscope @@ -393,8 +395,8 @@ recognition.start() try: audio_data: bytes = None - f = open("asr_example.wav", 'rb') - if os.path.getsize("asr_example.wav"): + f = open("{YOUR_AUDIO_FILE}", 'rb') + if os.path.getsize("{YOUR_AUDIO_FILE}"): while True: audio_data = f.read(3200) if not audio_data: diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md index ff800b01..697261be 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别Android SDK +# Paraformer非实时语音识别Android SDK -本文档提供了Paraformer录音文件识别Android SDK的详细使用指南，帮助您将语音转换为文本。 +本文档提供了Paraformer非实时语音识别Android SDK的详细使用指南，帮助您将语音转换为文本。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) @@ -17,7 +17,10 @@ - [下载最新SDK整合包](https://help.aliyun.com/zh/isi/sdk-selection-and-download)。 - 解压 ZIP 包。在 `app/libs` 目录中获取 AAR 格式 SDK，并添加到项目依赖。 - 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 + 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 + + + - 用 Android Studio 打开工程。示例代码位于`DashParaformerFileTranscriberActivity.java`，替换 API Key 后体验功能。 @@ -103,7 +106,7 @@ 是 - 运行模式。录音文件识别固定为 `"1"`。 + 运行模式。非实时语音识别固定为 `"1"`。 `device_id` @@ -174,7 +177,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { @@ -617,7 +620,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md index 2147dda6..0ecf2ea8 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别iOS SDK +# Paraformer非实时语音识别iOS SDK -本文档提供了Paraformer录音文件识别iOS SDK的详细使用指南，帮助您将语音转换为文本。 +本文档提供了Paraformer非实时语音识别iOS SDK的详细使用指南，帮助您将语音转换为文本。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) @@ -106,7 +106,7 @@ 是 - 运行模式。录音文件识别固定为 `"1"`。 + 运行模式。非实时语音识别固定为 `"1"`。 `device_id` @@ -177,7 +177,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { @@ -611,7 +611,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md index b3d42809..65269d08 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别Java SDK +# Paraformer非实时语音识别Java SDK -本文介绍Paraformer录音文件识别Java SDK的参数和接口细节。 +本文介绍Paraformer非实时语音识别Java SDK的参数和接口细节。 **重要** @@ -27,7 +27,7 @@ ## **快速开始** -[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行录音文件识别： +[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行非实时语音识别： - 异步提交任务+同步等待任务结束：提交任务后，阻塞当前线程直到任务结束并获取识别结果。 @@ -36,7 +36,7 @@ ### **异步提交任务+同步等待任务结束** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7695892871/CAEQURiBgMCO2_fRpxkiIDBlNzI4YmMyNTU3ODRlM2Y4NjUxZWU4YmUxNjliMmFl4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0509074871/CAEQURiBgMCO2_fRpxkiIDBlNzI4YmMyNTU3ODRlM2Y4NjUxZWU4YmUxNjliMmFl4709861_20241015153444.149.svg) 1. 配置[请求参数](#48ea212b1d08r)。 @@ -80,7 +80,7 @@ public class Main { .parameter("language_hints", new String[]{"zh", "en"}) .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -102,7 +102,7 @@ public class Main { ### **异步提交任务+异步查询任务执行结果** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7695892871/CAEQURiBgIDnxvjRpxkiIGI1NjJjOTgyNTVhMTRiMjM4OWVjYzFmZTExNGZjYzE14709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0509074871/CAEQURiBgIDnxvjRpxkiIGI1NjJjOTgyNTVhMTRiMjM4OWVjYzFmZTExNGZjYzE14709861_20241015153444.149.svg) 1. 配置[请求参数](#48ea212b1d08r)。 @@ -147,7 +147,7 @@ public class Main { .parameter("language_hints", new String[]{"zh", "en"}) .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -185,7 +185,7 @@ TranscriptionParam param = TranscriptionParam.builder() .parameter("language_hints", new String[]{"zh", "en"}) .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); ``` @@ -537,7 +537,7 @@ public JsonObject getOutput() "end_time":"2025-02-13 16:12:10.189", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/16%3A12/3baafe5f-d09d-46c6-8b01-724927670edb-1.json?Expires=1739520730&OSSAccessKeyId=yourOSSAccessKeyId&Signature=BF7vPxlsJN9hkJlY%2BLReezxOwK8%3D", "subtask_status":"SUCCEEDED" } @@ -563,7 +563,7 @@ public JsonObject getOutput() "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -645,7 +645,7 @@ public String getMessage() ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -807,7 +807,7 @@ TranscriptionParam param = .parameter("language_hints", new String[]{"zh", "en"}) .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -886,7 +886,7 @@ public TranscriptionResult fetch(TranscriptionQueryParam queryParam) ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **错误码** @@ -907,7 +907,7 @@ public TranscriptionResult fetch(TranscriptionQueryParam queryParam) "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md index e3ffe0b6..175befce 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别Python SDK +# Paraformer非实时语音识别Python SDK -本文介绍Paraformer录音文件识别Python SDK的参数和接口细节。 +本文介绍Paraformer非实时语音识别Python SDK的参数和接口细节。 **重要** @@ -27,7 +27,7 @@ ## **快速开始** -[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行录音文件识别： +[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行非实时语音识别： - 异步提交任务+同步等待任务结束：提交任务后，阻塞当前线程直到任务结束并获取识别结果。 @@ -36,7 +36,7 @@ ### **异步提交任务+同步等待任务结束** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2057892871/CAEQURiBgMCvo5zjpxkiIDQyNzUwZjVjMWM3MjQ5Nzg4ODBjNDRjNzE1ZGFiOGFj4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1609074871/CAEQURiBgMCvo5zjpxkiIDQyNzUwZjVjMWM3MjQ5Nzg4ODBjNDRjNzE1ZGFiOGFj4709861_20241015153444.149.svg) 1. 调用[核心类（Transcription）](#adcb5e9bddbyq)的`async_call`方法并设置[请求参数](#340f6879fci7d)。 @@ -69,7 +69,7 @@ dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co task_response = Transcription.async_call( model='paraformer-v2', - file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav'], + file_urls=['{YOUR_AUDIO_URL}'], language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型 ) @@ -81,7 +81,7 @@ if transcribe_response.status_code == HTTPStatus.OK: ### **异步提交任务+异步查询任务执行结果** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2057892871/CAEQURiBgMCN3qzkpxkiIGE0YmU4YTdjMWNiNzRmYjJhMjFlMWZkZmFmOWQ1NmEx4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2609074871/CAEQURiBgMCN3qzkpxkiIGE0YmU4YTdjMWNiNzRmYjJhMjFlMWZkZmFmOWQ1NmEx4709861_20241015153444.149.svg) 1. 调用[核心类（Transcription）](#adcb5e9bddbyq)的`async_call`方法并设置[请求参数](#340f6879fci7d)。 @@ -114,7 +114,7 @@ dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co transcribe_response = Transcription.async_call( model='paraformer-v2', - file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav'], + file_urls=['{YOUR_AUDIO_URL}'], language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型 ) @@ -451,7 +451,7 @@ int "end_time":"2025-02-13 17:31:21.867", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/17%3A31/20ee4e4f-0404-4806-b617-c7d4c62eed19-1.json?Expires=1739525481&OSSAccessKeyId=yourOSSAccessKeyId&Signature=3q%2B1uQmRwltd7FPn5HQM2mBKw74%3D", "subtask_status":"SUCCEEDED" } @@ -484,7 +484,7 @@ int "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -607,7 +607,7 @@ transcription\_url "end_time":"2025-02-13 17:59:28.828", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/17%3A59/70e737cc-bf8c-418b-b0c8-83fab192a0fa-1.json?Expires=1739527168&OSSAccessKeyId=yourOSSAccessKeyId&Signature=AtGjIKI%2BdgbzjJIu%2BHsr1R5nSAY%3D", "subtask_status":"SUCCEEDED" } @@ -633,7 +633,7 @@ transcription\_url "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -701,7 +701,7 @@ transcription\_url ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -903,7 +903,7 @@ def fetch(cls, ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **错误码** @@ -924,7 +924,7 @@ def fetch(cls, "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md index 971b8184..2fb481c4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别RESTful API +# Paraformer非实时语音识别HTTP API -本文介绍Paraformer录音文件识别RESTful API的参数和接口细节。 +本文介绍Paraformer非实时语音识别HTTP API的参数和接口细节。 **重要** @@ -59,7 +59,7 @@ X-DashScope-Async: enable // 请勿遗漏该请求头，否则无法提交任务 "model":"paraformer-v2", //模型名，必选 "input":{ "file_urls":[ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ] //待识别文件，必选 } "parameters":{ @@ -90,7 +90,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header "Content-Type: application/json" \ --header "X-DashScope-Async: enable" \ - --data '{"model":"paraformer-v2","input":{"file_urls":["https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav"]},"parameters":{"channel_id":[0]}}' + --data '{"model":"paraformer-v2","input":{"file_urls":["{YOUR_AUDIO_URL}"]},"parameters":{"channel_id":[0]}}' ``` **参数** @@ -450,7 +450,7 @@ string "end_time": "2024-09-12 15:11:40.903", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url": "{YOUR_AUDIO_URL}", "transcription_url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/pre/filetrans-16k/20240912/15%3A11/409a4b92-445b-4dd8-8c1d-f110954d82d8-1.json?Expires=1726211500&OSSAccessKeyId=yourOSSAccessKeyId&Signature=v5Owy5qoAfT7mzGmQgH0g8C****%3D", "subtask_status": "SUCCEEDED" } @@ -480,7 +480,7 @@ string "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -546,7 +546,7 @@ JSON数据中各字段含义请参见[识别结果说明](#a9021178ccl7s)。 ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -689,7 +689,7 @@ string ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **完整示例** @@ -704,7 +704,7 @@ import time api_key = "your-dashscope-api-key" # 在此处替换为您的API Key file_urls = [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "{YOUR_AUDIO_URL}", ] language_hints = ["zh", "en"] @@ -797,7 +797,7 @@ print("transcription result: ", result) "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md index 4a80c542..e847488f 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md @@ -1,4 +1,4 @@ -# 录音文件识别（Qwen-ASR）API参考 +# 非实时语音识别（Qwen-ASR）API参考 本文介绍 Qwen-ASR 模型的输入与输出参数。可通过OpenAI 兼容或DashScope协议调用 API。 @@ -70,7 +70,7 @@ try: # 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key = "sk-xxx", api_key=os.getenv("DASHSCOPE_API_KEY"), - # 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 + # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -84,7 +84,7 @@ try: { "type": "input_audio", "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "data": "{YOUR_AUDIO_URL}" } } ], @@ -130,7 +130,7 @@ const client = new OpenAI({ // 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey: "sk-xxx", apiKey: process.env.DASHSCOPE_API_KEY, - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 + // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", }); @@ -146,7 +146,7 @@ async function main() { { type: "input_audio", input_audio: { - data: "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + data: "{YOUR_AUDIO_URL}" } } ] @@ -157,11 +157,9 @@ async function main() { // stream_options: { // "include_usage": true // }, - extra_body: { - asr_options: { - // language: "zh", - enable_itn: false - } + asr_options: { + // language: "zh", + enable_itn: false } }); @@ -191,7 +189,7 @@ main(); #### cURL -以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 +以下为华北2（北京）地域的配置，调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)，各地域的配置不同。 ``` curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \ @@ -205,7 +203,7 @@ curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode { "type": "input_audio", "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "data": "{YOUR_AUDIO_URL}" } } ], @@ -245,7 +243,7 @@ curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode import base64, pathlib # input.mp3为用于声音复刻的本地音频文件，请替换为自己的音频文件路径，确保其符合音频要求 - file_path = pathlib.Path("input.mp3") + file_path = pathlib.Path("{YOUR_AUDIO_FILE}") base64_str = base64.b64encode(file_path.read_bytes()).decode() data_uri = f"data:audio/mpeg;base64,{base64_str}" ``` @@ -268,7 +266,7 @@ curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode // 使用示例 public static void main(String[] args) throws Exception { - System.out.println(toDataUrl("input.mp3")); + System.out.println(toDataUrl("{YOUR_AUDIO_FILE}")); } } ``` @@ -276,8 +274,6 @@ curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode #### Python SDK -示例中用到的音频文件为：[welcome.mp3](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260105/wotsae/welcome.mp3)。 - ``` import base64 from openai import OpenAI @@ -286,7 +282,7 @@ import pathlib try: # 请替换为实际的音频文件路径 - file_path = "welcome.mp3" + file_path = "{YOUR_AUDIO_FILE}" # 请替换为实际的音频文件MIME类型 audio_mime_type = "audio/mpeg" @@ -301,7 +297,7 @@ try: # 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key = "sk-xxx", api_key=os.getenv("DASHSCOPE_API_KEY"), - # 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 + # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -349,8 +345,6 @@ except Exception as e: #### Node.js SDK -示例中用到的音频文件为：[welcome.mp3](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260105/wotsae/welcome.mp3)。 - ``` // 运行前的准备工作: // Windows/Mac/Linux 通用: @@ -364,8 +358,8 @@ const client = new OpenAI({ // 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey: "sk-xxx", apiKey: process.env.DASHSCOPE_API_KEY, - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 + baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", }); const encodeAudioFile = (audioFilePath) => { @@ -374,7 +368,7 @@ const encodeAudioFile = (audioFilePath) => { }; // 请替换为实际的音频文件路径 -const dataUri = `data:audio/mpeg;base64,${encodeAudioFile("welcome.mp3")}`; +const dataUri = `data:audio/mpeg;base64,${encodeAudioFile("{YOUR_AUDIO_FILE}")}`; async function main() { try { @@ -399,11 +393,9 @@ async function main() { // stream_options: { // "include_usage": true // }, - extra_body: { - asr_options: { - // language: "zh", - enable_itn: false - } + asr_options: { + // language: "zh", + enable_itn: false } }); @@ -443,7 +435,7 @@ main(); System Message `_object_`（可选） -模型的目标或角色。如果设置系统消息，请放在messages列表的第一位。 +用于为语音识别提供上下文（Context），如背景文本和实体词表等参考信息，不支持设置模型角色等传统系统提示词。如果设置系统消息，请放在messages列表的第一位。 **属性** @@ -912,15 +904,7 @@ curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services { "content": [ { - "text": "" - } - ], - "role": "system" - }, - { - "content": [ - { - "audio": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "audio": "{YOUR_AUDIO_URL}" } ], "role": "user" @@ -961,7 +945,7 @@ public class Main { MultiModalMessage userMessage = MultiModalMessage.builder() .role(Role.USER.getValue()) .content(Arrays.asList( - Collections.singletonMap("audio", "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3"))) + Collections.singletonMap("audio", "{YOUR_AUDIO_URL}"))) .build(); Map asrOptions = new HashMap<>(); @@ -981,7 +965,7 @@ public class Main { } public static void main(String[] args) { try { - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 + // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; simpleMultiModalConversationCall(); } catch (ApiException | NoApiKeyException | UploadFileException e) { @@ -998,11 +982,11 @@ public class Main { import os import dashscope -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 +# 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' messages = [ - {"role": "user", "content": [{"audio": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3"}]} + {"role": "user", "content": [{"audio": "{YOUR_AUDIO_URL}"}]} ] response = dashscope.MultiModalConversation.call( @@ -1035,7 +1019,7 @@ print(response) System Message `_object_`（可选） -模型的目标或角色。如果设置系统消息，请放在messages列表的第一位。 +用于为语音识别提供上下文（Context），如背景文本和实体词表等参考信息，不支持设置模型角色等传统系统提示词。如果设置系统消息，请放在messages列表的第一位。 仅千问3-ASR-Flash支持该参数。 @@ -1460,7 +1444,7 @@ curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.c --data '{ "model": "qwen3-asr-flash-filetrans", "input": { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "file_url": "{YOUR_AUDIO_URL}" }, "parameters": { "channel_id":[ @@ -1498,7 +1482,7 @@ public class Main { { "model": "qwen3-asr-flash-filetrans", "input": { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "file_url": "{YOUR_AUDIO_URL}" }, "parameters": { "channel_id": [0], @@ -1514,7 +1498,7 @@ public class Main { { "model": "qwen3-asr-flash-filetrans", "input": { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "file_url": "{YOUR_AUDIO_URL}" }, "parameters": { "channel_id": [0], @@ -1595,7 +1579,7 @@ headers = { payload = { "model": "qwen3-asr-flash-filetrans", "input": { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "file_url": "{YOUR_AUDIO_URL}" }, "parameters": { "channel_id": [0], @@ -2010,6 +1994,14 @@ print(response.json()) + + + + + + + + 详情参见[异步调用识别结果说明](#2c27ad3e80p4y)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md index 9839c136..d5bc3a81 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md @@ -100,7 +100,7 @@ Authorization 鉴权在 WebSocket 握手阶段验证。如果 API Key 无效或 **启用方式：**配置客户端`[session.update](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#af43722339yva)`事件的`session.turn_detection`参数。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1592892871/CAEQaxiBgICW9b6H3RkiIGM5MDgwMTNkMjBjMDRlNTNiOGZlODNjZGJhNDQ3NGJm5812623_20251022102739.334.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7151064871/CAEQaxiBgICW9b6H3RkiIGM5MDgwMTNkMjBjMDRlNTNiOGZlODNjZGJhNDQ3NGJm5812623_20251022102739.334.svg) - 客户端通过发送`[input_audio_buffer.append](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#a42f8e9111n72)`事件将音频追加到缓冲区。 @@ -112,6 +112,10 @@ Authorization 鉴权在 WebSocket 握手阶段验证。如果 API Key 无效或 - 客户端在音频提交完后，发送`[session.finish](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#147ce70052d4z)`事件通知服务端结束当前会话。 + **警告** + + 在 VAD 模式下，推完音频后必须先发送`[session.finish](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#147ce70052d4z)`事件再关闭连接。如果客户端直接关闭 WebSocket 连接而未发送该事件，服务端将丢弃当前 in\_progress item，`conversation.item.input_audio_transcription.completed` 等事件将不会到达。建议在调用 `ws.Close()` 之前，先发送 `{"type":"session.finish"}`，并等待收到`[session.finished](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-server-events#6eaa77339djdv)`事件后再关闭连接。 + - 服务端在检测到语音结束时返回`[input_audio_buffer.speech_stopped](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-server-events#3d73b074cak7k)`事件。 - 服务端返回`[input_audio_buffer.committed](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-server-events#1108a3764an0e)`事件。 @@ -131,7 +135,7 @@ Authorization 鉴权在 WebSocket 握手阶段验证。如果 API Key 无效或 **启用方式：**将客户端`[session.update](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#af43722339yva)`事件的`session.turn_detection`设为null。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1592892871/CAEQaxiBgMDUp8qH3RkiIGEyYTc0NTI1ZmQ1OTQ5NjliNWE0OTYwYTAwMDBlMjBm5812623_20251022102739.334.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8151064871/CAEQaxiBgMDUp8qH3RkiIGEyYTc0NTI1ZmQ1OTQ5NjliNWE0OTYwYTAwMDBlMjBm5812623_20251022102739.334.svg) - 客户端通过发送`[input_audio_buffer.append](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#a42f8e9111n72)`事件将音频追加到缓冲区。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md index 3d4ec241..2c798bb3 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md @@ -277,7 +277,7 @@ API Key。建议使用时效性短、安全性更高的[临时API Key](https://h 语音合成所使用的音色。 -- **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -394,7 +394,7 @@ SSML 的使用限制（支持的模型、音色和接口），请参见[使用 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 > 时间戳结果在[INativeStreamInputTtsCallback](#secstreamcallback)的all\_response中。 @@ -440,7 +440,7 @@ cosyvoice-v1不支持该参数。 - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -460,9 +460,15 @@ cosyvoice-v1不支持该参数。 - vi：越南语 +- es：西班牙语 + - it：意大利语 -- ms：马来语 +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 `instruction` @@ -834,6 +840,7 @@ STREAM\_INPUT\_TTS\_EVENT\_TASK\_FAILED - 解压 ZIP 包。在 `app/libs` 目录中获取 AAR 格式 SDK，并添加到项目依赖。 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 + - 用 Android Studio 打开工程。示例代码位于`DashCosyVoiceStreamTtsActivity.java`，替换 API Key 后体验功能。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md index dabe4c85..69c32225 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md @@ -91,7 +91,7 @@ 语音合成所使用的音色。 -- **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -173,7 +173,7 @@ SSML 的使用限制（支持的模型、音色和接口），请参见[使用 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 **seed** `_integer_` （可选） @@ -209,7 +209,7 @@ cosyvoice-v1不支持该参数。 - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -229,9 +229,15 @@ cosyvoice-v1不支持该参数。 - vi：越南语 +- es：西班牙语 + - it：意大利语 -- ms：马来语 +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 **instruction** `_string_` （可选） @@ -365,7 +371,7 @@ qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v2、cosyvoice-v1 ## **finish-task** -**说明**：通知服务端文本发送完毕，请求结束任务。 +**说明**：通知服务端文本发送完毕，请求结束任务。如需取消当前轮次的语音合成任务，可在 `input` 中设置 `directive` 为 `cancel`。 **发送时机**：所有文本发送完毕后立即发送。 @@ -400,10 +406,41 @@ qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v2、cosyvoice-v1 } ``` +**取消任务示例**： + +``` +{ + "header": { + "action": "finish-task", + "task_id": "2bf83b9a-baeb-4fda-8d9a-xxxxxxxxxxxx", + "streaming": "duplex" + }, + "payload": { + "input": { + "directive": "cancel" + } + } +} +``` + **payload** `_object_` **（必选）** **属性** **input** `_object_` **（必选）** -固定为 `{}`。 +任务输入。为空对象 `{}` 时表示正常结束任务；包含 `directive` 时可用于取消当前轮次的语音合成任务。 + +**directive** `_string_` （可选） + +控制任务结束行为。当前仅支持取值为 `cancel`，表示取消当前轮次的语音合成任务，服务端会立即返回 `task-finished` 事件，且不会输出后续音频。 + +取消后，可在当前 WebSocket 连接上重新发起语音合成任务（发送新的 `run-task` 事件），无需重新建立连接。 + +**重要** + +**模型限制**： + +- 华北2（北京）地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型仅 v2 及以上版本支持该功能。 + +- 新加坡地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型不支持该功能。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md index 5a534022..6631da6e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md @@ -243,7 +243,7 @@ Qwen-Audio-TTS/CosyVoice 支持一次性输入和流式输入两种调用方式 语音合成所使用的音色。 - - **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) + - **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -360,7 +360,7 @@ Qwen-Audio-TTS/CosyVoice 支持一次性输入和流式输入两种调用方式 默认值：false。 - 仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 + 仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 > 时间戳结果在[onStreamInputTtsEventCallback](#bea29bbafcosq)的all\_response中。 @@ -406,7 +406,7 @@ Qwen-Audio-TTS/CosyVoice 支持一次性输入和流式输入两种调用方式 - zh：中文 - - en：英文 + - en：英语 - fr：法语 @@ -426,9 +426,15 @@ Qwen-Audio-TTS/CosyVoice 支持一次性输入和流式输入两种调用方式 - vi：越南语 + - es：西班牙语 + - it：意大利语 - - ms：马来语 + - ms：马来西亚语 + + - fil：菲律宾语 + + - ar：阿拉伯语 `instruction` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md index f3e4feda..d8dec276 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md @@ -121,6 +121,29 @@ public void streamingComplete() 结束双向流式调用，通知服务端所有文本已发送完毕。 +### **streamingCancel() - 取消双向流式调用** + +**方法签名**： + +``` +public void streamingCancel() +``` + +**说明**：取消当前轮次的双向流式语音合成任务。调用后，SDK 会立即结束当前任务。取消后可在当前连接上继续发起新的合成任务，无需重新初始化 `SpeechSynthesizer` 实例。 + +**重要** + +**版本要求**：使用该功能需要 Java SDK 版本不低于 2.22.26。 + +**重要** + +**模型限制**： + +- 华北2（北京）地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型仅 v2 及以上版本支持该功能。 + +- 新加坡地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型不支持该功能。 + + ### **callAsFlowable() - 单向流式合成（响应式）** **方法签名**： @@ -242,7 +265,7 @@ public long getFirstPackageDelay() ``` SpeechSynthesisParam param = SpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 模型 - .voice("longanlingxi") // 音色 + .voice("longanhuan_v3.6") // 音色 .format(SpeechSynthesisAudioFormat.WAV_8000HZ_MONO_16BIT) // 音频编码格式、采样率 .volume(50) // 音量，取值范围：[0, 100] .speechRate(1.0f) // 语速，取值范围：[0.5, 2] @@ -278,7 +301,7 @@ String 语音合成所使用的音色。 -- **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -345,7 +368,7 @@ boolean 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 `seed(int)` @@ -391,7 +414,7 @@ List - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -411,9 +434,15 @@ List - vi：越南语 +- es：西班牙语 + - it：意大利语 -- ms：马来语 +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 `instruction(String)` @@ -458,7 +487,7 @@ paramHotFix.setReplace(replaceItems); SpeechSynthesisParam param = SpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 模型 - .voice("your_voice") // 替换成qwen-audio-3.0-tts-flash复刻音色 + .voice("longanhuan_v3.6") // 音色 .hotFix(paramHotFix) .build(); ``` @@ -488,7 +517,7 @@ Map ``` SpeechSynthesisParam param = SpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_markdown_filter", true) .build(); ``` @@ -842,7 +871,7 @@ SDK提供了语音合成的关键接口，支持以下几种调用方式： ### **非流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1094204871/CAEQURiBgIDHpsn4phkiIDQ0ZGE2OTk3NmY5NTRhNDVhZDQwNWE3ZGZiMzk4Yjk54709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6538354871/CAEQURiBgIDHpsn4phkiIDQ0ZGE2OTk3NmY5NTRhNDVhZDQwNWE3ZGZiMzk4Yjk54709861_20241015153444.149.svg) 发送的文本长度不得超过20000字符。 @@ -864,7 +893,7 @@ public class Main { // 模型 private static String model = "qwen-audio-3.0-tts-flash"; // 音色 - private static String voice = "longanlingxi"; + private static String voice = "longanhuan_v3.6"; public static void streamAudioDataToSpeaker() { // 请求参数 @@ -917,7 +946,7 @@ public class Main { ### **单向流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1094204871/CAEQVRiBgMCfo..hrBkiIGEyMjNkZjVlMWZiYzRhZDU4ZjEyZjdjMmMzYjM1YzMz4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7538354871/CAEQVRiBgMCfo..hrBkiIGEyMjNkZjVlMWZiYzRhZDU4ZjEyZjdjMmMzYjM1YzMz4709861_20241015153444.149.svg) 发送的文本长度不得超过20000字符。 @@ -949,7 +978,7 @@ public class Main { // 模型 private static String model = "qwen-audio-3.0-tts-flash"; // 音色 - private static String voice = "longanlingxi"; + private static String voice = "longanhuan_v3.6"; public static void streamAudioDataToSpeaker() { CountDownLatch latch = new CountDownLatch(1); @@ -1024,7 +1053,7 @@ public class Main { ### 双向流式调用 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1094204871/CAEQVRiBgICHxPGhrBkiIGE3ZTVmMzY0YzI3NzQxYTFiYWE2MmU2NTBhMDgzZGM14709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7538354871/CAEQVRiBgICHxPGhrBkiIGE3ZTVmMzY0YzI3NzQxYTFiYWE2MmU2NTBhMDgzZGM14709861_20241015153444.149.svg) 单次发送文本长度不得超过 20000 字符，且累计发送文本总长度不得超过 20 万字符。 @@ -1076,7 +1105,7 @@ public class Main { "减少了用户等待时间。", "适用于调用大规模", "语言模型（LLM），以", "流式输入文本的方式", "进行语音合成的场景。"}; private static String model = "qwen-audio-3.0-tts-flash"; // 模型 - private static String voice = "longanlingxi"; // 音色 + private static String voice = "longanhuan_v3.6"; // 音色 public static void streamAudioDataToSpeaker() { // 配置回调函数 @@ -1177,7 +1206,7 @@ class TimeUtils { public class Main { private static String model = "qwen-audio-3.0-tts-flash"; // 模型 - private static String voice = "longanlingxi"; // 音色 + private static String voice = "longanhuan_v3.6"; // 音色 public static void streamAudioDataToSpeaker() throws NoApiKeyException { // 请求参数 @@ -1252,7 +1281,7 @@ public class Main { "减少了用户等待时间。", "适用于调用大规模", "语言模型（LLM），以", "流式输入文本的方式", "进行语音合成的场景。"}; private static String model = "qwen-audio-3.0-tts-flash"; - private static String voice = "longanlingxi"; + private static String voice = "longanhuan_v3.6"; public static void streamAudioDataToSpeaker() throws NoApiKeyException { // 模拟流式输入 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md index 8e94f9c5..4d00d326 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md @@ -125,6 +125,47 @@ def streaming_complete(self) -> None **说明**：通知服务端所有文本已发送完毕，阻塞当前线程直到剩余文本合成完成并返回所有音频数据。未调用此方法可能导致尾部文本无法转换为语音。 +### **streaming\_cancel() - 取消流式合成** + +**方法签名**： + +``` +def streaming_cancel(self, complete_timeout_millis: int = 10000) -> None +``` + +**参数说明**： + +**参数** + +**类型** + +**必填** + +**说明** + +complete\_timeout\_millis + +int + +否 + +等待服务端返回 task-finished 事件的超时时间，单位毫秒。默认值：10000。 + +**说明**：取消当前轮次的流式语音合成任务。调用后，SDK 会立即结束当前任务。取消后可在当前连接上继续发起新的合成任务，无需重新初始化 `SpeechSynthesizer` 实例。 + +**重要** + +**版本要求**：使用该功能需要 Python SDK 版本不低于 1.26.4。 + +**重要** + +**模型限制**： + +- 华北2（北京）地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型仅 v2 及以上版本支持该功能。 + +- 新加坡地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型不支持该功能。 + + ### **get\_last\_request\_id() - 获取请求ID** **方法签名**： @@ -185,7 +226,7 @@ str 语音合成所使用的音色。 -- **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -263,7 +304,7 @@ int ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="longanlingxi", + voice="longanhuan_v3.6", additional_params={"bit_rate": 128000} ) ``` @@ -278,7 +319,7 @@ bool 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 **说明** @@ -287,7 +328,7 @@ bool ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="your_voice", + voice="longanhuan_v3.6", additional_params={"word_timestamp_enabled": True} ) ``` @@ -334,7 +375,7 @@ list\[str\] - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -354,9 +395,15 @@ list\[str\] - vi：越南语 +- es：西班牙语 + - it：意大利语 -- ms：马来语 +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 instruction @@ -388,7 +435,7 @@ bool ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="longanlingxi", + voice="longanhuan_v3.6", additional_params={ "enable_aigc_tag": True, "aigc_propagator": "your_propagator", @@ -447,7 +494,7 @@ qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v2、cosyvoice-v1 ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="your_voice", # 替换成qwen-audio-3.0-tts-flash复刻音色 + voice="longanhuan_v3.6", # 音色 hot_fix={ "pronunciation": [{"天气": "tian1 qi4"}], "replace": [{"今天": "金天"}] @@ -483,7 +530,7 @@ bool ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="your_voice", # 替换成qwen-audio-3.0-tts-flash复刻音色 + voice="longanhuan_v3.6", # 音色 additional_params={"enable_markdown_filter": True} ) ``` @@ -693,7 +740,7 @@ SDK提供了语音合成的关键接口，支持以下几种调用方式： ### **非流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1194204871/CAEQURiBgMDRr9T4phkiIGNmYzBiZjFkZjQ4MDQzZGU4NDIyZDU2NWJjYjkyZTQ04709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0658354871/CAEQURiBgMDRr9T4phkiIGNmYzBiZjFkZjQ4MDQzZGU4NDIyZDU2NWJjYjkyZTQ04709861_20241015153444.149.svg) 单次调用发送的文本长度不得超过20000字符，超出限制将返回错误。 @@ -718,7 +765,7 @@ dashscope.base_websocket_api_url='wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.c # 模型 model = "qwen-audio-3.0-tts-flash" # 音色 -voice = "longanlingxi" +voice = "longanhuan_v3.6" # 实例化SpeechSynthesizer，并在构造方法中传入模型（model）、音色（voice）等请求参数 synthesizer = SpeechSynthesizer(model=model, voice=voice) @@ -736,7 +783,7 @@ with open('output.mp3', 'wb') as f: ### **单向流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1194204871/CAEQVRiBgIDv9fShrBkiIDhmNTk5YmQ1ZDgwNzRjZjRiN2VlMTU5YzI1ZGMwMTlm4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0658354871/CAEQVRiBgIDv9fShrBkiIDhmNTk5YmQ1ZDgwNzRjZjRiN2VlMTU5YzI1ZGMwMTlm4709861_20241015153444.149.svg) 单次调用发送的文本长度不得超过20000字符，超出限制将返回错误。 @@ -769,7 +816,7 @@ dashscope.base_websocket_api_url='wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.c # 模型 model = "qwen-audio-3.0-tts-flash" # 音色 -voice = "longanlingxi" +voice = "longanhuan_v3.6" # 定义回调接口 class Callback(ResultCallback): @@ -823,7 +870,7 @@ synthesizer.call("今天天气怎么样？") ### **双向流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1194204871/CAEQVRiBgMDb7PahrBkiIDVkNjEwOTMxYjEwOTRmOWFhMmI1OTRiY2Q3ZDgzZmE54709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0658354871/CAEQVRiBgMDb7PahrBkiIDVkNjEwOTMxYjEwOTRmOWFhMmI1OTRiY2Q3ZDgzZmE54709861_20241015153444.149.svg) 单次发送文本长度不得超过 20000 字符，且累计发送文本总长度不得超过 20 万字符。 @@ -888,7 +935,7 @@ dashscope.base_websocket_api_url='wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.c # 模型 model = "qwen-audio-3.0-tts-flash" # 音色 -voice = "longanlingxi" +voice = "longanhuan_v3.6" # 定义回调接口 class Callback(ResultCallback): diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md index 90389819..56124c1b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md @@ -87,7 +87,7 @@ Authorization 鉴权在 WebSocket 握手阶段验证。如果 API Key 无效或 ## 交互流程 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8684204871/CAEQaxiBgID50pCW3hkiIDVlOWNkODdhOGYyYjQ2ZDFiMzgyYjNmMmUzOGZkNGVh4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9336814871/CAEQaxiBgID50pCW3hkiIDVlOWNkODdhOGYyYjQ2ZDFiMzgyYjNmMmUzOGZkNGVh4709861_20241015153444.149.svg) 客户端事件和服务端事件的详细说明，请参见[客户端事件](https://help.aliyun.com/zh/model-studio/cosyvoice-client-events)和[服务端事件](https://help.aliyun.com/zh/model-studio/cosyvoice-server-events)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md index e78e4fb1..a25c2588 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md @@ -66,7 +66,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ "model": "qwen-audio-3.0-tts-flash", "input": { "text": "我家的后面有一个很大的花园。", - "voice": "longanlingxi", + "voice": "longanhuan_v3.6", "format": "wav", "sample_rate": 24000 } @@ -84,7 +84,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ "model": "qwen-audio-3.0-tts-flash", "input": { "text": "我家的后面有一个很大的花园。", - "voice": "longanlingxi", + "voice": "longanhuan_v3.6", "format": "wav", "sample_rate": 24000 } @@ -135,7 +135,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ 取值范围： -- 系统音色：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- 系统音色：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - 声音复刻音色：如何创建音色请参见[CosyVoice声音复刻/设计API](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api) @@ -211,7 +211,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 **seed** `_integer_` （可选） @@ -245,7 +245,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -265,6 +265,16 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ - vi：越南语 +- es：西班牙语 + +- it：意大利语 + +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 + **instruction** `_string_` （可选） diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md index 7e77c178..1c261549 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md @@ -207,7 +207,7 @@ String 取值范围： -- 系统音色：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- 系统音色：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - 声音复刻音色：如何创建音色请参见[CosyVoice声音复刻/设计API](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api) @@ -301,7 +301,7 @@ boolean HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("你好") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_ssml", true) .build(); ``` @@ -312,7 +312,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("你好") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("enable_ssml", true)) .build(); ``` @@ -327,7 +327,7 @@ boolean 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 **说明** @@ -339,7 +339,7 @@ boolean HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("word_timestamp_enabled", true) .build(); ``` @@ -350,7 +350,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("word_timestamp_enabled", true)) .build(); ``` @@ -377,7 +377,7 @@ int HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("seed", 1234) .build(); ``` @@ -388,7 +388,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("seed", 1234)) .build(); ``` @@ -421,7 +421,7 @@ List - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -441,6 +441,16 @@ List - vi：越南语 +- es：西班牙语 + +- it：意大利语 + +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 + **说明** @@ -452,7 +462,7 @@ List HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("language_hints", Arrays.asList("zh")) .build(); ``` @@ -463,7 +473,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("language_hints", Arrays.asList("zh"))) .build(); ``` @@ -488,7 +498,7 @@ String HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("instruction", "请用非常开心的语气说话。") .build(); ``` @@ -499,7 +509,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("instruction", "请用非常开心的语气说话。")) .build(); ``` @@ -530,7 +540,7 @@ int HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .format("opus") .parameter("bit_rate", 32) .build(); @@ -542,7 +552,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .format("opus") .parameters(Collections.singletonMap("bit_rate", 32)) .build(); @@ -570,7 +580,7 @@ boolean HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_aigc_tag", true) .build(); ``` @@ -581,7 +591,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("enable_aigc_tag", true)) .build(); ``` @@ -608,7 +618,7 @@ String HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_aigc_tag", true) .parameter("aigc_propagator", "xxxx") .build(); @@ -624,7 +634,7 @@ map.put("aigc_propagator", "xxxx"); HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(map) .build(); ``` @@ -651,7 +661,7 @@ String HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_aigc_tag", true) .parameter("aigc_propagate_id", "xxxx") .build(); @@ -667,7 +677,7 @@ map.put("aigc_propagate_id", "xxxx"); HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(map) .build(); ``` @@ -726,7 +736,7 @@ hotFix.put("replace", replace); HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("今天天气真好。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("hot_fix", hotFix) .build(); ``` @@ -738,7 +748,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("今天天气真好。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("hot_fix", hotFix)) .build(); ``` @@ -774,7 +784,7 @@ boolean HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("# 标题\n正文内容") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_markdown_filter", true) .build(); ``` @@ -785,7 +795,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("# 标题\n正文内容") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("enable_markdown_filter", true)) .build(); ``` @@ -837,7 +847,7 @@ public class CosyVoiceSyncExample { HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 更换模型时，需同步更换为对应版本的音色 .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 + .voice("longanhuan_v3.6") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 .format("wav") .sampleRate(24000) // 未配置环境变量时，将下行替换为：apiKey("sk-xxx")，即替换为实际的API Key @@ -877,7 +887,7 @@ public class CosyVoiceSyncExample { HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 更换模型时，需同步更换为对应版本的音色 .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 + .voice("longanhuan_v3.6") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 .format("wav") .sampleRate(24000) // 未配置环境变量时，将下行替换为：apiKey("sk-xxx")，即替换为实际的API Key @@ -938,7 +948,7 @@ public class CosyVoiceStreamExample { HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 更换模型时，需同步更换为对应版本的音色 .text("今天天气真好，适合出去玩。") - .voice("longanlingxi") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 + .voice("longanhuan_v3.6") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 .format("wav") .sampleRate(24000) // 未配置环境变量时，将下行替换为：apiKey("sk-xxx")，即替换为实际的API Key diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md index 7143911a..a3f50903 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md @@ -109,7 +109,7 @@ str 取值范围： -- 系统音色：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- 系统音色：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - 声音复刻音色：如何创建音色请参见[CosyVoice声音复刻/设计API](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api) @@ -222,7 +222,7 @@ bool - False：关闭。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 seed @@ -264,7 +264,7 @@ list - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -284,6 +284,16 @@ list - vi：越南语 +- es：西班牙语 + +- it：意大利语 + +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 + instruction @@ -443,7 +453,7 @@ dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co result = HttpSpeechSynthesizer.call( model="qwen-audio-3.0-tts-flash", # 更换模型时，需同步更换为对应版本的音色 text="今天是个好日子，适合构建人们喜爱的产品！", - voice="longanhuan", # 该音色适用于cosyvoice-v3系列，cosyvoice-v2请使用longxiaochun_v2等v2音色 + voice="longanhuan_v3.6", # 该音色适用于cosyvoice-v3系列，cosyvoice-v2请使用longxiaochun_v2等v2音色 format="wav", sample_rate=24000, stream=False, @@ -478,7 +488,7 @@ dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co stream_result = HttpSpeechSynthesizer.call( model="qwen-audio-3.0-tts-flash", # 更换模型时，需同步更换为对应版本的音色 text="今天是个好日子，适合构建人们喜爱的产品！", - voice="longanhuan", # 该音色适用于cosyvoice-v3系列，cosyvoice-v2请使用longxiaochun_v2等v2音色 + voice="longanhuan_v3.6", # 该音色适用于cosyvoice-v3系列，cosyvoice-v2请使用longxiaochun_v2等v2音色 format="wav", sample_rate=24000, stream=True, diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md index 18e9d458..37905ab0 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md @@ -63,7 +63,7 @@ public class Main { .apiKey(System.getenv("DASHSCOPE_API_KEY")) .text("Today is a wonderful day to build something people love!") .voice(AudioParameters.Voice.CHERRY) - .languageType("English") + .parameter("language_type", "English") // 如需使用指令控制功能，请取消下方注释，并将model替换为qwen3-tts-instruct-flash // .parameter("instructions","语速较快，带有明显的上扬语调，适合介绍时尚产品。") // .parameter("optimize_instructions",true) @@ -166,7 +166,7 @@ public class Main { .apiKey(System.getenv("DASHSCOPE_API_KEY")) .text("Today is a wonderful day to build something people love!") .voice(AudioParameters.Voice.CHERRY) - .languageType("English") + .parameter("language_type", "English") // 如需使用指令控制功能，请取消下方注释，并将model替换为qwen3-tts-instruct-flash // .parameter("instructions","语速较快，带有明显的上扬语调，适合介绍时尚产品。") // .parameter("optimize_instructions",true) @@ -270,7 +270,7 @@ curl -X POST 'https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-gen 支持语言：仅支持中文和英文。 -适用范围：该功能仅适用于千问3-TTS-Instruct-Flash-Realtime系列模型。 +适用范围：该功能仅适用于千问3-TTS-Instruct-Flash系列模型。 **optimize\_instructions** `_boolean_` （可选） @@ -363,140 +363,6 @@ HTTP状态码。遵循 [RFC 9110](https://www.rfc-editor.org/rfc/rfc9110.html#n - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - **request\_id** `_string_` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md index 2673fe86..a2402160 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md @@ -181,7 +181,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -189,11 +189,11 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ 取值范围（因模型而异）： -- qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： - zh：中文 - - en：英文 + - en：英语 - fr：法语 @@ -215,7 +215,13 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ - it：意大利语 - - ms：马来语 + - es：西班牙语 + + - ms：马来西亚语 + + - fil：菲律宾语 + + - ar：阿拉伯语 - cosyvoice-v3-plus： @@ -297,7 +303,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -307,7 +313,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md index be92e944..36aa6d87 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md @@ -121,7 +121,7 @@ customParam 否 -自定义参数，可指定languageHints、maxPromptAudioLength等。 +自定义参数，可通过 parameter() 方法指定 language\_hints、max\_prompt\_audio\_length 等参数。 **返回值**：`Voice` 对象，通过 `getVoiceId()` 方法获取音色ID。 @@ -283,108 +283,11 @@ String 声音复刻模型，固定为"voice-enrollment"。 -languageHints(List) - -List - -**重要** - -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 - -辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 - -此参数为数组，但当前版本仅处理第一个元素。 - -取值范围（因模型而异）： - -- qwen-audio-3.0-tts-flash： - - - zh：中文 - - - en：英文 - - - fr：法语 - - - de：德语 - - - ja：日语 - - - ko：韩语 - - - ru：俄语 - - - pt：葡萄牙语 - - - th：泰语 - - - id：印尼语 - - - vi：越南语 - - - it：意大利语 - - - ms：马来语 - -- cosyvoice-v3-plus： - - - zh：中文 - - - en：英文 - - - fr：法语 - - - de：德语 - - - ja：日语 - - - ko：韩语 - - - ru：俄语 - -- cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash： - - - zh：中文 - - - en：英文 - - - fr：法语 - - - de：德语 - - - ja：日语 - - - ko：韩语 - - - ru：俄语 - - - pt：葡萄牙语 - - - th：泰语 - - - id：印尼语 - - - vi：越南语 - - -默认值：\["zh"\]。 - -maxPromptAudioLength(Float) - -Float - -**重要** - -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 - -音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 - -默认值：10.0。 - parameter(String, Object) Object -设置[扩展参数](#a18b66cba924j)，如 parameter("enable\_preprocess", false)。 +设置自定义参数，如 parameter("language\_hints", Arrays.asList("zh"))、parameter("max\_prompt\_audio\_length", 10.0f)、parameter("enable\_preprocess", false)。 ### **扩展参数** @@ -404,7 +307,7 @@ boolean **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 @@ -422,7 +325,7 @@ import com.alibaba.dashscope.utils.Constants; import org.slf4j.Logger; import org.slf4j.LoggerFactory; -import java.util.Collections; +import java.util.Arrays; public class Main { private static final Logger logger = LoggerFactory.getLogger(Main.class); @@ -444,8 +347,8 @@ public class Main { fileUrl, VoiceEnrollmentParam.builder() .model(cloneModelName) - .languageHints(Collections.singletonList("zh")) - // .maxPromptAudioLength(10.0f) + .parameter("language_hints", Arrays.asList("zh")) + // .parameter("max_prompt_audio_length", 10.0f) // .parameter("enable_preprocess", false) .build()); diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md index 14b73cd2..42410c27 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md @@ -112,7 +112,7 @@ List\[str\] **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -120,11 +120,11 @@ List\[str\] 取值范围（因模型而异）： -- qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： - zh：中文 - - en：英文 + - en：英语 - fr：法语 @@ -146,7 +146,13 @@ List\[str\] - it：意大利语 - - ms：马来语 + - es：西班牙语 + + - ms：马来西亚语 + + - fil：菲律宾语 + + - ar：阿拉伯语 - cosyvoice-v3-plus： @@ -199,7 +205,7 @@ float **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -213,7 +219,7 @@ bool **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md index 80482553..7dfaaf7e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md @@ -2,7 +2,7 @@ Qwen-Audio Realtime API的客户端事件参考。 -**用户指南**：[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides)。如需了解事件交互时序，请参见[WebSocket API](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-websocket-api)。 +**用户指南**：[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides)。如需了解事件交互时序，请参见[WebSocket API](https://help.aliyun.com/zh/model-studio/fun-audiochat-realtime-websocket-api)。 ## **session.update** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-realtime-websocket-api.md similarity index 67% rename from skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md rename to skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-realtime-websocket-api.md index 25294168..75868574 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-realtime-websocket-api.md @@ -127,7 +127,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 下图展示了 server\_vad 模式下的典型交互时序： -服务端 客户端 服务端 客户端 会话初始化 语音输入 loop \[用户说话中\] loop \[用户说话中\] 响应生成 loop \[流式输出\] loop \[多轮交互\] connect session.created session.update session.updated input\_audio\_buffer.append input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.speech\_stopped conversation.item.input\_audio\_transcription.completed commit audio buffer input\_audio\_buffer.committed conversation.item.created response.created response.output\_item.added conversation.item.created response.content\_part.added response.audio\_transcript.delta response.audio.delta response.audio\_transcript.done response.audio.done response.content\_part.done response.output\_item.done response.done +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088432.svg) 按时间顺序，客户端与服务端的交互流程如下： @@ -148,7 +148,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 模型播报期间，若 VAD 检测到用户开始说话，服务端会取消当前响应（返回 `response.done`，状态为 `cancelled`），随后开始新一轮语音输入和响应。下图展示了用户打断的交互时序： -服务端 客户端 服务端 客户端 服务端正在流式输出 loop \[流式输出中\] 用户打断 新一轮语音输入 loop \[用户说话中\] 新一轮推理 response.audio.delta input\_audio\_buffer.append（用户开始说话） response.done（status=cancelled） input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.speech\_stopped conversation.item.input\_audio\_transcription.completed commit audio buffer input\_audio\_buffer.committed conversation.item.created response.created +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088435.svg) ### **smart\_turn 模式** @@ -160,7 +160,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 下图展示了 smart\_turn 模式下的典型交互时序： -服务端 客户端 服务端 客户端 会话初始化 loop \[用户说话中\] 无效语音（可能出现 0~N 次） loop \[用户说话中\] loop \[用户说话中\] 响应生成 loop \[流式输出\] loop \[多轮交互\] connect session.created session.update session.updated input\_audio\_buffer.append conversation.item.ambient\_audio\_transcription.delta conversation.item.ambient\_audio\_transcription.completed input\_audio\_buffer.append input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.speech\_stopped conversation.item.input\_audio\_transcription.completed commit audio buffer input\_audio\_buffer.committed conversation.item.created response.created response.output\_item.added conversation.item.created response.content\_part.added response.audio\_transcript.delta response.audio.delta response.audio\_transcript.done response.audio.done response.content\_part.done response.output\_item.done response.done +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088441.svg) 与 server\_vad 模式的主要区别： @@ -175,13 +175,13 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 与 server\_vad 模式的打断处理基本一致。下图展示了用户打断的交互时序： -服务端 客户端 服务端 客户端 服务端正在流式输出 loop \[流式输出中\] 用户打断 新一轮语音输入 loop \[用户说话中\] 新一轮推理 response.audio.delta input\_audio\_buffer.append（用户开始说话） response.done（status=cancelled） input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta conversation.item.input\_audio\_transcription.completed input\_audio\_buffer.speech\_stopped commit audio buffer input\_audio\_buffer.committed conversation.item.created response.created +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088443.svg) ## **无效轮次** 已判定有效的语音可能被撤回（`input_audio_buffer.speech_stopped` 返回 `reason=turn_invalid`），此时不触发推理，客户端应继续发送音频等待下一轮有效语音。下图展示了无效轮次的交互时序： -服务端 客户端 服务端 客户端 loop \[用户说话中\] loop \[用户说话中\] 轮次无效 继续发送音频，等待下一轮 input\_audio\_buffer.append input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.speech\_stopped（reason=turn\_invalid） +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088444.svg) ### **说话人增强配置流程** @@ -216,7 +216,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 下图展示了 push-to-talk 模式下的典型交互时序： -服务端 客户端 服务端 客户端 会话初始化 语音输入 loop \[用户说话中\] 响应生成 loop \[流式输出\] loop \[多轮交互\] connect session.created session.update session.updated input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.commit（用户松开按键） conversation.item.input\_audio\_transcription.completed input\_audio\_buffer.committed conversation.item.created response.create（手动触发推理） response.created response.output\_item.added conversation.item.created response.content\_part.added response.audio\_transcript.delta response.audio.delta response.audio\_transcript.done response.audio.done response.content\_part.done response.output\_item.done response.done +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088447.svg) 按时间顺序，客户端与服务端的交互流程如下： @@ -233,7 +233,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 客户端发送 `response.cancel` 取消当前响应，服务端返回 `response.done`（状态为 `cancelled`，原因为 `client_cancelled`）。下图展示了用户打断的交互时序： -服务端 客户端 服务端 客户端 服务端正在流式输出 loop \[流式输出中\] 用户打断 新一轮语音输入 loop \[用户说话中\] 新一轮推理 response.audio\_transcript.delta response.audio.delta response.cancel response.done（status=cancelled, reason=client\_cancelled） input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.commit conversation.item.input\_audio\_transcription.completed input\_audio\_buffer.committed conversation.item.created response.create response.created +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088449.svg) ## **各模式操作约束** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md index d74b004c..5cf16168 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md @@ -2,7 +2,7 @@ Qwen-Audio Realtime API 的服务端事件参考。所有服务端事件均包含 `event_id`（服务端自动生成）和 `type`（事件类型）公共字段。 -**用户指南**：[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides)。如需了解事件交互时序，请参见[WebSocket API](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-websocket-api)。 +**用户指南**：[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides)。如需了解事件交互时序，请参见[WebSocket API](https://help.aliyun.com/zh/model-studio/fun-audiochat-realtime-websocket-api)。 ## **error** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md index 0a687e48..96d7ae74 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md @@ -80,7 +80,6 @@ curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.c "generate_mode":"generate", "generate_num":1 }, - "auxiliary_parameters": "WMq4SC4......", "parameters":{} }' ``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md index 720eeeec..825a4dde 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md @@ -71,14 +71,16 @@ wanx-background-generation-v2 ### **步骤1：创建任务获取任务ID** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/background-generation/generation/` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation` #### **请求头（Headers）** ## 图像背景生成 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/background-generation/generation/' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -396,7 +398,7 @@ foreground\_edge图像列表和background\_edge图像列表之和不得超过10 ### **步骤2：根据任务ID查询结果** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` #### **请求头（Headers）** @@ -404,10 +406,10 @@ foreground\_edge图像列表和background\_edge图像列表之和不得超过10 请将`86ecf553-d340-4e21-xxxxxxxxx`替换为真实的task\_id。 -> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中WorkspaceId需替换为真实的业务空间ID。 +> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中{WorkspaceId}需替换为真实的业务空间ID。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md new file mode 100644 index 00000000..1537ef67 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md @@ -0,0 +1,459 @@ +# 千问-图像生成与编辑3.0 API参考 + +千问-图像生成与编辑3.0模型同时支持文生图（T2I）和图生图/图像编辑（I2I），可根据文本提示词直接生成图像，也可基于1-3张参考图结合编辑指令进行精确编辑。 + +**重要** + +该模型目前处于邀测阶段，您需要前往模型广场申请开通后方可使用。 + +## **模型概览** + +**模型名称** + +**模型简介** + +**输出图像规格** + +qwen-image-3.0-pro + +千问图像生成与编辑3.0模型，同时支持文生图（T2I）和图生图/图像编辑（I2I）。 + +图像分辨率： + +- **文生图（T2I）**：总像素需在512\*512至2048\*2048之间。 + +- **图生图（I2I）**：总像素需在512\*512至2048\*2048之间。 + +- **默认**：不指定`size`时，模型根据提示词自动推荐分辨率。 + + +图像格式：png + +## **前提条件** + +在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + +如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。目前，该SDK已支持Python和Java。 + +**重要** + +华北2（北京）和新加坡地域拥有独立的 **API Key** 与**请求地址**，不可混用，跨地域调用将导致鉴权失败或服务报错。 + +**重要** + +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： + +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 + +## HTTP调用 + +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +**新加坡地域**：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +#### 请求参数 + +## 文生图（T2I） + +``` +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ +--header 'Content-Type: application/json' \ +--header "Authorization: Bearer $DASHSCOPE_API_KEY" \ +--data '{ + "model": "qwen-image-3.0-pro", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "text": "画面是一张竖幅户外人像摄影，整体从上到下呈现温暖的午后街景氛围。顶部左侧到上方大面积被深绿色藤蔓和橙色小花覆盖，花叶从建筑檐口自然垂落，受阳光照射的叶片呈黄绿色高光，阴影处则偏深绿，形成浓密而柔和的背景层次。左上至中上区域是一块深蓝色横向招牌，招牌表面较暗、略带磨砂质感，上面以白色哥特体大字写着 Il Messaggero，文字位于画面左侧偏上，部分被前景花叶轻微遮挡，字体高对比、带装饰性尖角和粗细变化。招牌下方是报刊亭或书报摊的玻璃展示窗，黑色金属框架将橱窗分隔成多个矩形区域，内部陈列着许多报纸、杂志和书刊封面，但大多因景深虚化和光线反射而难以辨读，形成浅色纸张与深色边框交错的背景纹理。画面右上方是强烈的逆光区域，阳光从街道尽头照入，背景建筑被虚化成米灰色块面，边缘柔和，呈现明显的浅景深效果。画面中部偏右是一名年轻成年女性的半身至膝上人像，她回头面向镜头微笑，身体略向右转，肩背朝向观者，姿态自然放松。她有长而浓密的黑色波浪卷发，发丝被逆光勾勒出金色轮廓光，发梢在右侧向外散开，显得轻盈蓬松。她肤色白皙，脸型柔和偏鹅蛋形，眉形细致，眼睛明亮，眼妆清透，睫毛明显，面部带有自然高光，唇部为柔和珊瑚红色，笑容露齿，表情亲切明朗。她佩戴小巧耳饰，身穿黑色细肩带露背连衣裙，面料颜色深黑、轮廓简洁，细肩带从肩部向背部延伸，背部线条清晰。画面下部偏左到中部，她双手抱着一束玫瑰花，花束体积较大，主要由橙色、杏色、粉色和浅桃色玫瑰组成，花瓣层层卷曲，边缘被阳光照亮，绿色叶片和长花茎从花束下方垂出，花束与黑色裙装形成鲜明色彩对比。右侧背景是一条被阳光照亮的城市街道，地面呈暖灰与金黄色调，远处建筑、街边设施和一个模糊的红色圆形交通标志位于右下远景，均因焦外虚化而只保留色块和轮廓。整张照片采用暖色胶片感处理，带有细腻颗粒、柔和对比和明显逆光边缘光，人物位于视觉焦点，背景报刊亭、花藤、街道和阳光共同营造出浪漫、明亮、都市漫步式的氛围。" + } + ] + } + ] + }, + "parameters": { + "prompt_extend": true + } +}' +``` + +## 图生图/图像编辑（I2I） + +``` +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ +--header 'Content-Type: application/json' \ +--header "Authorization: Bearer $DASHSCOPE_API_KEY" \ +--data '{ + "model": "qwen-image-3.0-pro", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "image": "https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/yBRq1ZPYEaXdyOdv/img/33a80a19-7ac7-4c64-b0fa-7d685b7046a0.png" + }, + { + "text": "帮我生成一张充满高级感的都市风格女性写真，画面中人物完美保留输入图片中这位年轻女性的面部特征与一头柔顺的黑色长发。人物脱下原本的米色针织上衣，换上一套彰显高雅气质的都市职场穿搭，身穿一件质感垂顺的香槟色真丝衬衫，外搭一件剪裁利落的深灰色休闲西装外套，下身搭配同色系的高腰阔腿裤，整体造型既干练又富有女人味。场景设定在一家装修现代简约的高端咖啡店内，背景是通透的落地玻璃窗，窗外隐约可见繁华的城市街景，室内摆放着深色实木长桌和舒适的皮质座椅，桌面上放置着一台打开的银色笔记本电脑、一份文件和一杯热气腾腾的美式咖啡。人物呈现出慵懒而放松的办公姿态，身体微微后仰倚靠在椅背上，一只手臂自然搭在扶手上，另一只手轻轻握着咖啡杯置于桌边，头部微侧，眼神清澈从容且带有一丝慵懒地直视镜头，嘴角挂着一抹优雅自信的微笑。人物化着精致得体的正式场合妆容，底妆清透干净，眉眼线条清晰利落，唇部涂抹着显气色的豆沙色口红，展现出成熟知性的魅力。光线采用午后柔和的自然光，从侧面透过落地窗洒入，在人物的面部轮廓和衣物褶皱上留下细腻的光影过渡，背景呈现自然的景深虚化效果，色彩以大地色、灰色和暖白色为主调，营造出宁静、高级且充满故事感的都市办公氛围，构图采用经典的竖幅七分身人像视角，人物位于画面视觉中心略偏右，比例协调，画质清晰细腻。" + } + ] + } + ] + }, + "parameters": { + "prompt_extend": true + } +}' +``` + +##### 请求头（Headers） + +**Content-Type** `_string_` **（必选）** + +请求内容类型。此参数必须设置为`application/json`。 + +**Authorization** `_string_`**（必选）** + +请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 + +##### 请求体（Request Body） + +**model** `_string_` **（必选）** + +模型名称，当前可用模型为`qwen-image-3.0-pro`。 + +**input** `_object_` **（必选）** + +输入参数对象，包含以下字段： + +**属性** + +**messages** `_array_` **（必选）** + +请求内容数组。**当前仅支持单轮对话**，因此数组内**有且只有一个对象**，该对象包含`role`和`content`两个属性。 + +**属性** + +**role** `_string_` **（必选）** + +消息发送者角色，必须设置为`user`。 + +**content** `_array_` **（必选）** + +消息内容数组，根据使用场景有不同的组合方式： + +- **文生图（T2I）**：仅包含一个`{"text": "..."}`对象。 + +- **图生图（I2I）**：包含1-3个`{"image": "..."}`对象和1个`{"text": "..."}`对象。 + + +**属性** + +**image** `_string_` （可选） + +输入图像的 URL 或 Base64 编码数据。I2I场景下支持传入1-3张图像。多图输入时，按照数组顺序定义图像顺序。 + +**图像要求：** + +- 图像格式：JPG、JPEG、PNG、BMP、TIFF、WEBP和GIF。 + +- 图像分辨率：建议图像的宽和高均在384像素至2048像素之间。 + +- 图像大小：不超过10MB。 + + +**支持的输入格式** + +1. 公网URL：支持 HTTP 和 HTTPS 协议。您也可在此[获取临时公网URL](https://help.aliyun.com/zh/model-studio/get-temporary-file-url)。 + +2. Base64 编码：格式为`data:{MIME_type};base64,{base64_data}`。 + + +**text** `_string_` **（必选）** + +正向提示词，用于描述您期望生成或编辑的图像内容、风格和构图。支持中英文。 + +**注意**：仅支持传入一个text，不传或传入多个将报错。 + +**parameters** `_object_` （可选） + +控制图像生成的附加参数。 + +**属性** + +**prompt\_extend** `_boolean_` （可选） + +是否开启提示词智能改写，默认值为 `true`（建议开启）。开启后，模型会优化正向提示词，对描述较简单的提示词效果提升明显。 + +**n** `_integer_` （可选） + +输出图像的数量，支持输出1-6张图片，默认值为1。 + +**size** `_string_` （可选） + +设置输出图像的分辨率，格式为`宽*高`，例如`"1024*1024"`。未指定时由模型根据提示词自动推荐分辨率。 + +- **文生图（T2I）**：像素范围512\*512至2048\*2048。 + +- **图生图（I2I）**：像素范围512\*512至2048\*2048。 + + +**negative\_prompt** `_string_` （可选） + +反向提示词，用来描述不希望在画面中看到的内容，可以对画面进行限制。 + +**seed** `_integer_` （可选） + +随机数种子，取值范围`[0, 2147483647]`。固定种子可使生成结果相对稳定。 + +**watermark** `_boolean_` （可选） + +是否添加水印，默认值为 `false`。 + +#### 响应参数 + +## 任务执行成功 + +任务数据（如任务状态、图像URL等）仅保留24小时，超时后会被自动清除。请您务必及时保存生成的图像。 + +``` +{ + "output": { + "choices": [ + { + "finish_reason": "stop", + "message": { + "content": [ + { + "image": "https://dashscope-result-sz.oss-cn-shenzhen.aliyuncs.com/xxx.png?Expires=xxx" + } + ], + "role": "assistant" + } + } + ] + }, + "usage": { + "width": 1024, + "height": 1024, + "image_count": 1 + }, + "request_id": "571ae02f-5c9d-436c-83c2-f221e6df0xxx" +} +``` + +## 任务执行异常 + +如果因为某种原因导致任务执行失败，将返回相关信息，可以通过code和message字段明确指示错误原因。请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 + +``` +{ + "request_id": "31f808fd-8eef-9004-xxxxx", + "code": "InvalidApiKey", + "message": "Invalid API-key provided." +} +``` + +**output** `_object_` + +包含模型生成结果。 + +**属性** + +**choices** `_array_` + +结果选项列表。 + +**属性** + +**finish\_reason** `_string_` + +任务停止原因，自然停止时为`stop`。 + +**message** `_object_` + +模型返回的消息。 + +**属性** + +**role** `_string_` + +消息的角色，固定为`assistant`。 + +**content** `_array_` + +消息内容，包含生成的图像信息。 + +**属性** + +**image** `_string_` + +生成图像的 URL，格式为PNG。**链接有效期为24小时**，请及时下载并保存图像。 + +**usage** `_object_` + +本次调用的资源使用情况，仅调用成功时返回。 + +**属性** + +**width** `_integer_` + +生成图像的宽度（像素）。 + +**height** `_integer_` + +生成图像的高度（像素）。 + +**image\_count** `_integer_` + +生成图像的张数。 + +**request\_id** `_string_` + +请求唯一标识。可用于请求明细溯源和问题排查。 + +**code** `_string_` + +请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + +**message** `_string_` + +请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + +## SDK调用 + +以下以图生图/图像编辑（I2I）为示例，展示Python和Java SDK的调用方式。 + +## Python + +``` +import os +import base64 +import mimetypes +import dashscope +from dashscope import MultiModalConversation + +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' + +def encode_file(file_path): + mime_type, _ = mimetypes.guess_type(file_path) + if not mime_type or not mime_type.startswith("image/"): + raise ValueError("Unsupported or unrecognized image format") + with open(file_path, "rb") as image_file: + encoded_string = base64.b64encode(image_file.read()).decode('utf-8') + return f"data:{mime_type};base64,{encoded_string}" + +# [方法一] 使用公网图像URL +image_url = "https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/yBRq1ZPYEaXdyOdv/img/33a80a19-7ac7-4c64-b0fa-7d685b7046a0.png" + +# [方法二] 使用Base64编码图像 +# image_url = encode_file("./your_image.png") + +response = MultiModalConversation.call( + api_key=os.getenv("DASHSCOPE_API_KEY"), + model="qwen-image-3.0-pro", + messages=[{ + "role": "user", + "content": [ + {"image": image_url}, + {"text": "帮我生成一张充满高级感的都市风格女性写真，画面中人物完美保留输入图片中这位年轻女性的面部特征与一头柔顺的黑色长发。人物换上一套彰显高雅气质的都市职场穿搭，场景设定在一家装修现代简约的高端咖啡店内。"} + ] + }], + prompt_extend=True +) + +print(response) +if response.status_code == 200: + url = response.output.choices[0].message.content[0]["image"] + print(f"Generated image URL: {url}") +else: + print(f"Error: {response.code} - {response.message}") +``` + +## Java + +``` +import java.util.Arrays; +import java.util.Base64; +import java.util.Collections; +import java.io.IOException; +import java.nio.file.Files; +import java.nio.file.Path; +import java.nio.file.Paths; +import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; +import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; +import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; +import com.alibaba.dashscope.common.MultiModalMessage; +import com.alibaba.dashscope.common.Role; +import com.alibaba.dashscope.utils.Constants; + +public class ImageEditExample { + public static void main(String[] args) { + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + + // [方法一] 使用公网图像URL + String imageUrl = "https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/yBRq1ZPYEaXdyOdv/img/33a80a19-7ac7-4c64-b0fa-7d685b7046a0.png"; + + // [方法二] 使用Base64编码图像 + // String imageUrl = encodeFile("/path/to/your/image.png"); + + MultiModalConversation conv = new MultiModalConversation(); + MultiModalMessage userMessage = MultiModalMessage.builder() + .role(Role.USER.getValue()) + .content(Arrays.asList( + Collections.singletonMap("image", imageUrl), + Collections.singletonMap("text", "帮我生成一张充满高级感的都市风格女性写真，画面中人物完美保留输入图片中这位年轻女性的面部特征与一头柔顺的黑色长发。人物换上一套彰显高雅气质的都市职场穿搭，场景设定在一家装修现代简约的高端咖啡店内。") + )) + .build(); + MultiModalConversationParam param = MultiModalConversationParam.builder() + .apiKey(System.getenv("DASHSCOPE_API_KEY")) + .model("qwen-image-3.0-pro") + .messages(Arrays.asList(userMessage)) + .parameter("prompt_extend", true) + .build(); + try { + MultiModalConversationResult result = conv.call(param); + System.out.println(result); + } catch (Exception e) { + e.printStackTrace(); + } + } + + public static String encodeFile(String filePath) { + Path path = Paths.get(filePath); + if (!Files.exists(path)) { + throw new IllegalArgumentException("File does not exist: " + filePath); + } + String mimeType = null; + try { + mimeType = Files.probeContentType(path); + } catch (IOException e) { + throw new IllegalArgumentException("Cannot detect file type: " + filePath); + } + if (mimeType == null || !mimeType.startsWith("image/")) { + throw new IllegalArgumentException("Unsupported or unrecognized image format"); + } + byte[] fileBytes = null; + try { + fileBytes = Files.readAllBytes(path); + } catch (IOException e) { + throw new IllegalArgumentException("Cannot read file content: " + filePath); + } + String encodedString = Base64.getEncoder().encodeToString(fileBytes); + return "data:" + mimeType + ";base64," + encodedString; + } +} +``` + +## **错误码** + +如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/async-task-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/async-task-api.md index ef12bb72..2037a99d 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/async-task-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/async-task-api.md @@ -782,7 +782,7 @@ RocketMQ创建完成后，打开配置的事件目标界面，选择已配置的 - RocketMQ在线查看消息需要开通消息一键收发体验功能。 -- 消息一键收发体验功能是基于[函数计算](https://help.aliyun.com/zh/functioncompute/fc/product-overview/what-is-function-compute)实现的，如果超过了免费试用额度后将会产生少量费用，请查看[函数计算计费规则](https://help.aliyun.com/zh/functioncompute/fc-2-0/product-overview/billing-overview)。 +- 消息一键收发体验功能是基于[函数计算](https://help.aliyun.com/zh/functioncompute/what-is-function-compute)实现的，如果超过了免费试用额度后将会产生少量费用，请查看[函数计算计费规则](https://help.aliyun.com/zh/functioncompute/fc-2-0/product-overview/billing-overview)。 在**Topic 管理**页面找到已创建的Topic，单击其右侧**操作**列中的**详情**。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md deleted file mode 100644 index ac67d5f5..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md +++ /dev/null @@ -1,1726 +0,0 @@ -# GUI-Plus API参考 - -本文介绍通过 OpenAI 兼容接口 或 DashScope API 调用GUI-Plus模型的输入与输出参数。 - -> 相关文档：[界面交互专用模型（GUI-Plus）](https://help.aliyun.com/zh/model-studio/gui-automation) - -## OpenAI 兼容 - -SDK 调用配置的`base_url`为：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - -HTTP 调用配置的`endpoint`：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions` - -**重要** - -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - -- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - -- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` - - -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 - -> 您需要已[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。若通过OpenAI SDK进行调用，需要[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 - -### 请求体 - -## 非流式输出 - -## Python - -``` -import os -from openai import OpenAI - -system_prompt = """# Tools - -You may call one or more functions to assist with the user query. - -You are provided with function signatures within XML tags: - -{"type": "function", "function": {"name": "computer_use", "description": "Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.", "parameters": {"properties": {"action": {"description": "The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.", "enum": ["key", "type", "mouse_move", "left_click", "left_click_drag", "right_click", "middle_click", "double_click", "triple_click", "scroll", "hscroll", "wait", "terminate", "answer", "interact"], "type": "string"}, "keys": {"description": "Required only by `action=key`.", "type": "array"}, "text": {"description": "Required only by `action=type`, `action=answer` and `action=interact`.", "type": "string"}, "coordinate": {"description": "(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.", "type": "array"}, "pixels": {"description": "The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.", "type": "number"}, "time": {"description": "The seconds to wait. Required only by `action=wait`.", "type": "number"}, "status": {"description": "The status of the task. Required only by `action=terminate`.", "type": "string", "enum": ["success", "failure"]}}, "required": ["action"], "type": "object"}}} - - -For each function call, return a json object with function name and arguments within XML tags: - -{"name": , "arguments": } - - -# Response format - -Response format for every step: -1) Action: a short imperative describing what to do in the UI. -2) A single ... block containing only the JSON: {"name": , "arguments": }. - -Rules: -- Output exactly in the order: Action, . -- Be brief: one for Action. -- Do not output anything else outside those two parts. -- If finishing, use action=terminate in the tool call.""" - -messages = [ - { - "role": "system", - "content": system_prompt - }, - { - "role": "user", - "content": [ - {"type": "image_url", "image_url": {"url": "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png"}}, - {"type": "text", "text": "帮我打开浏览器"} - ] - } -] - -client = OpenAI( - # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key="sk-xxx", - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -) - -completion = client.chat.completions.create( - model="gui-plus-2026-02-26", - messages=messages, - extra_body={"vl_high_resolution_images": True} -) - -print(completion.choices[0].message.content) -``` - -## Node.js - -``` -import OpenAI from "openai"; - -const systemPrompt = `# Tools - -You may call one or more functions to assist with the user query. - -You are provided with function signatures within XML tags: - -{"type": "function", "function": {"name": "computer_use", "description": "Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.", "parameters": {"properties": {"action": {"description": "The action to perform. The available actions are:\\n* \`key\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* \`type\`: Type a string of text on the keyboard.\\n* \`mouse_move\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* \`left_click\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* \`left_click_drag\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* \`right_click\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* \`middle_click\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* \`double_click\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* \`triple_click\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* \`scroll\`: Performs a scroll of the mouse scroll wheel.\\n* \`hscroll\`: Performs a horizontal scroll (mapped to regular scroll).\\n* \`wait\`: Wait specified seconds for the change to happen.\\n* \`terminate\`: Terminate the current task and report its completion status.\\n* \`answer\`: Answer a question.\\n* \`interact\`: Resolve the blocking window by interacting with the user.", "enum": ["key", "type", "mouse_move", "left_click", "left_click_drag", "right_click", "middle_click", "double_click", "triple_click", "scroll", "hscroll", "wait", "terminate", "answer", "interact"], "type": "string"}, "keys": {"description": "Required only by \`action=key\`.", "type": "array"}, "text": {"description": "Required only by \`action=type\`, \`action=answer\` and \`action=interact\`.", "type": "string"}, "coordinate": {"description": "(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \`action=mouse_move\` and \`action=left_click_drag\`.", "type": "array"}, "pixels": {"description": "The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \`action=scroll\` and \`action=hscroll\`.", "type": "number"}, "time": {"description": "The seconds to wait. Required only by \`action=wait\`.", "type": "number"}, "status": {"description": "The status of the task. Required only by \`action=terminate\`.", "type": "string", "enum": ["success", "failure"]}}, "required": ["action"], "type": "object"}}} - - -For each function call, return a json object with function name and arguments within XML tags: - -{"name": , "arguments": } - - -# Response format - -Response format for every step: -1) Action: a short imperative describing what to do in the UI. -2) A single ... block containing only the JSON: {"name": , "arguments": }. - -Rules: -- Output exactly in the order: Action, . -- Be brief: one for Action. -- Do not output anything else outside those two parts. -- If finishing, use action=terminate in the tool call.`; - -const client = new OpenAI({ - apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -}); - -const messages = [ - { - role: "system", - content: systemPrompt, - }, - { - role: "user", - content: [ - { - type: "image_url", - image_url: { - url: "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png", - }, - }, - { type: "text", text: "帮我打开浏览器" }, - ], - }, -]; - -const completion = await client.chat.completions.create({ - model: "gui-plus-2026-02-26", - messages: messages, - extra_body: { vl_high_resolution_images: true }, -}); - -console.log(completion.choices[0].message.content); -``` - -## curl - -``` -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "gui-plus-2026-02-26", - "messages": [ - { - "role": "system", - "content": "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn'\''t open, try wait and taking another screenshot.\\n* The screen'\''s resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don'\''t click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it'\''s the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call." - }, - { - "role": "user", - "content": [ - { - "type": "image_url", - "image_url": { - "url": "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png" - } - }, - { - "type": "text", - "text": "帮我打开浏览器" - } - ] - } - ], - "vl_high_resolution_images": true - }' -``` - -**返回结果** - -``` -{ - "choices": [ - { - "message": { - "content": "\n{\"name\": \"computer_use\", \"arguments\": {\"action\": \"left_click\", \"coordinate\": [2530, 314]}}\n", - "role": "assistant" - }, - "finish_reason": "stop", - "index": 0, - "logprobs": null - } - ], - "object": "chat.completion", - "usage": { - "prompt_tokens": 7750, - "completion_tokens": 36, - "total_tokens": 7786, - "prompt_tokens_details": { - "image_tokens": 6743, - "text_tokens": 1007 - }, - "completion_tokens_details": { - "text_tokens": 36 - } - }, - "created": 1773133741, - "system_fingerprint": null, - "model": "gui-plus", - "id": "chatcmpl-8b375016-abb8-9791-856c-74b2825c22d5" -} -``` - -## 流式输出 - -## Python - -``` -import os -from openai import OpenAI - -system_prompt = """# Tools - -You may call one or more functions to assist with the user query. - -You are provided with function signatures within XML tags: - -{"type": "function", "function": {"name": "computer_use", "description": "Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.", "parameters": {"properties": {"action": {"description": "The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.", "enum": ["key", "type", "mouse_move", "left_click", "left_click_drag", "right_click", "middle_click", "double_click", "triple_click", "scroll", "hscroll", "wait", "terminate", "answer", "interact"], "type": "string"}, "keys": {"description": "Required only by `action=key`.", "type": "array"}, "text": {"description": "Required only by `action=type`, `action=answer` and `action=interact`.", "type": "string"}, "coordinate": {"description": "(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.", "type": "array"}, "pixels": {"description": "The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.", "type": "number"}, "time": {"description": "The seconds to wait. Required only by `action=wait`.", "type": "number"}, "status": {"description": "The status of the task. Required only by `action=terminate`.", "type": "string", "enum": ["success", "failure"]}}, "required": ["action"], "type": "object"}}} - - -For each function call, return a json object with function name and arguments within XML tags: - -{"name": , "arguments": } - - -# Response format - -Response format for every step: -1) Action: a short imperative describing what to do in the UI. -2) A single ... block containing only the JSON: {"name": , "arguments": }. - -Rules: -- Output exactly in the order: Action, . -- Be brief: one for Action. -- Do not output anything else outside those two parts. -- If finishing, use action=terminate in the tool call.""" - -messages = [ - { - "role": "system", - "content": system_prompt - }, - { - "role": "user", - "content": [ - {"type": "image_url", "image_url": {"url": "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png"}}, - {"type": "text", "text": "帮我打开浏览器"} - ] - } -] - -client = OpenAI( - # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key="sk-xxx", - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -) - -completion = client.chat.completions.create( - model="gui-plus-2026-02-26", - messages=messages, - stream=True, - stream_options={"include_usage":True} -) -for chunk in completion: - print(chunk.model_dump_json()) -``` - -## Node.js - -``` -import OpenAI from "openai"; - -const systemPrompt = `# Tools - -You may call one or more functions to assist with the user query. - -You are provided with function signatures within XML tags: - -{"type": "function", "function": {"name": "computer_use", "description": "Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.", "parameters": {"properties": {"action": {"description": "The action to perform. The available actions are:\\n* \`key\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* \`type\`: Type a string of text on the keyboard.\\n* \`mouse_move\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* \`left_click\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* \`left_click_drag\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* \`right_click\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* \`middle_click\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* \`double_click\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* \`triple_click\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* \`scroll\`: Performs a scroll of the mouse scroll wheel.\\n* \`hscroll\`: Performs a horizontal scroll (mapped to regular scroll).\\n* \`wait\`: Wait specified seconds for the change to happen.\\n* \`terminate\`: Terminate the current task and report its completion status.\\n* \`answer\`: Answer a question.\\n* \`interact\`: Resolve the blocking window by interacting with the user.", "enum": ["key", "type", "mouse_move", "left_click", "left_click_drag", "right_click", "middle_click", "double_click", "triple_click", "scroll", "hscroll", "wait", "terminate", "answer", "interact"], "type": "string"}, "keys": {"description": "Required only by \`action=key\`.", "type": "array"}, "text": {"description": "Required only by \`action=type\`, \`action=answer\` and \`action=interact\`.", "type": "string"}, "coordinate": {"description": "(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \`action=mouse_move\` and \`action=left_click_drag\`.", "type": "array"}, "pixels": {"description": "The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \`action=scroll\` and \`action=hscroll\`.", "type": "number"}, "time": {"description": "The seconds to wait. Required only by \`action=wait\`.", "type": "number"}, "status": {"description": "The status of the task. Required only by \`action=terminate\`.", "type": "string", "enum": ["success", "failure"]}}, "required": ["action"], "type": "object"}}} - - -For each function call, return a json object with function name and arguments within XML tags: - -{"name": , "arguments": } - - -# Response format - -Response format for every step: -1) Action: a short imperative describing what to do in the UI. -2) A single ... block containing only the JSON: {"name": , "arguments": }. - -Rules: -- Output exactly in the order: Action, . -- Be brief: one for Action. -- Do not output anything else outside those two parts. -- If finishing, use action=terminate in the tool call.`; - -const client = new OpenAI({ - apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -}); - -const messages = [ - { - role: "system", - content: systemPrompt, - }, - { - role: "user", - content: [ - { - type: "image_url", - image_url: { - url: "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png", - }, - }, - { type: "text", text: "帮我打开浏览器" }, - ], - }, -]; - -const openai = new OpenAI({ - // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: "sk-xxx" - apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" -}); - -async function main() { - const response = await openai.chat.completions.create({ - model: "gui-plus", - messages: messages, - stream:true, - stream_options:{"include_usage":true} - }); - - console.log("流式输出内容为：") - for await (const chunk of response) { - // 如果stream_options.include_usage为true，则最后一个chunk的choices字段为空数组，需要跳过（可以通过chunk.usage获取 Token 使用量） - if (chunk.choices[0] && chunk.choices[0].delta.content != null) { - console.log(chunk.choices[0].delta.content); - } - } -} -main() -``` - -## curl - -``` -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "gui-plus-2026-02-26", - "messages": [ - { - "role": "system", - "content": "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn'\''t open, try wait and taking another screenshot.\\n* The screen'\''s resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don'\''t click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it'\''s the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call." - }, - { - "role": "user", - "content": [ - { - "type": "image_url", - "image_url": { - "url": "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png" - } - }, - { - "type": "text", - "text": "帮我打开浏览器" - } - ] - } - ], - "vl_high_resolution_images": true, - "stream":true, - "stream_options":{"include_usage":true} - }' -``` - -**model** `_string_` **（必选）** - -模型名称。支持的模型：gui-plus。 - -**messages** `_array_` **（必选）**传递给大模型的上下文，按对话顺序排列。 - -**消息类型** - -System Message `_object_` （可选） - -系统消息，用于设定大模型的角色、语气、任务目标或约束条件等。一般放在`messages`数组的第一位。 - -**属性** - -**content** `_string_` **（必选）** - -系统指令，用于明确模型的角色、行为规范、回答风格和任务约束等。 - -**role** `_string_` **（必选）** - -系统消息的角色，固定为`system`。 - -User Message `_object_` **（必选）** - -用户消息。 - -**属性** - -**content** `_array_`**（必选）** - -消息内容。 - -**属性** - -**type** `_string_` **（必选）** - -可选值： - -- `text` - - 输入文本时需设为`text`。 - -- `image_url` - - 输入图片时需设为`image_url`。 - - -**text** `_string_` **（可选）** - -输入的文本。当`type`为`text`时，是必选参数。 - -**image\_url** `_object_` - -输入的图片信息。当`type`为`image_url`时是必选参数。 - -**属性** - -**url** `_string_`**（必选）** - -图片的 URL或 Base64 Data URL。传入本地文件请参考[传入本地文件](https://help.aliyun.com/zh/model-studio/vision#a63fbac15a8s8)。 - -**min\_pixels** `_integer_` （可选） - -用于设定输入图像的最小像素阈值，单位为像素。 - -当输入图像像素小于`min_pixels`时，会将图像进行放大，直到总像素高于`min_pixels`。 - -默认值和最小值均为 3136 。 - -**max\_pixels** `_integer_` （可选） - -用于设定输入图像的最大像素阈值，单位为像素。 - -当输入图像像素在`[min_pixels, max_pixels]`区间内时，模型会按原图进行识别。当输入图像像素大于`max_pixels`时，会将图像进行缩小，直到总像素低于`max_pixels`。 - -默认值和最大值和[vl\_high\_resolution\_images](#3b5c2d499e544)的取值有关： - -- 当 [vl\_high\_resolution\_images](#3b5c2d499e544) 为False时：默认值为1003520 ，最大值为12845056 - -- 当 [vl\_high\_resolution\_images](#3b5c2d499e544) 为True时：max\_pixels无效，输入图像的最大像素固定为12845056 - - -**role** `_string_` **（必选）** - -用户消息的角色，固定为`user`。 - -Assistant Message `_object_` （可选） - -模型的回复。通常用于在多轮对话中作为上下文回传给模型。 - -**属性** - -**content** `_string_` （必选） - -模型回复的文本内容。 - -**role** `_string_` **（必选）** - -助手消息的角色，固定为`assistant`。 - -**stream** `_boolean_` （可选） 默认值为 `false` - -是否以流式方式输出回复。 - -可选值： - -- `false`：等待模型生成完整回复后一次性返回。 - -- `true`：模型边生成边返回数据块。客户端需逐块读取，以还原完整回复。 - - -**stream\_options** `_object_` （可选） - -流式输出的配置项，仅在 `stream` 为 `true` 时生效。 - -**属性** - -**include\_usage** `_boolean_` （可选）默认值为 `false` - -是否在**最后一个数据块**包含Token消耗信息。 - -可选值： - -- `true`：包含； - -- `false`：不包含。 - - -**max\_tokens** `_integer_` （可选） - -用于限制模型输出的最大 Token 数。若生成内容超过此值，响应将被截断。 - -默认值与最大值均为模型的最大输出长度，请参见[模型选型](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#f4299b0a1ace4)。 - -**vl\_high\_resolution\_images** `_boolean_` （可选）默认值为`false` - -是否将输入图像的像素上限提升至 16384 Token 对应的像素值。 - -- `vl_high_resolution_images为true`，使用固定分辨率策略，像素上限固定为`12845056`，忽略 `max_pixels` 设置，超过此分辨率时会将图像总像素缩小至此上限内。 - -- `vl_high_resolution_images`为`false`，像素上限由`max_pixels`决定，输入图像的像素超过`max_pixels`会将图像缩小至`max_pixels`内。模型的默认像素上限即`max_pixels`的默认值。 - - -**enable\_thinking** `_boolean_` （可选） - -使用混合思考（回复前既可思考也可不思考）模型时，是否开启思考模式。在界面交互系列模型中，仅`gui-plus-2026-02-26`为混合思考模型。相关文档：[视觉推理](https://help.aliyun.com/zh/model-studio/visual-reasoning#02ccad9e41nsv) - -可选值： - -- `true`：开启 - - > 开启后，思考内容将通过`reasoning_content`字段返回。 - -- `false`：不开启 - - -> 该参数非OpenAI标准参数。通过 Python SDK调用时，请放入 **extra\_body** 对象中。配置方式为：`extra_body={"enable_thinking": xxx}`。 - -**seed** `_integer_` （可选） - -随机数种子。用于确保在相同输入和参数下生成结果可复现。若调用时传入相同的 `seed` 且其他参数不变，模型将尽可能返回相同结果。 - -取值范围：`[0,231−1]`。 - -**temperature** `_float_` （可选） 默认值为0.01 - -采样温度，控制模型生成文本的多样性。 - -temperature越高，生成的文本更多样，反之，生成的文本更确定。 - -取值范围： \[0, 2) - -temperature与top\_p均可以控制生成文本的多样性，建议只设置其中一个值。 - -**top\_p** `_float_` （可选）默认值为0.01 - -核采样的概率阈值，控制模型生成文本的多样性。 - -top\_p越高，生成的文本更多样。反之，生成的文本更确定。 - -取值范围：（0,1.0\] - -temperature与top\_p均可以控制生成文本的多样性，建议只设置其中一个值。 - -**top\_k** `_integer_` （可选）默认值为1 - -生成过程中采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个Token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。取值为None或当top\_k大于100时，表示不启用top\_k策略，此时仅有top\_p策略生效。 - -取值需要大于或等于0。 - -该参数非OpenAI标准参数。通过 Python SDK调用时，请放入 **extra\_body** 对象中，配置方式为：`extra_body={"top_k": xxx}`；通过 Node.js SDK 或 HTTP 方式调用时，请作为顶层参数传递。 - -**repetition\_penalty** `_float_` （可选）默认值为1.0 - -模型生成时连续序列中的重复度。提高repetition\_penalty时可以降低模型生成的重复度，1.0表示不做惩罚。该参数对模型效果影响较大，建议保持默认值。 - -**presence\_penalty** `_float_` （可选） - -控制模型生成文本时的内容重复度。默认值为1.5 - -取值范围：\[-2.0, 2.0\]。正值降低重复度，负值增加重复度。 - -在创意写作或头脑风暴等需要多样性、趣味性或创造力的场景中，建议调高该值；在技术文档或正式文本等强调一致性与术语准确性的场景中，建议调低该值。 - -**原理介绍** - -如果参数值是正数，模型将对目前文本中已存在的Token施加一个惩罚值（惩罚值与文本出现的次数无关），减少这些Token重复出现的几率，从而减少内容重复度，增加用词多样性。 - -**示例** - -提示词：把这句话翻译成中文“This movie is good. The plot is good, the acting is good, the music is good, and overall, the whole movie is just good. It is really good, in fact. The plot is so good, and the acting is so good, and the music is so good.” - -参数值为2.0：这部电影很好。剧情很棒，演技棒，音乐也非常好听，总的来说，整部电影都好得不得了。实际上它真的很优秀。剧情非常精彩，演技出色，音乐也是那么的动听。 - -参数值为0.0：这部电影很好。剧情好，演技好，音乐也好，总的来说，整部电影都很好。事实上，它真的很棒。剧情非常好，演技也非常出色，音乐也同样优秀。 - -参数值为-2.0：这部电影很好。情节很好，演技很好，音乐也很好，总的来说，整部电影都很好。实际上，它真的很棒。情节非常好，演技也非常好，音乐也非常好。 - -**stop** `_string 或 array_` （可选） - -用于指定停止词。当模型生成的文本中出现`stop` 指定的字符串或`token_id`时，生成将立即终止。 - -可传入敏感词以控制模型的输出。 - -> stop为数组时，不可将`token_id`和字符串同时作为元素输入，比如不可以指定为`["你好",104307]`。 - -### chat响应对象（非流式输出） - -``` -{ - "id": "chatcmpl-ef17511a-aceb-4c47-8757-13a87af2152d", - "choices": [ - { - "finish_reason": "stop", - "index": 0, - "logprobs": null, - "message": { - "content": "```json\n{\"thought\": \"用户想要打开浏览器，我观察到屏幕截图中有一个Google Chrome的图标，其位置在右上角一排的最后一个。因此，下一步操作应该是点击这个Chrome浏览器图标来启动它。\", \"action\": \"click\", \"parameters\": {\"x\": 1086, \"y\": 129}}\n```", - "refusal": null, - "role": "assistant", - "annotations": null, - "audio": null, - "function_call": null, - "tool_calls": null - } - } - ], - "created": 1763451557, - "model": "gui-plus", - "object": "chat.completion", - "service_tier": null, - "system_fingerprint": null, - "usage": { - "completion_tokens": 78, - "prompt_tokens": 2020, - "total_tokens": 2098, - "completion_tokens_details": { - "accepted_prediction_tokens": null, - "audio_tokens": null, - "reasoning_tokens": null, - "rejected_prediction_tokens": null, - "text_tokens": 78 - }, - "prompt_tokens_details": { - "audio_tokens": null, - "cached_tokens": null, - "image_tokens": 1244, - "text_tokens": 776 - } - } -} -``` - -**id** `_string_` - -本次请求的唯一标识符。 - -**choices** `_array_` - -模型生成内容的数组。 - -**属性** - -**finish\_reason** `_string_` - -模型停止生成的原因。 - -有两种情况： - -- 自然停止输出时为`stop`； - -- 生成长度过长而结束为`length`。 - - -**index** `_integer_` - -当前对象在`choices`数组中的索引。 - -**message** `_object_` - -模型输出的消息。 - -**属性** - -**content** `_string_` - -GUI任务的结果。 - -**refusal** `_string_` - -该参数当前固定为`null`。 - -**role** `_string_` - -消息的角色，固定为`assistant`。 - -**audio** `_object_` - -该参数当前固定为`null`。 - -**function\_call** `_object_` - -该参数当前固定为`null`。 - -**tool\_calls** `_array_` - -该参数当前固定为`null`。 - -**created** `_integer_` - -本次请求被创建时的时间戳。 - -**model** `_string_` - -本次请求使用的模型。 - -**object** `_string_` - -始终为`chat.completion`。 - -**service\_tier** `_string_` - -该参数当前固定为`null`。 - -**system\_fingerprint** `_string_` - -该参数当前固定为`null`。 - -**usage** `_object_` - -本次请求的 Token 消耗信息。 - -**属性** - -**completion\_tokens** `_integer_` - -模型输出的 Token 数。 - -**prompt\_tokens** `_integer_` - -输入的 Token 数。 - -**total\_tokens** `_integer_` - -消耗的总 Token 数，为`prompt_tokens`与`completion_tokens`的总和。 - -**completion\_tokens\_details** `_object_` - -模型输出Token的细粒度分类。 - -**属性** - -**accepted\_prediction\_tokens**`_integer_` - -该参数当前固定为`null`。 - -**audio\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**reasoning\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**text\_tokens** `_integer_` - -模型输出文本对应的 Token 数。 - -**rejected\_prediction\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**prompt\_tokens\_details** `_object_` - -输入 Token 的细粒度分类。 - -**属性** - -**audio\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**cached\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**text\_tokens** `_integer_` - -模型输入的文本对应的Token 数。 - -**image\_tokens** `_integer_` - -模型输入的图像对应的 Token数。 - -### chat响应chunk对象（流式输出） - -``` -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content='', function_call=None, refusal=None, role='assistant', tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content='```', function_call=None, refusal=None, role=None, tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content='json', function_call=None, refusal=None, role=None, tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content=None, function_call=None, refusal=None, role=None, tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content='\n{"thought": "', function_call=None, refusal=None, role=None, tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -... -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content=' 1086', function_call=None, refusal=None, role=None, tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content=', "y":', function_call=None, refusal=None, role=None, tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content=' 127', function_call=None, refusal=None, role=None, tool_calls=None), finish_reason=None, index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -ChatCompletionChunk(id='chatcmpl-9f3c627a-b0fc-4160-a558-3cc2cc7aa988', choices=[Choice(delta=ChoiceDelta(content='}}\n```', function_call=None, refusal=None, role=None, tool_calls=None), finish_reason='stop', index=0, logprobs=None)], created=1763452343, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=None) -ChatCompletionChunk(id='chatcmpl-bdb03054-42a2-459b-8a7e-5b94b39626f2', choices=[], created=1763452463, model='gui-plus', object='chat.completion.chunk', service_tier=None, system_fingerprint=None, usage=CompletionUsage(completion_tokens=78, prompt_tokens=2020, total_tokens=2098, completion_tokens_details=CompletionTokensDetails(accepted_prediction_tokens=None, audio_tokens=None, reasoning_tokens=None, rejected_prediction_tokens=None, text_tokens=78), prompt_tokens_details=PromptTokensDetails(audio_tokens=None, cached_tokens=None, image_tokens=1244, text_tokens=776))) -``` - -**id** `_string_` - -本次调用的唯一标识符。每个chunk对象有相同的 id。 - -**choices** `_array_` - -模型生成内容的数组。若设置`include_usage`参数为`true`，则在最后一个chunk中为空。 - -**属性** - -**delta** `_object_` - -流式返回的输出内容。 - -**属性** - -**content** `_string_` - -翻译结果，qwen-mt-flash为增量式更新，qwen-mt-plus和qwen-mt-turbo为非增量式更新。 - -**function\_call** `_object_` - -该参数当前固定为`null`。 - -**refusal** `_object_` - -该参数当前固定为`null`。 - -**role** `_string_` - -消息对象的角色，只在第一个chunk中有值。 - -**finish\_reason** `_string_` - -模型停止生成的原因。有三种情况： - -- 自然停止输出时为`stop`； - -- 生成未结束时为`null`； - -- 生成长度过长而结束为`length`。 - - -**index** `_integer_` - -当前响应在`choices`数组中的索引。 - -**created** `_integer_` - -本次请求被创建时的时间戳。每个chunk有相同的时间戳。 - -**model** `_string_` - -本次请求使用的模型。 - -**object** `_string_` - -始终为`chat.completion.chunk`。 - -**service\_tier** `_string_` - -该参数当前固定为`null`。 - -**system\_fingerprint**`_string_` - -该参数当前固定为`null`。 - -**usage** `_object_` - -本次请求消耗的Token。只在`include_usage`为`true`时，在最后一个chunk返回。 - -**属性** - -**completion\_tokens** `_integer_` - -模型输出的 Token 数。 - -**prompt\_tokens** `_integer_` - -输入的 Token 数。 - -**total\_tokens** `_integer_` - -消耗的总 Token 数，为`prompt_tokens`与`completion_tokens`的总和。 - -**completion\_tokens\_details** `_object_` - -模型输出Token的细粒度分类。 - -**属性** - -**accepted\_prediction\_tokens**`_integer_` - -该参数当前固定为`null`。 - -**audio\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**reasoning\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**text\_tokens** `_integer_` - -模型输出文本对应的 Token 数。 - -**rejected\_prediction\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**prompt\_tokens\_details** `_object_` - -输入 Token 的细粒度分类。 - -**属性** - -**audio\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**cached\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**text\_tokens** `_integer_` - -模型输入的文本对应的Token 数。 - -**image\_tokens** `_integer_` - -模型输入的图像对应的 Token数。 - -## DashScope - -- HTTP 请求地址：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -- SDK 调用：无需配置 `base_url` - - -> 您需要已[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。若通过DashScope SDK进行调用，需要[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk#f3e80b21069aa)。 - -### 请求体 - -## 非流式输出 - -## Python - -``` -import os -import dashscope -# 以下为华北2（北京）地域的配置，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的配置不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -system_prompt = """# Tools - -You may call one or more functions to assist with the user query. - -You are provided with function signatures within XML tags: - -{"type": "function", "function": {"name": "computer_use", "description": "Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.", "parameters": {"properties": {"action": {"description": "The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.", "enum": ["key", "type", "mouse_move", "left_click", "left_click_drag", "right_click", "middle_click", "double_click", "triple_click", "scroll", "hscroll", "wait", "terminate", "answer", "interact"], "type": "string"}, "keys": {"description": "Required only by `action=key`.", "type": "array"}, "text": {"description": "Required only by `action=type`, `action=answer` and `action=interact`.", "type": "string"}, "coordinate": {"description": "(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.", "type": "array"}, "pixels": {"description": "The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.", "type": "number"}, "time": {"description": "The seconds to wait. Required only by `action=wait`.", "type": "number"}, "status": {"description": "The status of the task. Required only by `action=terminate`.", "type": "string", "enum": ["success", "failure"]}}, "required": ["action"], "type": "object"}}} - - -For each function call, return a json object with function name and arguments within XML tags: - -{"name": , "arguments": } - - -# Response format - -Response format for every step: -1) Action: a short imperative describing what to do in the UI. -2) A single ... block containing only the JSON: {"name": , "arguments": }. - -Rules: -- Output exactly in the order: Action, . -- Be brief: one for Action. -- Do not output anything else outside those two parts. -- If finishing, use action=terminate in the tool call.""" - -messages = [ - { - "role": "system", - "content": system_prompt - }, - { - "role": "user", - "content": [ - {"image": "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png"}, - {"text": "帮我打开浏览器。"}] - }] - -response = dashscope.MultiModalConversation.call( - # 若没有配置环境变量， 请用百炼API Key将下行替换为： api_key = "sk-xxx" - api_key=os.getenv('DASHSCOPE_API_KEY'), - model='gui-plus-2026-02-26', - messages=messages, - vl_high_resolution_images=True -) - -print(response.output.choices[0].message.content[0]["text"]) -``` - -## Java - -``` -import java.util.Arrays; -import java.util.Collections; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - // 以下为华北2（北京）地域的配置，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - String systemPrompt = "# Tools\n\n" + - "You may call one or more functions to assist with the user query.\n\n" + - "You are provided with function signatures within XML tags:\n" + - "\n" + - "{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n" + - "\n\n" + - "For each function call, return a json object with function name and arguments within XML tags:\n" + - "\n" + - "{\"name\": , \"arguments\": }\n" + - "\n\n" + - "# Response format\n\n" + - "Response format for every step:\n" + - "1) Action: a short imperative describing what to do in the UI.\n" + - "2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\n" + - "Rules:\n" + - "- Output exactly in the order: Action, .\n" + - "- Be brief: one for Action.\n" + - "- Do not output anything else outside those two parts.\n" + - "- If finishing, use action=terminate in the tool call."; - MultiModalConversation conv = new MultiModalConversation(); - MultiModalMessage systemMsg = MultiModalMessage.builder().role(Role.SYSTEM.getValue()) - .content(Arrays.asList( - Collections.singletonMap("text",systemPrompt))).build(); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - Collections.singletonMap("image", "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png"), - Collections.singletonMap("text", "帮我打开浏览器。"))).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("gui-plus-2026-02-26") - .messages(Arrays.asList(systemMsg,userMessage)) - .vlHighResolutionImages(true) - .build(); - MultiModalConversationResult result = conv.call(param); - System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get("text")); - } - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -## curl - -``` -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "gui-plus-2026-02-26", - "input": { - "messages": [ - { - "role": "system", - "content": [ - { - "text": "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn'\''t open, try wait and taking another screenshot.\\n* The screen'\''s resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don'\''t click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it'\''s the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call." - } - ] - }, - { - "role": "user", - "content": [ - { - "image": "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png" - }, - { - "text": "帮我打开浏览器" - } - ] - } - ] - }, - "parameters": { - "vl_high_resolution_images": true - } - }' -``` - -## 流式输出 - -## Python - -``` -system_prompt = """# Tools - -You may call one or more functions to assist with the user query. - -You are provided with function signatures within XML tags: - -{"type": "function", "function": {"name": "computer_use", "description": "Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.", "parameters": {"properties": {"action": {"description": "The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.", "enum": ["key", "type", "mouse_move", "left_click", "left_click_drag", "right_click", "middle_click", "double_click", "triple_click", "scroll", "hscroll", "wait", "terminate", "answer", "interact"], "type": "string"}, "keys": {"description": "Required only by `action=key`.", "type": "array"}, "text": {"description": "Required only by `action=type`, `action=answer` and `action=interact`.", "type": "string"}, "coordinate": {"description": "(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.", "type": "array"}, "pixels": {"description": "The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.", "type": "number"}, "time": {"description": "The seconds to wait. Required only by `action=wait`.", "type": "number"}, "status": {"description": "The status of the task. Required only by `action=terminate`.", "type": "string", "enum": ["success", "failure"]}}, "required": ["action"], "type": "object"}}} - - -For each function call, return a json object with function name and arguments within XML tags: - -{"name": , "arguments": } - - -# Response format - -Response format for every step: -1) Action: a short imperative describing what to do in the UI. -2) A single ... block containing only the JSON: {"name": , "arguments": }. - -Rules: -- Output exactly in the order: Action, . -- Be brief: one for Action. -- Do not output anything else outside those two parts. -- If finishing, use action=terminate in the tool call.""" - -messages = [ - { - "role": "system", - "content": system_prompt - }, - { - "role": "user", - "content": [ - {"image": "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png"}, - {"text": "帮我打开浏览器。"}] - }] - -response = dashscope.MultiModalConversation.call( - # 若没有配置环境变量， 请用百炼API Key将下行替换为： api_key = "sk-xxx" - api_key = os.getenv('DASHSCOPE_API_KEY'), - model = 'gui-plus', - messages = messages, - stream=True -) -for chunk in response: - print(chunk) -``` - -## Java - -``` -import java.util.Arrays; -import java.util.Collections; - -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; -import io.reactivex.Flowable; - -public class Main { - // 以下为华北2（北京）地域的配置，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; - - public static void streamCall() - throws ApiException, NoApiKeyException, UploadFileException { - String systemPrompt = "# Tools\n\n" + - "You may call one or more functions to assist with the user query.\n\n" + - "You are provided with function signatures within XML tags:\n" + - "\n" + - "{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n" + - "\n\n" + - "For each function call, return a json object with function name and arguments within XML tags:\n" + - "\n" + - "{\"name\": , \"arguments\": }\n" + - "\n\n" + - "# Response format\n\n" + - "Response format for every step:\n" + - "1) Action: a short imperative describing what to do in the UI.\n" + - "2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\n" + - "Rules:\n" + - "- Output exactly in the order: Action, .\n" + - "- Be brief: one for Action.\n" + - "- Do not output anything else outside those two parts.\n" + - "- If finishing, use action=terminate in the tool call."; - MultiModalConversation conv = new MultiModalConversation(); - MultiModalMessage systemMsg = MultiModalMessage.builder().role(Role.SYSTEM.getValue()) - .content(Arrays.asList( - Collections.singletonMap("text",systemPrompt))).build(); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - Collections.singletonMap("image", "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png"), - Collections.singletonMap("text", "帮我打开浏览器。"))).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("gui-plus") - .messages(Arrays.asList(userMessage,systemMsg)) - .incrementalOutput(true) - .build(); - Flowable result = conv.streamCall(param); - result.blockingForEach(item -> { - try { - var content = item.getOutput().getChoices().get(0).getMessage().getContent(); - // 判断content是否存在且不为空 - if (content != null && !content.isEmpty()) { - System.out.println(content.get(0).get("text")); - } - } catch (Exception e) { - System.out.println(e.getMessage()); - } - }); - } - - public static void main(String[] args) { - try { - streamCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -## curl - -``` -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \ --H "Authorization: Bearer $DASHSCOPE_API_KEY" \ --H 'Content-Type: application/json' \ --H "X-DashScope-SSE: enable" \ --d '{ - "model": "gui-plus-2026-02-26", - "input": { - "messages": [ - { - "role": "system", - "content": [ - { - "text": "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn'\''t open, try wait and taking another screenshot.\\n* The screen'\''s resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don'\''t click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it'\''s the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call." - } - ] - }, - { - "role": "user", - "content": [ - { - "image": "https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png" - }, - { - "text": "帮我打开浏览器" - } - ] - } - ] - }, - "parameters": { - "vl_high_resolution_images": true - } - }' -``` - -**model** `_string_` **（必选）** - -模型名称。支持的模型：gui-plus。 - -**messages** `_array_` **（必选）** - -传递给大模型的上下文，按对话顺序排列。 - -> 通过HTTP调用时，请将**messages** 放入 **input** 对象中。 - -**消息类型** - -System Message `_object_` （可选） - -系统消息，用于设定大模型的角色、语气、任务目标或约束条件等。一般放在`messages`数组的第一位。 - -**属性** - -**content** `_string_` **（必选）** - -系统指令，用于明确模型的角色、行为规范、回答风格和任务约束等。 - -**role** `_string_` **（必选）** - -系统消息的角色，固定为`system`。 - -User Message `_object_`**（必选）** - -用户消息，用于向模型传递问题、指令或上下文等。 - -**属性** - -**content** `_string 或 array_`**（必选）** - -消息内容。若输入只有文本，则为 string 类型；若输入包含图像数据，则为 array 类型。 - -**属性** - -**text** `_string_`**（必选）** - -输入的文本。 - -**image** `_string_`（可选） - -传入的图片文件。可以为图片的URL或本地路径。传入本地文件请参见[传入本地文件](https://help.aliyun.com/zh/model-studio/vision#f18fc2bb52wxo)。 - -示例值：{"image":"https://xxxx.jpeg","max-pixels":1280\*28\*28} - -**min\_pixels** `_integer_` （可选） - -用于设定输入图像的最小像素阈值，单位为像素。 - -当输入图像像素小于`min_pixels`时，会将图像进行放大，直到总像素高于`min_pixels`。 - -默认值和最小值均为 3136 。 - -与 image 参数一起使用，示例：{"image":"https://xxxx.jpeg","min\_pixels":3136} - -**max\_pixels** `_integer_` （可选） - -用于设定输入图像的最大像素阈值，单位为像素。 - -当输入图像像素在`[min_pixels, max_pixels]`区间内时，模型会按原图进行识别。当输入图像像素大于`max_pixels`时，会将图像进行缩小，直到总像素低于`max_pixels`。 - -默认值和最大值和[vl\_high\_resolution\_images](#3b5c2d499e544)的取值有关： - -- 当 [vl\_high\_resolution\_images](#3b5c2d499e544) 为False时：默认值为1003520 ，最大值为12845056 - -- 当 [vl\_high\_resolution\_images](#3b5c2d499e544) 为True时：max\_pixels无效，输入图像的最大像素固定为12845056 - - -与 image 参数一起使用，示例：{"image":"https://xxxx.jpeg","max\_pixels":1003520 } - -Assistant Message `_object_` （可选） - -模型的回复。通常用于在多轮对话中作为上下文回传给模型。 - -**属性** - -**content** `_string_` （必选） - -模型回复的文本内容。 - -**role** `_string_` **（必选）** - -助手消息的角色，固定为`assistant`。 - -**vl\_high\_resolution\_images** `_boolean_` （可选）默认值为`false` - -是否将输入图像的像素上限提升至 16384 Token 对应的像素值。 - -- `vl_high_resolution_images为true`，使用固定分辨率策略，像素上限固定为`12845056`，忽略 `max_pixels` 设置，超过此分辨率时会将图像总像素缩小至此上限内。 - -- `vl_high_resolution_images`为`false`，像素上限由`max_pixels`决定，输入图像的像素超过`max_pixels`会将图像缩小至`max_pixels`内。模型的默认像素上限即`max_pixels`的默认值。 - - -**enable\_thinking** `_boolean_` （可选） - -使用混合思考模型时，是否开启思考模式。在界面交互系列模型中，仅`gui-plus-2026-02-26`为混合思考模型。相关文档：[视觉推理](https://help.aliyun.com/zh/model-studio/visual-reasoning#02ccad9e41nsv) - -可选值： - -- `true`：开启 - - > 开启后，思考内容将通过`reasoning_content`字段返回。 - -- `false`：不开启 - - -不同模型的默认值：[支持的模型](https://help.aliyun.com/zh/model-studio/deep-thinking#78286fdc35hlw) - -> Java SDK 为enableThinking；通过HTTP调用时，请将 **enable\_thinking** 放入 **parameters** 对象中。 - -**max\_tokens** `_integer_` （可选） - -用于限制模型输出的最大 Token 数。若生成内容超过此值，响应将被截断。 - -默认值与最大值均为模型的最大输出长度，请参见[模型选型](https://help.aliyun.com/zh/model-studio/machine-translation#efd59c2b9eosx)。 - -> Java SDK中为**maxTokens**_。_通过HTTP调用时，请将 **max\_tokens** 放入 **parameters** 对象中。 - -**seed** `_integer_` （可选） - -随机数种子。用于确保在相同输入和参数下生成结果可复现。若调用时传入相同的 `seed` 且其他参数不变，模型将尽可能返回相同结果。 - -取值范围：`[0,231−1]`。 - -> 通过HTTP调用时，请将 **seed** 放入 **parameters** 对象中。 - -**temperature** `_float_` （可选） 默认值为0.01 - -采样温度，控制模型生成文本的多样性。 - -temperature越高，生成的文本更多样，反之，生成的文本更确定。 - -取值范围： \[0, 2) - -temperature与top\_p均可以控制生成文本的多样性，建议只设置其中一个值。 - -> 通过HTTP调用时，请将 **temperature** 放入 **parameters** 对象中。 - -**top\_p** `_float_` （可选）默认值为0.01 - -核采样的概率阈值，控制模型生成文本的多样性。 - -top\_p越高，生成的文本更多样。反之，生成的文本更确定。 - -取值范围：（0,1.0\] - -temperature与top\_p均可以控制生成文本的多样性，建议只设置其中一个值。 - -> Java SDK中为**topP**_。_通过HTTP调用时，请将 **top\_p** 放入 **parameters** 对象中。 - -**repetition\_penalty** `_float_` （可选）默认值为1.0 - -模型生成时连续序列中的重复度。提高repetition\_penalty时可以降低模型生成的重复度，1.0表示不做惩罚。该参数对模型效果影响较大，建议保持默认值。 - -> Java SDK中为**repetitionPenalty**_。_通过HTTP调用时，请将 **repetition\_penalty** 放入 **parameters** 对象中。 - -**presence\_penalty** `_float_` （可选） - -控制模型生成文本时的内容重复度。默认值为1.5 - -取值范围：\[-2.0, 2.0\]。正值降低重复度，负值增加重复度。 - -在创意写作或头脑风暴等需要多样性、趣味性或创造力的场景中，建议调高该值；在技术文档或正式文本等强调一致性与术语准确性的场景中，建议调低该值。 - -**原理介绍** - -如果参数值是正数，模型将对目前文本中已存在的Token施加一个惩罚值（惩罚值与文本出现的次数无关），减少这些Token重复出现的几率，从而减少内容重复度，增加用词多样性。 - -**示例** - -提示词：把这句话翻译成中文“This movie is good. The plot is good, the acting is good, the music is good, and overall, the whole movie is just good. It is really good, in fact. The plot is so good, and the acting is so good, and the music is so good.” - -参数值为2.0：这部电影很好。剧情很棒，演技棒，音乐也非常好听，总的来说，整部电影都好得不得了。实际上它真的很优秀。剧情非常精彩，演技出色，音乐也是那么的动听。 - -参数值为0.0：这部电影很好。剧情好，演技好，音乐也好，总的来说，整部电影都很好。事实上，它真的很棒。剧情非常好，演技也非常出色，音乐也同样优秀。 - -参数值为-2.0：这部电影很好。情节很好，演技很好，音乐也很好，总的来说，整部电影都很好。实际上，它真的很棒。情节非常好，演技也非常好，音乐也非常好。 - -**top\_k** `_integer_` （可选）默认值为1 - -生成过程中采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个Token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。取值为None或当top\_k大于100时，表示不启用top\_k策略，此时仅有top\_p策略生效。 - -取值需要大于或等于0。 - -该参数非OpenAI标准参数。通过 Python SDK调用时，请放入 **extra\_body** 对象中，配置方式为：`extra_body={"top_k": xxx}`；通过 Node.js SDK 或 HTTP 方式调用时，请作为顶层参数传递。 - -> Java SDK中为**topK**_。_通过HTTP调用时，请将 **top\_k** 放入 **parameters** 对象中。 - -**repetition\_penalty** `_float_` （可选）默认值为1.0 - -模型生成时连续序列中的重复度。提高repetition\_penalty时可以降低模型生成的重复度，1.0表示不做惩罚。该参数对模型效果影响较大，建议保持默认值。 - -**stream** `_boolean_` （可选） - -是否以流式方式输出回复。 - -可选值： - -- `false`：等待模型生成完整回复后一次性返回。 - -- `true`：模型边生成边返回数据块。客户端需逐块读取，以还原完整回复。 - - -> 该参数仅支持Python SDK。通过Java SDK实现流式输出请通过`streamCall`接口调用；通过HTTP实现流式输出请在Header中指定`X-DashScope-SSE`为`enable`。 - -**incremental\_output** `_boolean_` （可选）默认为`false` - -在流式输出模式下是否开启增量输出。推荐您优先设置为`true`。 - -参数值： - -- false：每次输出为当前已经生成的整个序列，最后一次输出为生成的完整结果。 - - ``` - I - I like - I like apple - I like apple. - ``` - -- true（推荐）：增量输出，即后续输出内容不包含已输出的内容。您需要实时地逐个读取这些片段以获得完整的结果。 - - ``` - I - like - apple - . - ``` - - -> Java SDK中为**incrementalOutput**_。_通过HTTP调用时，请将 **incremental\_output** 放入 **parameters** 对象中。 - -**stop** `_string 或 array_` （可选） - -用于指定停止词。当模型生成的文本中出现`stop` 指定的字符串或`token_id`时，生成将立即终止。 - -可传入敏感词以控制模型的输出。 - -> stop为数组时，不可将`token_id`和字符串同时作为元素输入，比如不可以指定为`["你好",104307]`。 - -### chat响应对象（流式与非流式输出格式一致） - -``` -{ - "status_code": 200, - "request_id": "b74b3a25-3968-4059-8c44-63d793c07f02", - "code": "", - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "text": "```json\n{\"thought\": \"用户想要打开浏览器，我观察到屏幕截图中有一个Google Chrome的图标，其位置在右上角一排的最后一个。因此，下一步操作应该是点击这个Chrome浏览器图标来启动它。\", \"action\": \"CLICK\", \"parameters\": {\"x\": 1086, \"y\": 127}}\n```" - } - ] - } - } - ], - "audio": null - }, - "usage": { - "input_tokens": 2021, - "output_tokens": 78, - "characters": 0, - "image_tokens": 1244, - "input_tokens_details": { - "image_tokens": 1244, - "text_tokens": 777 - }, - "output_tokens_details": { - "text_tokens": 78 - }, - "total_tokens": 2099 - } -} -``` - -**status\_code** `_string_` - -本次请求的状态码。200 表示请求成功，否则表示请求失败。 - -> Java SDK不会返回该参数。调用失败会抛出异常，异常信息为**status\_code**和**message**的内容。 - -**request\_id** `_string_` - -本次调用的唯一标识符。 - -> Java SDK返回参数为**requestId。** - -**code** `_string_` - -错误码，调用成功时为空值。 - -> 只有Python SDK返回该参数。 - -**output** `_object_` - -调用结果信息。 - -**属性** - -**text** `_string_` - -该参数当前固定为`null`。 - -**finish\_reason** `_string_` - -模型结束生成的原因。有以下情况： - -- 正在生成时为`null`； - -- 模型输出自然结束为`stop`； - -- 因生成长度过长而结束为`length`； - - -**choices** `_array_` - -模型的输出信息。 - -**属性** - -**finish\_reason** `_string_` - -有以下情况： - -- 正在生成时为`null`； - -- 因模型输出自然结束为`stop`； - -- 因生成长度过长而结束为`length`； - - -**message** `_object_` - -模型输出的消息对象。 - -**属性** - -**role** `_string_` - -输出消息的角色，固定为`assistant`。 - -**content** `_string_` - -模型输出结果 - -**audio** `_string_` - -该参数当前固定为`null`。 - -**usage** `_object_` - -本次请求使用的Token信息。 - -**属性** - -**input\_tokens** `_integer_` - -输入 Token 数。 - -**output\_tokens** `_integer_` - -输出 Token 数。 - -**image\_tokens** `_integer_` - -输入内容包含`image`时返回该字段。为用户输入图片内容转换成Token后的长度。 - -**characters** `_integer_` - -该参数当前固定为`null`。 - -**input\_tokens\_details** `_object_` - -输入 Token 的细粒度分类。 - -**属性** - -**image\_tokens** `_integer_` - -模型输入的图像对应的 Token数。 - -**text\_tokens** `_integer_` - -模型输入的文本对应的Token 数。 - -**output\_tokens\_details** `_object_` - -输出 Token 的细粒度分类。 - -**属性** - -**text\_tokens** `_integer_` - -模型输入的文本对应的Token 数。 - -**total\_tokens** `_integer_` - -消耗的总 Token 数，为`input_tokens`与`output_tokens`的总和。 - -## **错误码** - -如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/client-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/client-events.md index bb4d9f20..0441479a 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/client-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/client-events.md @@ -89,7 +89,7 @@ Qwen-Omni-Realtime API的客户端事件参考。 默认音色： -- Qwen3.5-Omni-Realtime 系列：`Tina` +- Qwen3.5-Omni-Realtime 系列模型：`Tina` - Qwen3-Omni-Flash-Realtime：`Cherry` @@ -135,7 +135,7 @@ VAD 类型，可选值： - `server_vad`（默认值）：基于声学特征检测语音结束。 -- `semantic_vad`：基于语义有效性检测语音结束，可过滤回应语、背景音等无意义声音。仅 `qwen3.5-omni-realtime` 模型支持。 +- `semantic_vad`：基于语义有效性检测语音结束，可过滤回应语、背景音等无意义声音。仅 `qwen3.5-omni-realtime` 系列模型支持。 **threshold** `_float_`（可选） @@ -160,7 +160,7 @@ VAD 灵敏度。值越低，VAD 越灵敏，越容易将微弱声音（包括背 **enable\_search** `_boolean_`（可选） -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 是否启用联网搜索。设为 `true` 启用，默认为 `false`。启用后，模型可自主判断是否需要联网搜索来回答用户问题。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md index 0aece341..5d4f89d4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md @@ -176,7 +176,7 @@ VAD类型，取值如下： - `server_vad`（默认值）：基于声学特征检测用户语音结束。 -- `semantic_vad`：基于语义有效性检测用户语音结束，可过滤无意义语音（如回应语、背景音）。仅`qwen3.5-omni-realtime`模型支持。 +- `semantic_vad`：基于语义有效性检测用户语音结束，可过滤无意义语音（如回应语、背景音）。仅`qwen3.5-omni-realtime`系列模型支持。 turnDetectionThreshold @@ -210,7 +210,7 @@ enable\_search Boolean -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 是否启用联网搜索功能。设置为 `true` 启用，默认为 `false`。启用后，模型可自主判断是否需要搜索来回应用户的即时问题。 @@ -230,7 +230,7 @@ tools List> -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 工具定义列表。启用后，模型可自主判断是否需要调用外部工具来回应用户的问题。命中工具调用时，模型不生成音频，仅返回工具调用参数。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md index 869f563c..cd0f1593 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md @@ -95,7 +95,7 @@ str 默认音色： -- `Qwen3.5-Omni-Realtime`: `"Tina"` +- `Qwen3.5-Omni-Realtime`系列模型: `"Tina"` - `Qwen3-Omni-Flash-Realtime`: `"Cherry"` @@ -157,7 +157,7 @@ VAD类型，取值如下： - `server_vad`（默认值）：基于声学特征检测用户语音结束。 -- `semantic_vad`：基于语义有效性检测用户语音结束，可过滤无意义语音（如回应语、背景音）。仅`Qwen3.5-Omni-Realtime`模型支持。 +- `semantic_vad`：基于语义有效性检测用户语音结束，可过滤无意义语音（如回应语、背景音）。仅`Qwen3.5-Omni-Realtime`系列模型支持。 turn\_detection\_threshold @@ -191,7 +191,7 @@ enable\_search bool -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 是否启用联网搜索功能。设置为 `true` 启用，默认为 `false`。启用后，模型可自主判断是否需要搜索来回应用户的即时问题。 @@ -209,7 +209,7 @@ tools list\[dict\] -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 工具定义列表。启用后，模型可自主判断是否需要调用外部工具来回应用户的问题。命中工具调用时，模型不生成音频，仅返回工具调用参数。 @@ -244,7 +244,7 @@ temperature越高，生成的内容更多样，反之，生成的内容更确定 由于temperature与top\_p均可以控制生成内容的多样性，因此建议只设置其中一个值。 -- `Qwen3.5-Omni-Realtime`系列：0.7 +- `Qwen3.5-Omni-Realtime`系列模型：0.7 - `Qwen3-Omni-Flash-Realtime`系列：0.9 @@ -267,7 +267,7 @@ top\_p越高，生成的内容更多样。反之，生成的内容更确定。 top\_p默认值： -- `Qwen3.5-Omni-Realtime`系列：0.8 +- `Qwen3.5-Omni-Realtime`系列模型：0.8 - `Qwen3-Omni-Flash-Realtime`系列：1.0 @@ -286,7 +286,7 @@ integer top\_k默认值： -- `Qwen3.5-Omni-Realtime`系列：20 +- `Qwen3.5-Omni-Realtime`系列模型：20 - `Qwen3-Omni-Flash-Realtime`系列：50 @@ -317,7 +317,7 @@ float repetition\_penalty默认值： -- `Qwen3.5-Omni-Realtime`系列：1.0 +- `Qwen3.5-Omni-Realtime`系列模型：1.0 - 其他模型：1.05 @@ -334,7 +334,7 @@ float presence\_penalty默认值： -- `Qwen3.5-Omni-Realtime`系列：1.5 +- `Qwen3.5-Omni-Realtime`系列模型：1.5 - 其他模型：0.0 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md index 7fe83582..f35b6a89 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md @@ -10,9 +10,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## **音频要求** @@ -54,7 +59,7 @@ WAV (16bit)、MP3、M4A ## 快速开始：复刻与使用音色 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1397661871/CAEQbxiBgICd6_Do8BkiIDM3NjYwZDQxMGIyMTQzMDdhOGMyY2YwNWFhMmM2NjVi5899512_20251120114927.389.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2365984871/CAEQbxiBgICd6_Do8BkiIDM3NjYwZDQxMGIyMTQzMDdhOGMyY2YwNWFhMmM2NjVi5899512_20251120114927.389.svg) ### 1\. 工作流程 @@ -134,6 +139,7 @@ DEFAULT_AUDIO_MIME_TYPE = "audio/mpeg" VOICE_FILE_PATH = "voice.mp3" # 用于声音复刻的本地音频文件的相对路径 def create_voice(file_path: str, + url: str, target_model: str = DEFAULT_TARGET_MODEL, preferred_name: str = DEFAULT_PREFERRED_NAME, audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str: @@ -150,9 +156,6 @@ def create_voice(file_path: str, base64_str = base64.b64encode(file_path_obj.read_bytes()).decode() data_uri = f"data:{audio_mime_type};base64,{base64_str}" - - # 以下为华北2（北京）地域的URL，各地域的URL不同。 - url = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization" payload = { "model": "qwen-voice-enrollment", "input": { @@ -198,17 +201,18 @@ class SimpleCallback(OmniRealtimeCallback): if __name__ == '__main__': # 若没有配置环境变量，请用百炼API Key将下行替换为：dashscope.api_key = "sk-xxx" dashscope.api_key = os.getenv("DASHSCOPE_API_KEY") - # 以下为华北2（北京）地域的URL，各地域的URL不同。 - url = "wss://dashscope.aliyuncs.com/api-ws/v1/realtime" + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com" + ws_url = base_url.replace("https://", "wss://") + "/api-ws/v1/realtime" # 1. 声音复刻：创建专属音色 - voice = create_voice(VOICE_FILE_PATH) + voice = create_voice(VOICE_FILE_PATH, url=f"{base_url}/api/v1/services/audio/tts/customization") print(f"声音复刻完成，音色: {voice}") # 2. 使用复刻音色进行实时对话 pya = pyaudio.PyAudio() callback = SimpleCallback(pya) - conv = OmniRealtimeConversation(model=DEFAULT_TARGET_MODEL, callback=callback, url=url) + conv = OmniRealtimeConversation(model=DEFAULT_TARGET_MODEL, callback=callback, url=ws_url) conv.connect() conv.update_session( output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], @@ -234,6 +238,7 @@ if __name__ == '__main__': ``` import com.alibaba.dashscope.audio.omni.*; import com.alibaba.dashscope.exception.NoApiKeyException; +import com.alibaba.dashscope.utils.Constants; import com.google.gson.Gson; import com.google.gson.JsonObject; @@ -258,6 +263,8 @@ public class Main { // 用于声音复刻的本地音频文件的相对路径 private static final String AUDIO_FILE = "voice.mp3"; private static final String AUDIO_MIME_TYPE = "audio/mpeg"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + private static final String BASE_URL = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com"; // 生成 data URI public static String toDataUrl(String filePath) throws IOException { @@ -285,8 +292,7 @@ public class Main { + "}" + "}"; - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - String url = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization"; + String url = BASE_URL + "/api/v1/services/audio/tts/customization"; HttpURLConnection con = (HttpURLConnection) new URL(url).openConnection(); con.setRequestMethod("POST"); con.setRequestProperty("Authorization", "Bearer " + apiKey); @@ -366,8 +372,7 @@ public class Main { // 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key // 若没有配置环境变量，请用百炼API Key将下行替换为：.apikey("sk-xxx") .apikey(System.getenv("DASHSCOPE_API_KEY")) - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - .url("wss://dashscope.aliyuncs.com/api-ws/v1/realtime") + .url(BASE_URL.replace("https://", "wss://") + "/api-ws/v1/realtime") .build(); OmniRealtimeConversation conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() { @@ -466,8 +471,7 @@ def create_voice(file_path: str, base64_str = base64.b64encode(file_path_obj.read_bytes()).decode() data_uri = f"data:{audio_mime_type};base64,{base64_str}" - # 以下为华北2（北京）地域的URL，各地域的URL不同。 - url = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization" + url = dashscope.base_http_api_url + "/services/audio/tts/customization" payload = { "model": "qwen-voice-enrollment", "input": { @@ -492,8 +496,8 @@ def create_voice(file_path: str, raise RuntimeError(f"解析 voice 响应失败: {e}") if __name__ == '__main__': - # 以下为华北2（北京）地域的URL，各地域的URL不同。 - dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 1. 声音复刻：创建专属音色 voice = create_voice(VOICE_FILE_PATH) @@ -556,6 +560,8 @@ public class Main { // 用于声音复刻的本地音频文件的相对路径 private static final String AUDIO_FILE = "voice.mp3"; private static final String AUDIO_MIME_TYPE = "audio/mpeg"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + private static final String BASE_URL = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com"; // 将 PCM 数据写入标准 WAV 文件 public static void writeWav(String path, byte[] pcmData, int sampleRate) throws IOException { @@ -601,8 +607,7 @@ public class Main { + "}" + "}"; - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - String url = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization"; + String url = BASE_URL + "/api/v1/services/audio/tts/customization"; HttpURLConnection con = (HttpURLConnection) new URL(url).openConnection(); con.setRequestMethod("POST"); con.setRequestProperty("Authorization", "Bearer " + apiKey); @@ -650,8 +655,7 @@ public class Main { + "\"stream_options\": {\"include_usage\": true}" + "}"; - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - String url = "https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions"; + String url = BASE_URL + "/compatible-mode/v1/chat/completions"; HttpURLConnection con = (HttpURLConnection) new URL(url).openConnection(); con.setRequestMethod("POST"); con.setRequestProperty("Authorization", "Bearer " + apiKey); @@ -712,13 +716,13 @@ public class Main { - **URL** - 中国内地： + 华北2（北京）： ``` - POST https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization + POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization ``` - 国际： + 新加坡： ``` POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/audio/tts/customization @@ -1024,11 +1028,11 @@ public class Main { ``` # ======= 重要提示 ======= - # 以下为华北2（北京）地域的URL，各地域的URL不同。 + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 # 新加坡地域和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # === 执行时请删除该注释 === - curl -X POST https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization \ + curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -1062,8 +1066,8 @@ public class Main { # 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = "sk-xxx" api_key = os.getenv("DASHSCOPE_API_KEY") - # 以下为华北2（北京）地域的URL，各地域的URL不同。 - url = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization" + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization" payload = { "model": "qwen-voice-enrollment", # 不要修改这个值 @@ -1121,8 +1125,8 @@ public class Main { // 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key // 若没有配置环境变量，请用百炼API Key将下行替换为：String apiKey = "sk-xxx" String apiKey = System.getenv("DASHSCOPE_API_KEY"); - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - String apiUrl = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + String apiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization"; try { // 构造 JSON 请求体（注意内部的引号需转义） @@ -1188,13 +1192,13 @@ public class Main { - **URL** - 中国内地： + 华北2（北京）： ``` - POST https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization + POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization ``` - 国际： + 新加坡： ``` POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/audio/tts/customization @@ -1387,11 +1391,11 @@ public class Main { ``` # ======= 重要提示 ======= - # 以下为华北2（北京）地域的URL，各地域的URL不同。 + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 # 新加坡地域和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # === 执行时请删除该注释 === - curl --location --request POST 'https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization' \ + curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization' \ --header 'Authorization: Bearer $DASHSCOPE_API_KEY' \ --header 'Content-Type: application/json' \ --data '{ @@ -1413,8 +1417,8 @@ public class Main { # 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = "sk-xxx" api_key = os.getenv("DASHSCOPE_API_KEY") - # 以下为华北2（北京）地域的URL，各地域的URL不同。 - url = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization" + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization" payload = { "model": "qwen-voice-enrollment", # 不要修改该值 @@ -1463,8 +1467,8 @@ public class Main { // 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key // 若没有配置环境变量，请用百炼API Key将下行替换为：String apiKey = "sk-xxx" String apiKey = System.getenv("DASHSCOPE_API_KEY"); - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - String apiUrl = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + String apiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization"; // JSON 请求体（旧版本 Java 无 """ 多行字符串） String jsonPayload = @@ -1533,13 +1537,13 @@ public class Main { - **URL** - 中国内地： + 华北2（北京）： ``` - POST https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization + POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization ``` - 国际： + 新加坡： ``` POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/audio/tts/customization @@ -1678,11 +1682,11 @@ public class Main { ``` # ======= 重要提示 ======= - # 以下为华北2（北京）地域的URL，各地域的URL不同。 + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 # 新加坡地域和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # === 执行时请删除该注释 === - curl --location --request POST 'https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization' \ + curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization' \ --header 'Authorization: Bearer $DASHSCOPE_API_KEY' \ --header 'Content-Type: application/json' \ --data '{ @@ -1703,8 +1707,8 @@ public class Main { # 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = "sk-xxx" api_key = os.getenv("DASHSCOPE_API_KEY") - # 以下为华北2（北京）地域的URL，各地域的URL不同。 - url = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization" + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization" voice_to_delete = "yourVoice" # 要删除的音色（替换为真实值） @@ -1752,8 +1756,8 @@ public class Main { // 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key // 若没有配置环境变量，请用百炼API Key将下行替换为：String apiKey = "sk-xxx" String apiKey = System.getenv("DASHSCOPE_API_KEY"); - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - String apiUrl = "https://dashscope.aliyuncs.com/api/v1/services/audio/tts/customization"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + String apiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization"; String voiceToDelete = "yourVoice"; // 要删除的音色（替换为真实值） // 构造 JSON 请求体（字符串拼接，兼容 Java 8） diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/server-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/server-events.md index 1b9a731f..98d52374 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/server-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/server-events.md @@ -162,7 +162,7 @@ VAD检测阈值。 **enable\_search** `_boolean_` -是否启用联网搜索功能。仅 Qwen3.5-Omni-Realtime 模型支持。 +是否启用联网搜索功能。仅 Qwen3.5-Omni-Realtime 系列模型支持。 **search\_options** `_object_` @@ -308,7 +308,7 @@ VAD检测阈值。 **enable\_search** `_boolean_`（可选） -是否启用联网搜索功能。仅 Qwen3.5-Omni-Realtime 模型支持。 +是否启用联网搜索功能。仅 Qwen3.5-Omni-Realtime 系列模型支持。 **search\_options** `_object_`（可选） diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/error-code.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/error-code.md index 59f0dda7..1ef9f9e0 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/error-code.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/error-code.md @@ -213,6 +213,17 @@ AI 助理准确分析出原因，并给出解决方案： - **使用正确的模型名称：**请对照模型列表中的模型名称，检查输入的`model`是否正确。请勿混用开源社区的模型名与百炼模型ID，如应该使用`qwen3-235b-a22b-instruct-2507`，而非`Qwen/Qwen3-235B-A22B-Instruct-2507`。 +### **The product is not activated, please confirm that you have activated products and try again after activation.** + +**原因：** 所调用的模型服务未开通（未激活）。通过 OpenAI 兼容接口调用时，若目标模型未在模型市场开通，网关会返回该错误，错误码为 `invalid_parameter_error`。 + +**解决方案：** + +- **开通模型服务：**请前往阿里云百炼控制台[模型市场](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market)，确认所调用的模型（如 `kimi-k3`）处于已开通状态后再行调用。 + +- **开通百炼服务：**若尚未开通阿里云百炼服务，请先开通百炼服务后再调用。 + + ### The result\_format parameter must be \\"message\\" when enable\_thinking is tru**e** **原因：** 调用思考模式模型，`result_format`参数未设置为`"message"`。 @@ -767,13 +778,13 @@ AI 助理准确分析出原因，并给出解决方案： ### **request timeout after 23 seconds.** -**原因：** 超过23秒未向服务发送数据。该报错信息在使用[实时语音合成（Sambert）](https://help.aliyun.com/zh/model-studio/sambert-speech-synthesis/)、[语音识别（Paraformer）](https://help.aliyun.com/zh/model-studio/paraformer-speech-recognition)和[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)时产生。 +**原因：** 超过23秒未向服务发送数据。该报错信息在使用[实时语音合成（Sambert）](https://help.aliyun.com/zh/model-studio/sambert-speech-synthesis/)、[语音识别（Paraformer）](https://help.aliyun.com/zh/model-studio/paraformer-speech-recognition)和[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)时产生。 **解决方案：** 请检查为什么长时间未向服务器发送数据。如果长时间（超过23秒）不向服务端发送消息，请及时结束任务。 ### **Please ensure input text is valid.** -**原因：** 若您使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，此错误通常是由于未发送待合成文本引起的。可能原因包括：参数遗漏（未为 `text` 参数赋值）或代码异常（导致对 `text` 参数的赋值失败）。 +**原因：** 若您使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，此错误通常是由于未发送待合成文本引起的。可能原因包括：参数遗漏（未为 `text` 参数赋值）或代码异常（导致对 `text` 参数的赋值失败）。 **解决方案：** 请排查代码，确保 `text` 参数被正确赋值并发送。 @@ -816,7 +827,7 @@ AI 助理准确分析出原因，并给出解决方案： ### **\[tts:\]Engine return error code: 418** -**原因：** 使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，请求参数 `voice`（音色）不正确，或 `model`（模型）与 `voice`（音色）版本不匹配。 +**原因：** 使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，请求参数 `voice`（音色）不正确，或 `model`（模型）与 `voice`（音色）版本不匹配。 **解决方案：** @@ -831,7 +842,7 @@ AI 助理准确分析出原因，并给出解决方案： ### **Request voice is invalid!** -**原因：** 若您使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，此错误通常是因为未设置音色。 +**原因：** 若您使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，此错误通常是因为未设置音色。 **解决方案：** 请检查是否对`voice`参数赋值。若您使用[WebSocket API参考](https://help.aliyun.com/zh/model-studio/cosyvoice-websocket-api)，请参照API文档按照正确JSON格式配置参数。 @@ -1015,7 +1026,7 @@ AI 助理准确分析出原因，并给出解决方案： **解决方案：**前往[费用与成本](https://usercenter2.aliyun.com/home)查看是否欠费： -- 未欠费：请确认该 API Key 是否属于当前账号； +- 未欠费：请确认该 API Key 是否属于当前账号。如果账号不存在欠费的情况，可能账户出现异常，详情请联系客服进一步排查。 - 欠费：请及时充值。充值后，系统余额可能存在延迟，请稍等后重试。 @@ -1745,7 +1756,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 - **填写错误**：阿里云百炼的 API Key 以 `sk-` 开头，请确认未误填其他模型提供商的密钥，且复制时未包含多余空格或换行符。 -- **套餐专属 API Key（Coding Plan / Token Plan 团队版）**：Coding Plan 与 Token Plan 团队版均提供以 `sk-sp-` 开头的专属 API Key，**必须配合各自的专属 Base URL 使用**，不可与通用 API Key/Base URL 混用（混用会返回本鉴权错误）。其中 Coding Plan 的专属地址为 https://coding.dashscope.aliyuncs.com/v1；Token Plan 团队版的专属 Base URL 可在控制台**我的订阅**的 API Key 区域查看。请确认同时更新了 API Key 和 Base URL，具体配置方法请分别参见[接入AI工具](https://help.aliyun.com/zh/model-studio/use-coding-plan-in-ai-tools/)与[快速开始](https://help.aliyun.com/zh/model-studio/token-plan-quickstart)。 +- **套餐专属 API Key（Coding Plan / Token Plan 团队版）**：Coding Plan 与 Token Plan 团队版均提供以 `sk-sp-` 开头的专属 API Key，**必须配合各自的专属 Base URL 使用**，不可与通用 API Key/Base URL 混用（混用会返回本鉴权错误）。其中 Coding Plan 的专属地址为 https://coding.dashscope.aliyuncs.com/v1；Token Plan 团队版的专属 Base URL 可在控制台**我的订阅**的 API Key 区域查看。请确认同时更新了 API Key 和 Base URL，具体配置方法请分别参见[接入AI工具](https://help.aliyun.com/zh/model-studio/use-coding-plan-in-ai-tools/)与[快速开始](https://help.aliyun.com/zh/model-studio/token-plan-team-quickstart)。 - **地域不匹配**：API Key 和 Base URL 属于不同的地域，例如使用了华北2（北京）地域的 API Key 和新加坡地域的 Base URL。请确认您使用的 API Key 位于[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/api_key)地域页面或[新加坡](https://modelstudio.console.aliyun.com/?tab=globalset#/efm/api_key)地域页面，或[美国](https://modelstudio.console.aliyun.com/us-east-1)地域页面。各地域对应的 Base URL 如下： @@ -1806,6 +1817,8 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 - 中国站用户请使用**华北2（北京）地域的接入地址；国际站用户请使用新加坡**地域的接入地址。使用[在线调试](https://api.aliyun.com/api/bailian/2023-12-29/CreateIndex)时，确认服务地址正确。 + ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7513215871/p1090718.png) + ## **401-**invalid access token or token expired @@ -1958,6 +1971,8 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 - 请前往模型广场开通模型服务。 +- 如果您通过国际站 API 端点（如 `dashscope-us.aliyuncs.com`）发起调用，请注意不同地域可用的模型列表不同。调用前请确认目标模型是否在该地域可用，部分模型在美国地域需使用带 `-us` 后缀的模型名称（如 `qwen-max-us`）。 + ## **404-**model\_not\_supported @@ -2038,7 +2053,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 - 如需更高调用频率，可参考[限流](https://help.aliyun.com/zh/model-studio/rate-limit)申请提额。 -## **429-**Throttling.RateQuota/LimitRequests/limit\_requests +## **429-**Throttling.RateQuota/LimitRequests/limit\_requests/ResourceExhausted/Too many requests ### **You have exceeded your request limit./Requests rate limit exceeded, please try again later.** /You exceeded your current requests list. @@ -2186,7 +2201,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 **原因：** 语音合成中使用的音色不存在。 -**解决方案：** 请检查`voice`参数，确保指定了正确的音色名称。可用音色请参见[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)。 +**解决方案：** 请检查`voice`参数，确保指定了正确的音色名称。可用音色请参见[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)。 ## **500-**InternalError.FileUpload @@ -2438,7 +2453,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 ### **Cannot resolve symbol 'ttsv2'** -**原因：** 若您使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，出现该问题的原因是DashScope SDK版本过低。 +**原因：** 若您使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，出现该问题的原因是DashScope SDK版本过低。 **解决方案：** 请[安装最新版 DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk#f80a232bb24v7)。 @@ -2458,7 +2473,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 ### **InputRequiredException: Parameter invalid: text is null** -**原因**：使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)时未发送待合成文本。 +**原因**：使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)时未发送待合成文本。 **解决方案：**调用语音合成接口时为 `text` 参数赋值。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/get-api-key.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/get-api-key.md index 3c29a67e..1b74ba58 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/get-api-key.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/get-api-key.md @@ -4,7 +4,7 @@ **说明** -本文介绍的是百炼按量付费的 API Key。如果您使用的是 Token Plan 或 Coding Plan，请使用对应的专属 API Key（以`sk-sp-`开头），获取方式请参见[Token Plan API Key](https://help.aliyun.com/zh/model-studio/token-plan-quickstart#tp04-h-step2)和[Coding Plan 的 API Key](https://help.aliyun.com/zh/model-studio/coding-plan#2531c37fd64f9)。 +本文介绍的是百炼按量付费的 API Key。如果您使用的是 Token Plan 或 Coding Plan，请使用对应的专属 API Key（以`sk-sp-`开头），获取方式请参见[Token Plan API Key](https://help.aliyun.com/zh/model-studio/token-plan-team-quickstart#tp04-h-step2)和[Coding Plan 的 API Key](https://help.aliyun.com/zh/model-studio/coding-plan#2531c37fd64f9)。 ## **获取API Key** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md new file mode 100644 index 00000000..8eaec477 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md @@ -0,0 +1,17 @@ +# AOQ SDK简介 + +阿里云AOQ SDK面向实时多模态场景，帮助开发者快速搭建基于阿里云实时多模态模型的解决方案。 + +欢迎使用AOQ SDK来实现您的业务需求，阿里云AOQ基于阿里云多年以来服务全球客户的深厚技术沉淀，以面向实时多模态场景向全球开发者开放的产品，致力于帮助全球开发者快速高效的搭建基于阿里云的实时多模态模型的解决方案。 + +## **API设计与回调机制** + +阿里云AOQ SDK的API设计遵循状态式API设计原则。客户只需要把想要的状态通过API传至SDK，SDK内部会进行状态判断，并在适当时机达成用户想要的目标状态，同时在执行时通过回调机制及时向客户反馈SDK当前执行状态，使客户使用我们的SDK时享受到无忧的体验。您只需设置期望的状态，SDK便会自动处理并通过回调通知您，无需重复调用，也不需要关心具体的调用时机和场景。 + +## **异常处理机制** + +遇到异常情况时，AOQ SDK会优先尝试内部解决，仅在内部无法恢复时，才会提示客户介入处理。目前这些无法恢复的内部错误主要分为物理限制（如网络问题、音频设备故障、视频设备故障、CPU或内存资源限制）和外部因素（token无效）两类。 + +## **下载指南** + +阿里云AOQ SDK是阿里云自研产品，请通过阿里云官网进行下载，详情请参见[SDK下载](https://help.aliyun.com/zh/model-studio/realtime-sdk-download)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md new file mode 100644 index 00000000..f675adaa --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md @@ -0,0 +1,1179 @@ +# 音频常用功能介绍 + +AOQ Client SDK 提供了完整的音频能力，覆盖音频采集、播放、编解码配置、扬声器管理、文件混音、外部音频流注入、音频帧数据回调等核心场景。本文档基于 Android（Java）、iOS（Objective-C）、Ohos（ArkTS）三个平台的公开 API，对音频常用功能进行统一介绍。 + +## **音频采集** + +音频采集用于打开设备麦克风，将实时音频数据送入 SDK 编码推流管线。SDK 支持两种采集模式： + +- **内部采集**（默认）：SDK 自动管理麦克风设备的打开、录音和关闭。 + +- **外部采集**：由应用自行管理麦克风，采集到的 PCM 数据通过外部音频流接口输入 SDK。 + + +### 配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +isExternal + +bool + +false + +是否使用外部采集模式 + +isVoipMode + +bool + +false + +是否启用 VoIP 模式（硬件 AEC），移动端有效，采集播放参数先到为准 + +channel + +int + +1 + +采集通道数，支持 1（单声道）/ 2（立体声） + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +开启采集 + +`startAudioCapture(config)` + +`startAudioCapture:config:` + +`startAudioCapture(config)` + +关闭采集 + +`stopAudioCapture()` + +`stopAudioCapture` + +`stopAudioCapture()` + +静音/取消静音 + +`muteAudioCapture(mute)` + +`muteAudioCapture:` + +`muteAudioCapture(mute)` + +### 使用示例 + +**Android** + +``` +AoqAudioCaptureConfig config = new AoqAudioCaptureConfig(); +config.isVoipMode = true; +config.channel = 1; +engine.startAudioCapture(config); +``` + +**iOS** + +``` +AoqAudioCaptureConfig *config = [[AoqAudioCaptureConfig alloc] init]; +config.isVoipMode = YES; +config.channel = 1; +[engine startAudioCapture:config]; +``` + +**Ohos** + +``` +const config: AoqAudioCaptureConfig = { isVoipMode: true, channel: 1 }; +engine.startAudioCapture(config); +``` + +## **音频播放** + +音频播放用于将接收到的远端音频数据渲染到本地扬声器或耳机。SDK 支持播放暂停/恢复（带淡入淡出）、打断当前轮音频通话等高级控制。 + +### 配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +isVoipMode + +bool + +false + +是否启用 VoIP 模式（硬件AEC），移动端有效，采集播放参数先到为准 + +isDefaultSpeaker + +bool + +true + +是否默认使用扬声器（移动端有效，非VoIP时无效） + +isExternal + +bool + +false + +是否使用外部播放模式 + +channel + +int + +1 + +播放通道数，支持 1（单声道）/ 2（立体声） + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +开始播放 + +`startAudioPlayer(config)` + +`startAudioPlayer:config:` + +`startAudioPlayer(config)` + +停止播放 + +`stopAudioPlayer()` + +`stopAudioPlayer` + +`stopAudioPlayer()` + +暂停播放 + +`pauseAudioPlayer(fadeMs)` + +`pauseAudioPlayer:` + +`pauseAudioPlayer(fadeMs)` + +恢复播放 + +`resumeAudioPlayer(fadeMs)` + +`resumeAudioPlayer:` + +`resumeAudioPlayer(fadeMs)` + +打断通话 + +`interruptAudioPlayer(trackType, fadeMs)` + +`interruptAudioPlayer:fadeMs:` + +`interruptAudioPlayer(trackType, fadeMs)` + +**说明** + +**fadeMs 参数**：暂停和恢复播放时的淡入/淡出时长（毫秒），设为 0 则立即切换。 + +## **扬声器管理** + +控制音频输出设备在扬声器和听筒之间切换。 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +切换扬声器 + +`enableSpeakerphone(enable)` + +`enableSpeakerphone:` + +`enableSpeakerphone(enable)` + +查询扬声器状态 + +`isSpeakerphoneEnabled()` + +`isSpeakerphoneEnabled` + +`isSpeakerphoneEnabled()` + +**说明** + +需要在 VoIP 模式下才允许切换，非 VoIP 时，`enableSpeakerphone` 调用有 OnError(AoqECAudioDeviceEarpieceRequiresVoipMode) 错误通知。 + +**说明** + +**iOS 特殊行为**：iPad 设备只有扬声器模式；当 AVAudioSession 不是 PlayAndRecord 类别时，也始终返回 YES。 + +## **音频编解码配置** + +设置音频上行（编码器）和下行（解码器）的编码格式、采样率、声道数和码率。表示推流/拉流的格式。 + +### 配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +Audio + +音频轨道类型，当前只支持一条音频流 + +codecType + +AoqEncoderType + +AudioPCM + +编码类型：AudioPCM(1) 或 AudioOpus(2) + +sampleRate + +int + +48000 + +采样率，Opus 支持 8K/16K/48K，PCM 支持 8K/16K/32K/48K + +channel + +int + +1 + +声道数，支持 1（单声道）/ 2（立体声） + +bitrate + +int + +32000 + +码率（bps） + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +设置编码参数 + +`setAudioEncoderConfig(config)` + +`setAudioEncoderConfig:` + +`setAudioEncoderConfig(config)` + +设置解码参数 + +`setAudioDecoderConfig(config)` + +`setAudioDecoderConfig:` + +`setAudioDecoderConfig(config)` + +### 支持的编码格式 + +**枚举值** + +**数值** + +**说明** + +AoqEncoderTypeAudioPCM + +1 + +PCM 裸音频 + +AoqEncoderTypeAudioOpus + +2 + +Opus 编码 + +## **音频文件混音** + +支持将本地音频文件混入当前音频流中一起推流和/或本地播放。每个音频文件通过业务自分配的 `fileId` 标识，可同时管理多个文件实例。 + +### 混音配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +fileName + +String + +\- + +音频文件路径（含文件名） + +cycles + +int + +\-1 + +循环次数，-1 表示无限循环 + +startPosMs + +long + +0 + +起始播放位置（毫秒） + +publishVolume + +int + +100 + +推流音量 \[0-100\] + +playoutVolume + +int + +100 + +本地播放音量 \[0-100\] + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +开始播放 + +`startAudioFile(fileId, config)` + +`startAudioFile:config:` + +`startAudioFile(fileId, config)` + +停止播放 + +`stopAudioFile(fileId)` + +`stopAudioFile:` + +`stopAudioFile(fileId)` + +暂停 + +`pauseAudioFile(fileId)` + +`pauseAudioFile:` + +`pauseAudioFile(fileId)` + +恢复 + +`resumeAudioFile(fileId)` + +`resumeAudioFile:` + +`resumeAudioFile(fileId)` + +获取文件时长 + +`getAudioFileDuration(fileId)` + +`getAudioFileDuration:` + +`getAudioFileDuration(fileId)` + +获取当前位置 + +`getAudioFileCurrentPosition(fileId)` + +`getAudioFileCurrentPosition:` + +`getAudioFileCurrentPosition(fileId)` + +设置播放位置 + +`setAudioFilePositionMillis(fileId, pos)` + +`setAudioFilePositionMillis:positionMillis:` + +`setAudioFilePositionMillis(fileId, pos)` + +设置音量 + +`setAudioFileVolume(fileId, type, vol)` + +`setAudioFileVolume:type:volume:` + +`setAudioFileVolume(fileId, type, vol)` + +获取音量 + +`getAudioFileVolume(fileId, type)` + +`getAudioFileVolume:type:` + +`getAudioFileVolume(fileId, type)` + +**说明** + +**音量方向（type）**：`AoqAudioStreamPublish(0)` 控制推流音量；`AoqAudioStreamPlayout(1)` 控制本地播放音量。 + +### 状态回调 + +**状态码** + +**数值** + +**说明** + +AoqAudioFileNone + +0 + +初始状态 + +AoqAudioFileStarted + +1 + +已开始播放 + +AoqAudioFileStopped + +2 + +已停止 + +AoqAudioFilePaused + +3 + +已暂停 + +AoqAudioFileResumed + +4 + +已恢复 + +AoqAudioFileEnded + +5 + +播放结束 + +AoqAudioFileBuffering + +6 + +缓冲中 + +AoqAudioFileBufferingEnd + +7 + +缓冲结束 + +AoqAudioFileFailed + +8 + +播放失败 + +## **外部音频流** + +外部音频流允许将应用生成的 PCM 音频数据注入到 SDK 的音频管线中，支持推流和/或本地播放。典型场景包括 TTS 语音合成输出、AI 模型音频输出、背景音效等。每个外部音频流通过业务自分配的 `streamId` 标识。 + +### 配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +Audio + +音频轨道类型 + +codecType + +AoqEncoderType + +AudioPCM + +音频流格式 + +channels + +int + +1 + +声道数 + +sampleRate + +int + +48000 + +采样率，支持 8/12/16/24/32/44.1/48/64/88.2/96/176.4/192K + +playoutVolume + +int + +100 + +本地播放音量 \[0-100\] + +publishVolume + +int + +100 + +推流音量 \[0-100\] + +maxBufferDuration + +int + +600000 + +最大缓冲时长（毫秒），取值范围 \[100, ~\]，超过时 Push 失败 + +enable3A + +bool + +false + +输入 PCM 是否经过 3A 处理 + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +新增外部音频流 + +`addAudioExternalStream(streamId, config)` + +`addAudioExternalStream:config:` + +`addAudioExternalStream(streamId, config)` + +输入音频数据 + +`pushAudioExternalStreamData(streamId, data)` + +`pushAudioExternalStreamData:data:` + +`pushAudioExternalStreamData(streamId, data)` + +设置音量 + +`setAudioExternalStreamVolume(streamId, type, vol)` + +`setAudioExternalStreamVolume:type:volume:` + +`setAudioExternalStreamVolume(streamId, type, vol)` + +获取音量 + +`getAudioExternalStreamVolume(streamId, type)` + +`getAudioExternalStreamVolume:type:` + +`getAudioExternalStreamVolume(streamId, type)` + +清空缓存 + +`clearAudioExternalStreamBuffer(streamId, fadeoutMs)` + +`clearAudioExternalStreamBuffer:fadeoutMs:` + +`clearAudioExternalStreamBuffer(streamId, fadeoutMs)` + +移除流 + +`removeAudioExternalStream(streamId)` + +`removeAudioExternalStream:` + +`removeAudioExternalStream(streamId)` + +### Push 数据最佳实践 + +- 需要循环调用 `pushAudioExternalStreamData`，保证数据 push 成功 + +- 返回错误码 110（缓冲区满）时短暂 Sleep 30ms 后重试，不要丢弃数据 + +- 引擎退出前先停止推送循环，再调用 `removeAudioExternalStream` + +- 实时采集每帧 10ms 长，有数据就调用 push；从文件解析每帧 40ms 长，间隔 30ms 调用 push 一次 + + +## **音频帧数据回调** + +音频帧回调允许开发者在音频管线的不同位置获取原始 PCM 数据，用于音频分析、自定义处理、录制等场景。 + +### 支持的数据源位置 + +**数据源** + +**枚举值** + +**说明** + +Captured + +0 + +采集后的原始音频数据（未经 3A 处理） + +ProcessCaptured + +1 + +经过 3A 处理后的音频数据，需要 Connect 成功后才回调数据 + +Publish + +2 + +即将推流的音频数据（需要 Connect 成功） + +Playback + +3 + +即将播放的音频数据（远端下行） + +### 回调配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +sampleRate + +int + +48000 + +回调音频的采样率 + +channels + +int + +1 + +回调音频的声道数，支持 1/2 + +mode + +AoqAudioObserverMode + +ReadOnly + +只读(0)/读写(1) 模式 + +### 使用步骤 + +1. **注册观察者**：调用 `setAudioFrameObserver` 设置音频帧回调监听器 + +2. **启用数据源**：调用 `enableAudioFrameObserver` 选择需要监听的数据源位置，开启回调 + +3. **处理回调数据**：在回调函数中获取 PCM 数据 + + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +注册观察者 + +`setAudioFrameObserver(listener)` + +`setAudioFrameObserver:` + +`setAudioFrameObserver(observer)` + +启用回调 + +`enableAudioFrameObserver(enabled, source, config)` + +`enableAudioFrameObserver:audioSource:config:` + +`enableAudioFrameObserver(enabled, source, config)` + +### 回调方法 + +**回调** + +**Android** + +**iOS** + +**Ohos** + +采集数据 + +`onCapturedAudioFrame(frame)` + +`onCapturedAudioFrame:` + +`onCapturedAudioFrame(frame)` + +3A 后数据 + +`onProcessCapturedAudioFrame(frame)` + +`onProcessCapturedAudioFrame:` + +`onProcessCapturedAudioFrame(frame)` + +推流数据 + +`onPublishAudioFrame(trackType, frame)` + +`onPublishAudioFrame:frame:` + +`onPublishAudioFrame(trackType, frame)` + +播放数据 + +`onPlaybackAudioFrame(frame)` + +`onPlaybackAudioFrame:` + +`onPlaybackAudioFrame(frame)` + +## **音频状态与路由** + +SDK 自动监测音频设备的状态变化和路由切换，并通过回调通知应用层。 + +### 设备状态码 + +**状态码** + +**值** + +**说明** + +AoqAudioDeviceNone + +0 + +初始状态 + +RecordStarting + +1 + +采集启动中 + +RecordStarted + +2 + +采集已启动 + +RecordStopping + +3 + +采集停止中 + +RecordStopped + +4 + +采集已停止 + +RecordFail + +5 + +采集失败 + +PlayStarting + +6 + +播放启动中 + +PlayStarted + +7 + +播放已启动 + +PlayStopping + +8 + +播放停止中 + +PlayStopped + +9 + +播放已停止 + +PlayFail + +10 + +播放失败 + +### 设备路由类型 + +**路由** + +**值** + +**说明** + +Default + +0 + +默认 + +Headset + +1 + +有线耳机 + +Earpiece + +2 + +听筒 + +HeadsetNoMic + +3 + +无麦克风耳机 + +SpeakerPhone + +4 + +扬声器 + +Usb + +5 + +USB 设备 + +Bluetooth + +6 + +蓝牙 SCO + +BluetoothA2dp + +7 + +蓝牙 A2DP + +### 回调对照 + +**回调** + +**Android** + +**iOS** + +**Ohos** + +设备状态变化 + +`onAudioDeviceStateChanged(state)` + +`onAudioDeviceStateChanged:` + +`onAudioDeviceStateChanged(state, reason)` + +路由变化 + +`onAudioDeviceRouteChanged(routeType)` + +`onAudioDeviceRouteChanged:` + +`onAudioDeviceRouteChanged(routeType)` + +设备中断 + +`onAudioDeviceInterrupted(interrupt)` + +`onAudioDeviceInterrupted:` + +`onAudioDeviceInterrupted(interrupt)` + +文件状态 + +`onAudioFileState(state)` + +`onAudioFileState:` + +`onAudioFileState(fileId, stateCode, errorCode)` + +## **音频错误码与警告码** + +### 音频错误码 + +**错误码** + +**值** + +**说明** + +AoqErrorCodeAudio + +100 + +通用音频错误 + +AudioExternalBufferFull + +110 + +外部缓冲区已满 + +AudioDevice + +120 + +设备通用错误 + +RecordingAuthFailed + +121 + +麦克风权限失败 + +RecordingOccupied + +122 + +麦克风被占用 + +RecordingBackgroundStart + +123 + +后台启动录音 + +RecordingStartFail + +124 + +录音启动失败 + +PlayoutOccupied + +125 + +播放设备被占用 + +PlayoutBackgroundStart + +126 + +后台启动播放 + +PlayoutStartFail + +127 + +播放启动失败 + +EarpieceRequiresVoipMode + +128 + +听筒需要启用 VoIP 模式 + +### 音频警告码 + +**警告码** + +**值** + +**说明** + +AoqWCAudio + +100 + +通用音频警告 + +AudioHowling + +101 + +啸叫检测 + +AudioDevice + +120 + +设备通用警告 + +MicEnumerateError + +121 + +麦克风枚举错误 + +MicStartTimeout + +122 + +麦克风启动超时 + +RecordingError + +123 + +录音错误 + +SpeakerEnumerateError + +124 + +扬声器枚举错误 + +SpeakerStartTimeout + +125 + +扬声器启动超时 + +PlayoutError + +126 + +播放错误 + +## **iOS 专有：AVAudioSession 控制** + +iOS 平台提供了 `setAudioSessionRestriction` 接口，可精细控制 SDK 对系统 AVAudioSession 的管理权限。 + +**控制项** + +**说明** + +SetCategory + +SDK 是否有权设置 Session 类别 + +ConfigureSession + +SDK 是否有权配置 Session 参数 + +DeactivateSession + +SDK 是否有权停用 Session + +ActivateSession + +SDK 是否有权激活 Session + +通过按位组合传入 restriction 值，可限制 SDK 对 AVAudioSession 的控制范围，避免与应用层其他音频组件冲突。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md new file mode 100644 index 00000000..433f6552 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md @@ -0,0 +1,171 @@ +# 连接状态管理 + +本章节介绍 AOQ Client SDK 的连接状态机及其对应的 API 调用。 + +## **连接状态图** + +下图描述了 AOQ Client SDK 的连接状态迁移关系： + +AOQ Client SDK 连接状态迁移图 + +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3811364871/p1088081.svg) + +## **状态说明** + +**状态** + +**枚举值** + +**说明** + +链接中（Connecting） + +1 + +调用 `connect` 后进入，正在与 AI Service 建立连接 + +已连接（Connected） + +2 + +连接建立成功，可正常收发音视频和数据消息 + +失败（Failed） + +3 + +连接异常（鉴权失败/超时/服务端拒绝等），SDK 内部会自动迁移到已断开 + +已断开（Disconnected） + +0 + +初始状态/主动断开/异常断开后的终态 + +## **状态迁移规则** + +1. **App 调用** `connect` → 进入 `Connecting` 状态。 + +2. **链接成功** → 从 `Connecting` 迁移到 `Connected`。 + +3. **链接异常** → 从 `Connecting` 迁移到 `Failed`，随后 SDK 自动迁移到 `Disconnected`。 + +4. **链接异常** → 从 `Connected` 迁移到 `Failed`，随后 SDK 自动迁移到 `Disconnected`。 + +5. **App 调用** `disconnect` → 从 `Connected` 迁移到 `Disconnected`。 + + +**说明**：`Failed` 是瞬态，SDK 触发 `onConnectionStatusChange(Failed)` 后会自动迁移到 `Disconnected`，业务层无需手动调用 disconnect。 + +## **connect API** + +调用 `connect` 发起与 AI Service 的连接，传入由 AppServer allocate 接口返回的鉴权凭证。 + +### **方法签名** + +**Android：** + +``` +public abstract int connect(@NonNull AoqConnectConfig config); +``` + +**iOS：** + +``` +- (int)connect:(AoqConnectConfig * _Nonnull)config; +``` + +**Ohos：** + +``` +connect(config: AoqConnectConfig): number; +``` + +**返回值：**`0` 表示调用成功（异步建连）；`< 0` 表示失败。 + +### **行为说明** + +- 调用后触发 `onConnectionStatusChange(connecting)` 回调。 + +- 如果链接成功时触发 `onConnectionStatusChange(connected)` 回调。 + +- 如果链接失败时触发 `onConnectionStatusChange(failed)` 回调。 + + +## **disconnect API** + +调用 `disconnect` 主动断开与 AI Service 的连接。 + +### **方法签名** + +**Android：** + +``` +public abstract int disconnect(); +``` + +**iOS：** + +``` +- (int)disconnect; +``` + +**Ohos：** + +``` +disconnect(): number; +``` + +**返回值：**`0` 表示成功；`< 0` 表示失败。 + +### **行为说明** + +- 调用后触发 `onConnectionStatusChange(Disconnected)` 回调。 + +- 引擎不会自动释放，可重新调用 `connect` 进行重连。 + +- 未连接状态下调用 `disconnect` 是安全的，返回 0。 + + +## **onConnectionStatusChange 回调** + +连接状态变化时，SDK 通过此回调通知业务层。 + +**Android：** + +``` +public void onConnectionStatusChange( + @NonNull AoqClientEngine.AoqConnectionStatus status) {} +``` + +**iOS：** + +``` +- (void)onConnectionStatusChange:(AoqConnectionStatus)status; +``` + +**Ohos：** + +``` +onConnectionStatusChange?: (status: AoqConnectionStatus) => void; +``` + +### **示例代码** + +``` +func onConnectionStatusChange(_ status: AoqConnectionStatus) { + switch status { + case .connecting: + print("正在连接...") + case .connected: + print("连接成功") + // 连接成功后可发送 session.update + case .failed: + print("连接失败") + // SDK 会自动迁移到 disconnected，无需手动 disconnect + case .disconnected: + print("已断开") + // 可根据业务决定是否重连 + } +} +``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md new file mode 100644 index 00000000..670fc6f0 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md @@ -0,0 +1,253 @@ +# 自定义音频采集 + +介绍如何使用 AOQ Client SDK 实现自定义音频采集功能，包括外部音频流的添加、PCM 数据推送和管理。 + +## **功能介绍** + +AOQ Client SDK 内部音频模块可满足应用中对基本音频功能的需求，但在特定场景中，SDK 内部的音频采集模块可能无法满足开发需求，需要实现自定义音频采集功能，例如： + +- 解决音频采集设备被占用问题。 + +- 需要从定制的采集系统、音频文件中获取音频数据后交给 SDK 传输。 + +- 需要将 AI TTS 生成的音频数据通过 SDK 推流传输。 + + +AOQ Client SDK 支持灵活的自定义采集功能，允许用户根据业务场景自行管理音频设备与音频源。外部音频流的数据会与内部采集的音频数据混音后一起推流发送。 + +## **示例代码** + +暂无 + +## **前提条件** + +- 已创建引擎实例（调用 `createEngine`）。 + +- 已成功连接服务器（`onConnectionStatusChange` 回调状态为 `AoqConnectionStatusConnected`）。 + + +## **功能实现** + +### **1\. 打开或关闭音频采集** + +需要先开启音频采集，外部音频流输入的数据会与内部采集数据混音后一起推流。如果不需要内部麦克风采集，可以设置 `isExternal=true` 关闭内部采集设备。 + +``` +// 方式一：开启内部采集，外部音频流数据会与麦克风数据混音推流 +AoqClientEngine.AoqAudioCaptureConfig config = new AoqClientEngine.AoqAudioCaptureConfig(); +config.isExternal = false; // 使用内部麦克风采集 +config.isVoipMode = false; +engine.startAudioCapture(config); + +// 方式二：关闭内部采集，仅推送外部音频流数据 +AoqClientEngine.AoqAudioCaptureConfig config = new AoqClientEngine.AoqAudioCaptureConfig(); +config.isExternal = true; // 不打开麦克风，由外部音频流提供数据 +engine.startAudioCapture(config); +``` + +### **2\. 连接成功后，添加外部音频流** + +在 `onConnectionStatusChange` 回调状态变为 `AoqConnectionStatusConnected` 后，调用 `addAudioExternalStream` 添加外部音频流。需要指定一个唯一的 `streamId` 用于后续推送数据和管理。 + +如果需要音频 3A 处理（回声消除、噪声抑制、自动增益），请配置 `AoqAudioExternalStreamConfig` 中的 `enable3A` 参数。 + +``` +// 在 onConnectionStatusChange 回调中确认连接成功后添加 +@Override +public void onConnectionStatusChange(AoqClientEngine.AoqConnectionStatus status) { + if (status == AoqClientEngine.AoqConnectionStatus.AoqConnectionStatusConnected) { + addExternalAudioStream(); + } +} + +private void addExternalAudioStream() { + AoqClientEngine.AoqAudioExternalStreamConfig config = new AoqClientEngine.AoqAudioExternalStreamConfig(); + config.sampleRate = 48000; // 采样率，需与实际音频数据一致 + config.channels = 1; // 声道数 + config.publishVolume = 100; // 推流音量 [0-100] + config.playoutVolume = 0; // 本地播放音量 [0-100]，0 表示不本地播放 + config.maxBufferDuration = 1000; // 最大缓冲时长（毫秒） + config.enable3A = true; // 是否对输入 PCM 进行 3A 处理 + + String streamId = "external_audio_1"; + int ret = engine.addAudioExternalStream(streamId, config); + if (ret == 0) { + mExternalStreamId = streamId; + } +} +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +AoqTrackTypeAudio + +音频轨道类型 + +codecType + +AoqEncoderType + +AoqEncoderTypeAudioPCM + +音频流格式 + +channels + +int + +1 + +声道数 + +sampleRate + +int + +48000 + +采样率（Hz） + +playoutVolume + +int + +100 + +播放音量 \[0-100\] + +publishVolume + +int + +100 + +推流音量 \[0-100\] + +maxBufferDuration + +int + +1000 + +最大缓冲时长（毫秒） + +enable3A + +boolean + +false + +是否对输入 PCM 进行 3A 处理 + +### **3\. 实现自采集模块或从文件获取 PCM 数据** + +自定义采集功能需要根据业务场景自行采集并处理音频数据，之后将数据传入 SDK 进行传输。常见的数据来源： + +- **麦克风采集**：通过 Android AudioRecord 采集 PCM 数据。 + +- **文件读取**：从本地 PCM/WAV 音频文件中解析获取 PCM 数据。 + +- **AI TTS**：从语音合成引擎获取 PCM 数据。 + +- **网络流**：从网络音频流中解码获取 PCM 数据。 + + +音频数据需要为 PCM 格式，并记录对应的采样率、声道数等参数，用于构造 `AoqAudioFrameData` 对象。 + +### **4\. 通过外部音频流 ID 推送音频数据到 SDK** + +调用 `pushAudioExternalStreamData` 接口，将采集到的 PCM 音频数据传入 SDK。 + +- 从硬件采集：建议采集 10ms 为一帧数据，采集到数据就 push 给 SDK。 + +- 从文件解析：建议 40ms 为一帧数据，每 push 一帧 Sleep 30ms 后 push 下一帧。 + +- 需要维护一个 `running` 标记，当引擎退出或 stream ID 被删除时退出推送循环。 + + +``` +// 成员变量：控制推送循环的运行标记 +private volatile boolean mPushRunning = false; + +// 推送单帧音频数据 + +private void pushAudioData(byte[] audioData, int bytesRead) { + + if (engine == null || mExternalStreamId == null || bytesRead <= 0) { + return; + } + + int channels = 1; + int bytesPerSample = 2; // 16bit PCM + int sampleRate = 48000; + + // 构造音频帧数据 + AoqClientEngine.AoqAudioFrameData frameData = new AoqClientEngine.AoqAudioFrameData(); + frameData.dataPtr = audioData; + frameData.dataSize = bytesRead; + frameData.numOfSamples = bytesRead / (channels * bytesPerSample); + frameData.bytesPerSample = bytesPerSample; + frameData.numOfChannels = channels; + frameData.samplesPerSec = sampleRate; + + // 推送数据，处理缓冲区满的情况 + int ret; + final int WAIT_MS = 30; + + do { + // 检查运行标记和 stream ID 是否仍有效 + if (!mPushRunning || mExternalStreamId == null) { + break; + } + ret = engine.pushAudioExternalStreamData(mExternalStreamId, frameData); + if (ret == 110) { // AoqErrorCodeAudioExternalBufferFull + try { + Thread.sleep(WAIT_MS); + } catch (InterruptedException e) { + break; + } + } else { + break; + } + } while (true); +} +``` + +**注意事项：** + +- 需要在连接成功且添加外部音频流之后再开始推送数据。 + +- 需要按照数据的实际长度设置 `AoqAudioFrameData` 的 `numOfSamples`。 + +- 调用 `pushAudioExternalStreamData` 时，可能出现内部缓冲区满（错误码 110）而导致失败，需要等待重试。 + +- 实时采集建议 10ms 一帧数据 push，有数据就调用 push，注意处理内部缓冲区满（错误码 110）。 + +- 从文件解析建议 40ms 一帧数据，间隔 30ms 调用 push，注意处理内部缓冲区满（错误码 110）。 + +- 引擎退出（`destroy`）或 stream ID 被移除前，必须先设置 `mPushRunning = false` 停止推送循环，避免在已释放的资源上操作。 + + +### **5\. 移除外部音频流** + +当不再需要发布自定义采集的音频时，先停止推送循环，再调用 `removeAudioExternalStream` 接口移除外部音频流。 + +``` +// 先停止推送 +stopPushAudio(); +// 再移除外部音频流 +engine.removeAudioExternalStream(mExternalStreamId); +mExternalStreamId = null; +``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md new file mode 100644 index 00000000..1d94d8e7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md @@ -0,0 +1,235 @@ +# 自定义音频播放 + +AOQ Client SDK 支持自定义音频播放功能，通过音频帧回调机制将解码后的 PCM 数据回调给应用层，由开发者自行实现音频渲染播放。 + +## **功能介绍** + +AOQ Client SDK 内部音频模块默认会将接收到的远端音频数据通过系统扬声器/听筒播放，但在特定场景中，SDK 内部的音频播放模块可能无法满足开发需求，需要实现自定义音频播放功能，例如： + +- 需要将接收到的音频数据输出到自定义的播放设备或音频处理管线。 + +- 需要对接收到的音频数据进行二次处理（如 AI 语音识别、音效处理等）。 + +- 解决音频播放设备被占用的问题。 + + +AOQ Client SDK 支持灵活的自定义播放功能，通过音频帧回调机制，将解码后的 PCM 数据回调给应用层，由开发者自行实现音频渲染播放。 + +## **示例代码** + +暂无 + +## **前提条件** + +- 已创建引擎实例（调用 `createEngine`）。 + +- 已成功连接服务器（`onConnectionStatusChange` 回调状态为 `AoqConnectionStatusConnected`）。 + + +## **功能实现** + +### **1\. 开启音频播放（外部模式）** + +调用 `startAudioPlayer` 时设置 `isExternal=true`，关闭 SDK 内部的音频渲染设备，由应用层自行处理音频播放。 + +``` +AoqClientEngine.AoqAudioPlaybackConfig config = new AoqClientEngine.AoqAudioPlaybackConfig(); +config.isExternal = true; // 关闭 SDK 内部播放，由应用层自行渲染 +config.channel = 1; // 声道数 +engine.startAudioPlayer(config); +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +isVoipMode + +boolean + +false + +是否启用 VoIP 模式（硬件AEC），移动端有效 + +isDefaultSpeaker + +boolean + +true + +是否默认扬声器，移动端有效 + +isExternal + +boolean + +false + +是否外部播放模式，true 时 SDK 不打开播放设备 + +channel + +int + +1 + +声道数 + +### **2\. 设置音频帧回调监听** + +调用 `setAudioFrameObserver` 设置音频帧数据回调监听器，实现 `onPlaybackAudioFrame` 回调方法接收播放 PCM 数据。 + +``` +engine.setAudioFrameObserver(new AoqClientListener.AoqAudioFrameListener() { + @Override + public void onPlaybackAudioFrame(@NonNull AoqClientEngine.AoqAudioFrameData frame) { + // 在此处理接收到的播放音频数据 + // frame.dataPtr: PCM 数据 + // frame.numOfSamples: 采样点数 + // frame.numOfChannels: 声道数 + // frame.samplesPerSec: 采样率 + // frame.bytesPerSample: 每采样点字节数 + playPcmData(frame); + } +}); +``` + +### **3\. 开启播放数据回调** + +调用 `enableAudioFrameObserver` 开启播放位置的音频帧回调，指定数据源为 `AoqAudioSourcePlayback`。 + +``` +AoqClientEngine.AoqAudioObserverConfig observerConfig = new AoqClientEngine.AoqAudioObserverConfig(); +observerConfig.sampleRate = 48000; // 回调音频采样率 +observerConfig.channels = 1; // 回调音频声道数 +observerConfig.mode = AoqClientEngine.AoqAudioObserverMode.AoqAudioObserverModeReadOnly; // 只读模式 + +engine.enableAudioFrameObserver( + true, // 开启回调 + AoqClientEngine.AoqAudioSource.AoqAudioSourcePlayback, // 播放数据源 + observerConfig +); +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +sampleRate + +int + +48000 + +回调音频采样率（Hz） + +channels + +int + +1 + +回调音频声道数 + +mode + +AoqAudioObserverMode + +AoqAudioObserverModeReadOnly + +读写模式 + +### **4\. 实现自定义音频渲染** + +在 `onPlaybackAudioFrame` 回调中接收到 PCM 数据后，由应用层自行实现音频渲染播放。常见的实现方式： + +- **Android AudioTrack**：通过 AudioTrack 将 PCM 数据写入系统音频设备播放。 + +- **AI 语音识别**：将 PCM 数据传入 ASR 引擎进行语音识别。 + +- **音效处理**：对 PCM 数据进行音效处理后再播放。 + +- **文件存储**：将接收到的音频数据保存到本地文件。 + + +``` +// 示例：使用 Android AudioTrack 播放 +private AudioTrack mAudioTrack; +private volatile boolean mPlayRunning = false; + +private void initAudioTrack(int sampleRate, int channels) { + int channelConfig = (channels == 2) + ? AudioFormat.CHANNEL_OUT_STEREO + : AudioFormat.CHANNEL_OUT_MONO; + int bufferSize = AudioTrack.getMinBufferSize( + sampleRate, channelConfig, AudioFormat.ENCODING_PCM_16BIT); + + mAudioTrack = new AudioTrack( + AudioManager.STREAM_VOICE_CALL, + sampleRate, + channelConfig, + AudioFormat.ENCODING_PCM_16BIT, + bufferSize, + AudioTrack.MODE_STREAM); + mAudioTrack.play(); + mPlayRunning = true; +} + +private void playPcmData(AoqClientEngine.AoqAudioFrameData frame) { + if (!mPlayRunning || mAudioTrack == null) { + return; + } + if (frame.dataPtr != null && frame.dataSize > 0) { + mAudioTrack.write(frame.dataPtr, 0, frame.dataSize); + } +} +``` + +**注意事项：** + +- `onPlaybackAudioFrame` 回调在 SDK 内部线程触发，回调中的 `frame.dataPtr` 仅在回调期间有效，异步使用需自行拷贝。 + +- AudioTrack.write 是阻塞操作，在回调中直接写入即可，SDK 内部会按节奏回调。 + +- 需要维护 `mPlayRunning` 标记，当引擎退出或停止播放时退出处理逻辑。 + + +### **5\. 停止自定义播放** + +当不再需要自定义播放时，先关闭音频帧回调，再停止播放设备，释放 AudioTrack 资源。 + +``` +// 1. 关闭播放位置的音频帧回调 +AoqClientEngine.AoqAudioObserverConfig observerConfig = new AoqClientEngine.AoqAudioObserverConfig(); +engine.enableAudioFrameObserver( + false, // 关闭回调 + AoqClientEngine.AoqAudioSource.AoqAudioSourcePlayback, + observerConfig +); + +// 2. 移除音频帧回调监听 +engine.setAudioFrameObserver(null); + +// 3. 停止 SDK 音频播放 +engine.stopAudioPlayer(); + +// 4. 释放 AudioTrack 资源 +mPlayRunning = false; +if (mAudioTrack != null) { + mAudioTrack.stop(); + mAudioTrack.release(); + mAudioTrack = null; +} +``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md new file mode 100644 index 00000000..358f530f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md @@ -0,0 +1,647 @@ +# 自定义视频输入 + +介绍 AOQ Client SDK 自定义视频输入的两种模式：原始帧模式和编码帧模式，以及各模式的配置方法和示例代码。 + +## **功能介绍** + +AOQ Client SDK 内部视频模块可满足应用中对基本视频功能的需求，但在特定场景中，SDK 内部的视频采集模块可能无法满足开发需求，需要实现自定义视频采集功能，例如： + +- 解决摄像头设备被占用或不兼容问题。 + +- 需要从定制的采集系统、视频文件中获取视频数据后交给 SDK 传输。 + +- 需要将 AI 生成的画面、屏幕录制、虚拟摄像头等内容通过 SDK 推流传输。 + + +AOQ Client SDK 支持两种自定义视频采集模式： + +- **原始帧模式**：自行采集原始视频帧（BGRA、I420、NV12、NV21 等格式），通过 `pushExternalVideoCapturedFrame` 推送给 SDK 进行编码和传输。SDK 内部完成编码、传输等完整流程。 + +- **编码帧模式**：自行完成视频编码（目前支持 JPEG），通过 `pushExternalVideoEncodedFrame` 直推已编码数据给 SDK，跳过 SDK 内部编码器，直接打包发送。 + + +## **示例代码** + +暂无 + +## **前提条件** + +- 已创建引擎实例（调用 `createEngine`）。 + +- 已成功连接服务器（`onConnectionStatusChange` 回调状态为 `AoqConnectionStatusConnected`）。 + + +## **功能实现** + +根据业务场景选择以下两种模式之一。两种模式不可混用：同一时间只能使用其中一种推送接口。 + +## **模式一：原始帧模式** + +自行采集原始视频帧（BGRA、I420、NV12、NV21 等格式），推送给 SDK 进行编码和传输。SDK 内部完成编码、传输等完整流程。 + +### **1\. 配置视频编码参数** + +SDK 内部编码器会对推送的原始帧进行编码，可根据业务需要调整编码参数。 + +``` +AoqClientEngine.AoqVideoCodecConfig config = new AoqClientEngine.AoqVideoCodecConfig(); +config.width = 1280; +config.height = 720; +config.fps = 2; +config.bitrate = 500000; // 起始码率 500kbps +config.minBitrate = 128000; // 最小码率 128kbps +config.keyframeInterval = 2; +// isExternal 保持默认 false，SDK 内部编码 + +engine.setVideoEncoderConfig(config); +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +AoqTrackTypeVideo + +视频轨道类型 + +codecType + +AoqEncoderType + +AoqEncoderTypeVideoH264 + +编码器类型 + +width + +int + +720 + +编码宽度（像素） + +height + +int + +1280 + +编码高度（像素） + +fps + +int + +5 + +帧率 + +bitrate + +int + +500000 + +起始码率（bps） + +minBitrate + +int + +128000 + +最小码率（bps） + +keyframeInterval + +int + +2 + +关键帧间隔（秒） + +isExternal + +boolean + +false + +原始帧模式保持 false + +mirrorMode + +AoqMirrorMode + +AoqMirrorModeDisabled + +镜像模式 + +orientationMode + +AoqOrientationMode + +AoqOrientationModeAuto + +画面方向模式 + +### **2\. 以外部采集模式启动视频采集** + +调用 `startVideoCapture` 并设置 `isExternal=true`，告知 SDK 不打开摄像头，由外部源提供视频帧。这是原始帧模式的前置条件，未调用则 SDK 不会消费推送的帧数据。 + +``` +AoqClientEngine.AoqVideoCaptureConfig config = new AoqClientEngine.AoqVideoCaptureConfig(); +config.isExternal = true; // 不打开摄像头，由外部源推送视频帧 +// isExternal=true 时 width/height/fps 无效，实际分辨率和帧率由推送数据决定 +int ret = engine.startVideoCapture(config); +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +width + +int + +1280 + +采集宽度（`isExternal=true` 时无效） + +height + +int + +720 + +采集高度（`isExternal=true` 时无效） + +fps + +int + +15 + +采集帧率（`isExternal=true` 时无效） + +isExternal + +boolean + +false + +true：不打开摄像头，由外部源推送帧数据 + +cameraDirection + +AoqCameraDirection + +AoqCameraDirectionFront + +摄像头方向（`isExternal=true` 时无效） + +### **3\. 推送原始视频帧** + +调用 `pushExternalVideoCapturedFrame` 接口，将采集到的原始视频帧传入 SDK。SDK 内部完成编码和传输。 + +支持的视频帧格式：BGRA、I420、NV12、NV21、RGBA。Apple 平台额外支持 CVPixelBuffer 零拷贝格式。 + +#### **3.1 BGRA 格式** + +BGRA 为打包格式，每个像素 4 字节（Blue、Green、Red、Alpha），一帧数据量 = width x height x 4。 + +``` +// 构造 BGRA 视频帧 +AoqClientEngine.AoqVideoFrame frame = new AoqClientEngine.AoqVideoFrame(); +frame.format = AoqClientEngine.AoqVideoPixelFormat.AoqVideoPixelFormatBGRA; +frame.width = 1280; +frame.height = 720; + +frame.data = bgraBytes; // byte[]，长度 = width * height * 4 + +frame.timeStamp = System.currentTimeMillis(); + +int ret = engine.pushExternalVideoCapturedFrame( + AoqClientEngine.AoqTrackType.AoqTrackTypeVideo, frame); +``` + +#### **3.2 I420 格式** + +I420 为三平面格式（Y、U、V 分离），Y 平面大小 = width x height，U/V 平面各为 (width/2) x (height/2)。 + +``` +// 构造 I420 视频帧 +AoqClientEngine.AoqVideoFrame frame = new AoqClientEngine.AoqVideoFrame(); +frame.format = AoqClientEngine.AoqVideoPixelFormat.AoqVideoPixelFormatI420; +frame.width = 1280; +frame.height = 720; + +frame.dataY = yPlane; // byte[]，长度 = width * height + +frame.dataU = uPlane; // byte[]，长度 = (width/2) * (height/2) + +frame.dataV = vPlane; // byte[]，长度 = (width/2) * (height/2) + +frame.strideY = 1280; // Y 平面行字节数 +frame.strideU = 640; // U 平面行字节数 +frame.strideV = 640; // V 平面行字节数 +frame.timeStamp = System.currentTimeMillis(); + +int ret = engine.pushExternalVideoCapturedFrame( + AoqClientEngine.AoqTrackType.AoqTrackTypeVideo, frame); +``` + +#### **3.3 NV12 / NV21 格式** + +NV12 和 NV21 为半平面格式，Y 平面 + UV 交错平面。NV12 为 UV 交替排列，NV21 为 VU 交替排列。数据量 = width x height x 3 / 2，打包在 `data` 字段中。 + +``` +// 构造 NV12 视频帧（NV21 同理，修改 format 即可） +AoqClientEngine.AoqVideoFrame frame = new AoqClientEngine.AoqVideoFrame(); +frame.format = AoqClientEngine.AoqVideoPixelFormat.AoqVideoPixelFormatNV12; +frame.width = 1280; +frame.height = 720; + +frame.data = nv12Bytes; // byte[]，长度 = width * height * 3 / 2 + +frame.timeStamp = System.currentTimeMillis(); + +int ret = engine.pushExternalVideoCapturedFrame( + AoqClientEngine.AoqTrackType.AoqTrackTypeVideo, frame); +``` + +#### **3.4 CVPixelBuffer 格式（Apple 平台）** + +iOS / macOS 平台支持直接传递 `CVPixelBufferRef`，实现零拷贝传输，避免内存拷贝带来的性能开销。 + +``` +// iOS / macOS 平台 +let frame = AoqVideoFrame() +frame.format = .cvPixelBuffer +frame.width = 1280 +frame.height = 720 +frame.pixelBuffer = pixelBuffer // CVPixelBufferRef +frame.timeStamp = Int64(Date().timeIntervalSince1970 * 1000) + +// SDK 内部异步持有 pixelBuffer，需要额外 +1 引用计数 +// SDK 消费完毕后会自行释放 +let _ = Unmanaged.passRetained(pixelBuffer) + +engine.pushExternalVideoCapturedFrame(.video, frame: frame) +``` + +### **4\. 停止原始帧采集** + +当不再需要推送视频帧时，先停止推帧定时器，再调用 `stopVideoCapture` 关闭视频采集。 + +``` +// 1. 停止推帧定时器 +stopExternalFramePush(); +// 2. 停止视频采集 +engine.stopVideoCapture(); +``` + +## **模式二：编码帧模式** + +自行完成视频编码（目前支持 JPEG），直推已编码数据给 SDK，跳过 SDK 内部编码器，直接打包发送。此模式**不需要调用** `startVideoCapture` 等采集相关接口。 + +### **1\. 配置视频编码参数并启用外部编码** + +调用 `setVideoEncoderConfig` 并设置 `isExternal=true`，告知 SDK 跳过内部编码器，由外部提供已编码数据。 + +``` +AoqClientEngine.AoqVideoCodecConfig config = new AoqClientEngine.AoqVideoCodecConfig(); +config.width = 1280; +config.height = 720; +config.fps = 2; +config.isExternal = true; // 跳过内部编码，由外部推送已编码帧 + +engine.setVideoEncoderConfig(config); +``` + +设置完成后即可直接推送编码帧，无需调用 `startVideoCapture`。 + +### **2\. 推送编码视频帧** + +调用 `pushExternalVideoEncodedFrame` 接口，将已编码的视频数据直传给 SDK。目前仅支持 JPEG 编码格式。 + +``` +// 从 Bitmap 生成 JPEG 数据 +android.graphics.Bitmap bmp = android.graphics.Bitmap.createBitmap( + width, height, android.graphics.Bitmap.Config.ARGB_8888); +// ... 填充 Bitmap 内容 ... + +java.io.ByteArrayOutputStream baos = new java.io.ByteArrayOutputStream(); +bmp.compress(android.graphics.Bitmap.CompressFormat.JPEG, 85, baos); +bmp.recycle(); + +// 构造编码帧并推送 +AoqClientEngine.AoqVideoEncodedFrame frame = new AoqClientEngine.AoqVideoEncodedFrame(); +frame.codec = AoqClientEngine.AoqVideoCodecType.AoqVideoCodecTypeJPEG; +frame.data = baos.toByteArray(); +frame.width = width; +frame.height = height; +frame.timeStamp = System.currentTimeMillis(); + +int ret = engine.pushExternalVideoEncodedFrame( + AoqClientEngine.AoqTrackType.AoqTrackTypeVideo, frame); +``` + +**AoqVideoEncodedFrame 参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +codec + +AoqVideoCodecType + +AoqVideoCodecTypeJPEG + +编码格式，目前仅支持 JPEG + +data + +byte\[\] + +null + +编码后的数据 + +width + +int + +0 + +画面宽度（像素） + +height + +int + +0 + +画面高度（像素） + +timeStamp + +long + +0 + +时间戳（毫秒），为 0 时 SDK 使用本地时钟补充 + +### **3\. 停止编码帧推送** + +编码帧模式无需管理采集设备，停止推帧定时器即可。 + +``` +stopExternalFramePush(); +``` + +## **视频帧格式参考** + +### **AoqVideoFrame（原始帧模式使用）** + +**字段** + +**类型** + +**说明** + +format + +AoqVideoPixelFormat + +像素格式 + +width + +int + +画面宽度（像素） + +height + +int + +画面高度（像素） + +data + +byte\[\] + +打包格式数据（NV12/NV21/BGRA/RGBA） + +dataY + +byte\[\] + +I420 Y 平面数据 + +dataU + +byte\[\] + +I420 U 平面数据 + +dataV + +byte\[\] + +I420 V 平面数据 + +strideY + +int + +I420 Y 平面行字节数 + +strideU + +int + +I420 U 平面行字节数 + +strideV + +int + +I420 V 平面行字节数 + +textureId + +int + +Android 纹理 ID（TextureOES/Texture2D） + +transformMatrix + +float\[\] + +纹理变换矩阵（4x4 行优先） + +eglContext + +EGLContext + +Android 共享 EGL context（纹理模式使用） + +pixelBuffer + +CVPixelBufferRef + +Apple 零拷贝 CVPixelBuffer（仅 iOS/macOS） + +timeStamp + +long + +时间戳（毫秒），为 0 时 SDK 用本地时钟补充 + +### **AoqVideoPixelFormat 枚举值** + +**枚举值** + +**数值** + +**说明** + +AoqVideoPixelFormatUnknown + +0 + +未知格式 + +AoqVideoPixelFormatI420 + +1 + +I420 三平面格式 + +AoqVideoPixelFormatNV12 + +2 + +NV12 半平面格式（UV 交替） + +AoqVideoPixelFormatNV21 + +3 + +NV21 半平面格式（VU 交替） + +AoqVideoPixelFormatBGRA + +4 + +BGRA 打包格式 + +AoqVideoPixelFormatRGBA + +5 + +RGBA 打包格式 + +AoqVideoPixelFormatCVPixelBuffer + +6 + +Apple CVPixelBuffer（仅 iOS/macOS） + +AoqVideoPixelFormatTextureOES + +7 + +Android OES 外部纹理 + +AoqVideoPixelFormatTexture2D + +8 + +Android 2D 纹理 + +### **AoqVideoEncodedFrame（编码帧模式使用）** + +**字段** + +**类型** + +**说明** + +codec + +AoqVideoCodecType + +编码格式 + +data + +byte\[\] + +编码后的数据 + +width + +int + +画面宽度（像素） + +height + +int + +画面高度（像素） + +timeStamp + +long + +时间戳（毫秒），为 0 时 SDK 用本地时钟补充 + +### **AoqVideoCodecType 枚举值** + +**枚举值** + +**数值** + +**说明** + +AoqVideoCodecTypeJPEG + +0 + +JPEG 编码格式 + +## **注意事项** + +- 原始帧模式：必须先调用 `startVideoCapture(isExternal=true)` 再推送帧，否则 SDK 返回参数错误。 + +- 编码帧模式：只需调用 `setVideoEncoderConfig(isExternal=true)` 即可推送，**不需要**调用 `startVideoCapture`。 + +- 原始帧模式与编码帧模式不可混用：同一时间只能使用其中一种推送接口。 + +- 编码帧模式目前仅支持 JPEG 格式。 + +- 视频帧数据在推送后由 SDK 内部管理生命周期，调用方无需在推送后继续持有数据引用。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md new file mode 100644 index 00000000..69b625b7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md @@ -0,0 +1,115 @@ +# 媒体流发送管理 + +`enableSendMediaStream` 用于控制客户端是否向 AI 服务发送音频/视频媒体流，在 AOQ 协议接入场景中精确控制发送时机。 + +## **概述** + +`enableSendMediaStream` 用于控制客户端是否向 AI 服务发送音频/视频媒体流。在 AOQ 协议接入场景中，部分模型要求在收到 `session.updated` 确认后才能接收媒体数据，因此需要通过此接口精确控制发送时机。 + +## **API 定义** + +**iOS / Mac** + +``` +// iOS / Mac +func enableSendMediaStream(_ trackType: AoqTrackType, enable: Bool) +``` + +**Android** + +``` +// Android +void enableSendMediaStream(AoqTrackType trackType, boolean enable) +``` + +**OHOS (ArkTS)** + +``` +// OHOS (ArkTS) +enableSendMediaStream(trackType: AoqTrackType, enable: boolean): void +``` + +**参数说明：** + +**参数** + +**类型** + +**说明** + +trackType + +AoqTrackType + +媒体轨道类型：`.audio` 或 `.video` + +enable + +Bool + +`true` = 开启发送，`false` = 暂停发送 + +## **控制流程** + +典型的媒体流控制流程如下： + +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0575914871/p1088082.svg) + +## **使用示例（iOS Swift）** + +### **connect 前禁用发送** + +``` +// 连接前关闭音视频发送，避免模型未就绪时收到数据 +engine.enableSendMediaStream(.audio, enable: false) +engine.enableSendMediaStream(.video, enable: false) + +// 发起连接 +engine.connect(config) +``` + +### **收到 session.updated 后开启** + +``` +func onDataMsg(_ msg: AoqDataMsg) { + guard let obj = try? JSONSerialization.jsonObject(with: msg.data) as? [String: Any], + let type = obj["type"] as? String else { return } + + if type == "session.updated" { + // AI 侧已确认会话配置，开启媒体发送 + engine.enableSendMediaStream(.audio, enable: true) + engine.enableSendMediaStream(.video, enable: true) + } +} +``` + +## **注意事项** + +- **调用时机**：必须在 `createEngine` 之后调用，引擎未创建时调用无效。 + +- **默认行为**：如果不调用此接口，connect 成功后 SDK 会立即开始发送媒体流。 + +- **模型兼容性**：部分模型要求先收到 `session.updated` 再接收媒体数据，建议统一采用"先禁用、后开启"模式。 + +- **独立控制**：音频和视频可独立控制，例如仅发送音频不发送视频。 + + +## **常见场景** + +**场景** + +**操作** + +**说明** + +连接模型前 + +`enable(.audio, false)` + +等待 session.updated 再发送 + +收到 session.updated + +`enable(.audio, true)` + +AI 已就绪，开始发送 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md new file mode 100644 index 00000000..f71bf047 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md @@ -0,0 +1,1186 @@ +# 视频常用功能介绍 + +AOQ Client SDK 提供了完整的视频能力，覆盖视频采集、渲染显示、编码配置、帧数据回调、外部视频输入等核心场景。本文档基于 Android（Java）、iOS（Objective-C）、Ohos（ArkTS）三个平台的公开 API，对视频常用功能进行统一介绍。 + +## **1\. 视频采集** + +### 1.1 功能说明 + +视频采集用于打开设备摄像头，将实时视频帧数据送入 SDK 编码推流管线。SDK 支持两种采集模式： + +- **内部采集（默认）**：SDK 自动管理摄像头设备的打开、帧采集和关闭，支持前后置摄像头切换。 + +- **外部采集**：由应用自行管理摄像头或其他视频源，采集到的帧数据通过 `pushExternalVideoCapturedFrame` 接口输入 SDK。 + + +### 1.2 采集配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +width + +int + +1280 + +采集宽度（像素），外部采集时无效 + +height + +int + +720 + +采集高度（像素），外部采集时无效 + +fps + +int + +15 + +采集帧率，外部采集时由送帧节奏决定 + +isExternal + +bool + +false + +是否使用外部采集模式 + +cameraDirection + +AoqCameraDirection + +Front(0) + +摄像头方向，外部采集时无效 + +### 1.3 摄像头方向枚举 + +**枚举值** + +**数值** + +**说明** + +AoqCameraDirectionFront + +0 + +前置摄像头 + +AoqCameraDirectionBack + +1 + +后置摄像头 + +### 1.4 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +开启采集 + +startVideoCapture(config) + +startVideoCapture:config: + +startVideoCapture(config) + +关闭采集 + +stopVideoCapture() + +stopVideoCapture + +stopVideoCapture() + +切换摄像头 + +switchCamera(direction) + +switchCamera: + +switchCamera(direction) + +### 1.5 使用示例 + +**Android** + +``` +AoqVideoCaptureConfig config = new AoqVideoCaptureConfig(); +config.width = 1280; +config.height = 720; +config.fps = 15; +config.cameraDirection = AoqCameraDirection.AoqCameraDirectionFront; +engine.startVideoCapture(config); +``` + +**iOS** + +``` +AoqVideoCaptureConfig *config = [[AoqVideoCaptureConfig alloc] init]; +config.width = 1280; +config.height = 720; +config.fps = 15; +config.cameraDirection = AoqCameraDirectionFront; +[engine startVideoCapture:config]; +``` + +**Ohos** + +``` +let config: AoqVideoCaptureConfig = { + width: 1280, + height: 720, + fps: 15, + cameraDirection: AoqCameraDirection.AoqCameraDirectionFront +}; +engine.startVideoCapture(config); +``` + +## **2\. 视频渲染** + +### 2.1 功能说明 + +视频渲染用于将本地采集或远端接收的视频帧数据显示到屏幕上。SDK 支持设置本地预览窗口和远端渲染窗口，通过 `trackType` 区分视频流（Video）和屏幕共享流（Screen）。 + +### 2.2 渲染模式 + +**枚举值** + +**数值** + +**说明** + +AoqRenderModeAuto + +0 + +自动模式 + +AoqRenderModeStretch + +1 + +拉伸平铺，画面可能变形 + +AoqRenderModeFill + +2 + +填充黑边，画面完整显示 + +AoqRenderModeCrop + +3 + +裁剪模式，画面内容可能丢失 + +### 2.3 画布配置 + +**参数** + +**类型** + +**默认值** + +**说明** + +view + +平台视图 + +null + +渲染视图（Android: SurfaceView/TextureView, iOS: UIView, Ohos: XComponent） + +renderMode + +AoqRenderMode + +Auto(0) + +渲染显示模式 + +### 2.4 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +设置本地预览 + +setLocalView(trackType, canvas) + +setLocalView:trackType:canvas: + +setLocalView(trackType, canvas) + +设置远端渲染 + +setRemoteView(trackType, canvas) + +setRemoteView:trackType:canvas: + +setRemoteView(trackType, canvas) + +**说明** + +**平台差异**：Android 使用 SurfaceView 或 TextureView 作为渲染容器；iOS 使用 UIView（内部通过 AoqRenderView 封装，支持 Metal 加速）；Ohos 使用 XComponent（通过 AoqXComponentController 管理 native 渲染视图）。 + +## **3\. 视频编码配置** + +### 3.1 功能说明 + +设置视频编码参数，包括编码格式、分辨率、帧率、码率、关键帧间隔、镜像和方向等。通过 `trackType` 区分视频轨道和屏幕共享轨道的编码配置。 + +### 3.2 编码配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +Video(1) + +轨道类型：Video + +codecType + +AoqEncoderType + +VideoH264(3) + +编码格式 + +width + +int + +720 + +编码宽度 + +height + +int + +1280 + +编码高度 + +fps + +int + +5 + +编码帧率 + +bitrate + +int + +500000 + +目标码率（bps） + +minBitrate + +int + +128000 + +最小码率（bps） + +keyframeInterval + +int + +2 + +关键帧间隔（秒） + +mirrorMode + +AoqMirrorMode + +Disabled(0) + +镜像模式 + +orientationMode + +AoqOrientationMode + +Auto(0) + +方向模式 + +isExternal + +bool + +false + +外部编码模式（true 时由应用推送已编码帧） + +### 3.3 编码格式枚举 + +**枚举值** + +**数值** + +**说明** + +AoqEncoderTypeVideoH264 + +3 + +H.264 编码 + +AoqEncoderTypeVideoJpeg + +4 + +JPEG 编码（用于外部编码帧） + +### 3.4 镜像模式 + +**枚举值** + +**数值** + +**说明** + +AoqMirrorModeDisabled + +0 + +禁用镜像 + +AoqMirrorModeEnabled + +1 + +启用镜像 + +### 3.5 方向模式 + +**枚举值** + +**数值** + +**说明** + +AoqOrientationModeAuto + +0 + +自动方向 + +AoqOrientationModePortrait + +1 + +竖屏方向 + +AoqOrientationModeLandscape + +2 + +横屏方向 + +### 3.6 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +设置编码参数 + +setVideoEncoderConfig(config) + +setVideoEncoderConfig: + +setVideoEncoderConfig(config) + +## **4\. 外部视频帧输入** + +### 4.1 功能说明 + +外部视频帧输入允许应用将自定义的视频帧数据推送到 SDK，用于外部采集或外部编码场景。支持两种推送方式： + +- **推送原始帧**：将未编码的像素数据（I420/NV12/NV21/BGRA/RGBA 等格式）推送给 SDK，由 SDK 进行编码。 + +- **推送已编码帧**：将已编码的数据（如 JPEG）直推给 SDK，SDK 不做二次编码，直接打包发送。 + + +通过 `trackType` 路由，`AoqTrackTypeVideo` 对应视频采集的外部帧，`AoqTrackTypeScreen` 对应屏幕共享的外部帧。 + +### 4.2 像素格式枚举 + +**枚举值** + +**数值** + +**说明** + +**平台支持** + +AoqVideoPixelFormatI420 + +1 + +I420 三平面 + +全平台 + +AoqVideoPixelFormatNV12 + +2 + +NV12 双平面 + +全平台 + +AoqVideoPixelFormatNV21 + +3 + +NV21 双平面 + +全平台 + +AoqVideoPixelFormatBGRA + +4 + +BGRA 打包 + +全平台 + +AoqVideoPixelFormatRGBA + +5 + +RGBA 打包 + +全平台 + +AoqVideoPixelFormatCVPixelBuffer + +6 + +Apple 零拷贝 + +仅 iOS + +AoqVideoPixelFormatTextureOES + +7 + +OES 纹理 + +仅 Android + +AoqVideoPixelFormatTexture2D + +8 + +2D 纹理 + +仅 Android + +### 4.3 原始视频帧数据结构 (AoqVideoFrame) + +**字段** + +**类型** + +**说明** + +format + +AoqVideoPixelFormat + +像素格式 + +width + +int + +宽度（像素） + +height + +int + +高度（像素） + +data + +byte\[\] / ArrayBuffer + +打包格式数据（NV12/NV21/BGRA/RGBA） + +dataY / dataU / dataV + +byte\[\] / ArrayBuffer + +I420 三平面数据 + +strideY / strideU / strideV + +int + +I420 三平面步长 + +textureId + +int + +纹理 ID（Android TextureOES/Texture2D 时有效） + +transformMatrix + +float\[16\] + +4x4 纹理变换矩阵（Android） + +eglContext + +EGLContext + +共享 EGL 上下文（Android） + +pixelBuffer + +CVPixelBufferRef + +Apple 零拷贝（iOS） + +timeStamp + +long + +时间戳（ms），0 时 SDK 用本地时钟补 + +### 4.4 已编码视频帧数据结构 (AoqVideoEncodedFrame) + +**字段** + +**类型** + +**默认值** + +**说明** + +codec + +AoqVideoCodecType + +JPEG(0) + +编码格式 + +data + +byte\[\] / ArrayBuffer + +\- + +编码后数据 + +width + +int + +\- + +宽度（像素） + +height + +int + +\- + +高度（像素） + +timeStamp + +long + +0 + +时间戳（ms） + +### 4.5 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +推送原始帧 + +pushExternalVideoCapturedFrame(trackType, frame) + +pushExternalVideoCapturedFrame:frame: + +pushExternalVideoCapturedFrame(trackType, frame) + +推送已编码帧 + +pushExternalVideoEncodedFrame(trackType, frame) + +pushExternalVideoEncodedFrame:frame: + +pushExternalVideoEncodedFrame(trackType, frame) + +## **5\. 视频帧数据回调** + +### 5.1 功能说明 + +视频帧回调允许开发者在视频管线的不同位置获取原始帧数据，用于视频分析、自定义处理、录制等场景。支持只读和读写两种模式，读写模式下可修改帧数据并写回 SDK。 + +### 5.2 支持的数据源位置 + +**数据源** + +**枚举值** + +**说明** + +Captured + +0 + +采集后的视频数据（前处理前） + +PreEncode + +1 + +编码前的视频数据（前处理后） + +Remote + +2 + +远端解码后、渲染前的视频数据 + +### 5.3 回调配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +format + +AoqVideoPixelFormat + +I420(1) + +期望回调的像素格式 + +alignment + +AoqVideoObserverAlignment + +Default(0) + +宽度对齐策略 + +mode + +AoqVideoObserverMode + +ReadOnly(0) + +只读(0)/读写(1) 模式 + +mirrorApplied + +bool + +false + +是否对回调数据应用镜像 + +### 5.4 宽度对齐枚举 + +**枚举值** + +**数值** + +**说明** + +AoqVideoObserverAlignmentDefault + +0 + +默认对齐 + +AoqVideoObserverAlignmentEven + +1 + +2 字节对齐 + +AoqVideoObserverAlignment4 + +2 + +4 字节对齐 + +AoqVideoObserverAlignment8 + +3 + +8 字节对齐 + +AoqVideoObserverAlignment16 + +4 + +16 字节对齐 + +### 5.5 使用步骤 + +1. **注册观察者**：调用 `setVideoFrameObserver` 设置视频帧回调监听器 + +2. **启用数据源**：调用 `enableVideoFrameObserver` 选择需要监听的数据源位置，开启回调 + +3. **处理回调数据**：在回调函数中获取帧数据（仅回调期间有效，异步使用需自行拷贝） + + +### 5.6 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +注册观察者 + +setVideoFrameObserver(listener) + +setVideoFrameObserver: + +setVideoFrameObserver(observer) + +启用回调 + +enableVideoFrameObserver(enabled, source, config) + +enableVideoFrameObserver:videoSource:config: + +enableVideoFrameObserver(enabled, source, config) + +### 5.7 回调方法 + +**回调** + +**Android** + +**iOS** + +**Ohos** + +采集后数据 + +onCapturedVideoFrame(frame) + +onCapturedVideoFrame: + +onCapturedVideoFrame(frame) + +编码前数据 + +onPreEncodeVideoFrame(trackType, frame) + +onPreEncodeVideoFrame:frame: + +onPreEncodeVideoFrame(trackType, frame) + +远端数据 + +onRemoteVideoFrame(trackType, frame) + +onRemoteVideoFrame:frame: + +onRemoteVideoFrame(trackType, frame) + +**说明** + +回调方法返回 `true`/`YES` 表示数据已修改、需写回 SDK（仅 ReadWrite 模式且 I420 格式时生效）。 + +## **7\. 媒体流发送控制** + +### 7.1 功能说明 + +控制本地媒体流的发送开关，通过 `trackType` 路由到不同轨道（Audio/Video/Screen）。停用发送后，采集和编码继续运行，但数据不会发送到远端。 + +### 7.2 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +控制流发送 + +enableSendMediaStream(trackType, enable) + +enableSendMediaStream:enable: + +enableSendMediaStream(trackType, enable) + +### 7.3 轨道类型枚举 + +**枚举值** + +**数值** + +**说明** + +AoqTrackTypeAudio + +0 + +音频轨道 + +AoqTrackTypeVideo + +1 + +视频轨道 + +AoqTrackTypeData + +2 + +数据轨道 + +## **8\. 视频设备状态监控** + +### 8.1 功能说明 + +SDK 自动监测视频采集设备（摄像头）的状态变化，并通过 `onVideoDeviceStateChanged` 回调通知应用层。 + +### 8.2 设备状态码 + +**状态码** + +**值** + +**说明** + +AoqVideoDeviceNone + +0 + +初始状态 + +AoqVideoDeviceCaptureStarting + +1 + +采集启动中 + +AoqVideoDeviceCaptureStarted + +2 + +采集已启动 + +AoqVideoDeviceCaptureStopping + +3 + +采集停止中 + +AoqVideoDeviceCaptureStopped + +4 + +采集已停止 + +AoqVideoDeviceCaptureFail + +5 + +采集失败 + +### 8.3 回调对照 + +**回调** + +**Android** + +**iOS** + +**Ohos** + +设备状态变化 + +onVideoDeviceStateChanged(state) + +onVideoDeviceStateChanged: + +onVideoDeviceStateChanged(state) + +## **9\. 视频错误码与警告码** + +### 9.1 视频错误码 + +**错误码** + +**值** + +**说明** + +AoqErrorCodeVideo + +200 + +通用视频错误 + +VideoExternalBufferFull + +210 + +视频外部缓冲区已满 + +VideoDevice + +220 + +视频设备通用错误 + +CameraOpenFail + +221 + +摄像头打开失败 + +CameraAuthFailed + +222 + +摄像头权限被拒绝 + +CameraOccupied + +223 + +摄像头被占用 + +CameraRunningError + +224 + +摄像头运行错误 + +VideoCodec + +230 + +视频编解码通用错误 + +EncoderInitFail + +231 + +编码器初始化失败 + +VideoRender + +240 + +视频渲染通用错误 + +RenderCreateFail + +241 + +渲染器创建失败 + +RenderDrawError + +242 + +渲染绘制错误 + +Screen + +300 + +屏幕共享通用错误 + +**说明** + +Android 额外错误码：ScreenPermissionDenied(310) 屏幕共享权限被拒绝、ScreenForegroundServiceFailed(311) 前台服务启动失败。 + +### 9.2 视频警告码 + +**警告码** + +**值** + +**说明** + +AoqWCVideo + +200 + +通用视频警告 + +CameraEnumerateError + +201 + +摄像头枚举错误 + +EncoderSwitched + +202 + +编码器切换警告 + +RenderDowngrade + +203 + +渲染降级警告 + +## **附录：完整视频 API 方法列表** + +**分类** + +**方法名** + +**说明** + +采集控制 + +startVideoCapture + +打开视频采集设备 + +采集控制 + +stopVideoCapture + +关闭视频采集设备 + +采集控制 + +switchCamera + +切换前后置摄像头 + +渲染控制 + +setLocalView + +设置本地预览窗口 + +渲染控制 + +setRemoteView + +设置远端渲染窗口 + +编解码 + +setVideoEncoderConfig + +设置视频编码参数 + +外部输入 + +pushExternalVideoCapturedFrame + +推送原始视频帧 + +外部输入 + +pushExternalVideoEncodedFrame + +推送已编码视频帧 + +屏幕共享 + +startScreenCapture + +启动屏幕采集 + +屏幕共享 + +stopScreenCapture + +停止屏幕采集 + +流控制 + +enableSendMediaStream + +控制媒体流发送 + +帧回调 + +setVideoFrameObserver + +注册视频帧观察者 + +帧回调 + +enableVideoFrameObserver + +启用/禁用视频帧回调 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md new file mode 100644 index 00000000..a291cc23 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md @@ -0,0 +1,520 @@ +# 通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话 + +本文档说明如何在 Android、iOS、HarmonyOS 平台接入 AOQ Client SDK，实现 AOQ+qwen3.5-omni-plus-realtime 音视频通话功能。 + +## **SDK 获取** + +AOQ Client SDK 及音频 Opus 插件请参见[SDK下载](https://help.aliyun.com/zh/model-studio/realtime-sdk-download)。Opus 编码以独立插件形式提供，请根据您的场景按需引入。 + +## **SDK 导入** + +请根据不同平台将核心 SDK 产物导入工程依赖目录，并在工程配置中声明相关权限。 + +### **Android** + +将 `AoqClientSdk-release.aar` 放入工程 `app/libs/` 目录，将 `libPluginOpus.so` 按 ABI 放入 `app/libs/armeabi-v7a/` 和 `app/libs/arm64-v8a/`，并在 `app/build.gradle` 中： + +``` +android { + defaultConfig { + minSdk 21 + ndk { abiFilters 'armeabi-v7a', 'arm64-v8a' } + } + sourceSets { main { jniLibs.srcDirs = ['libs'] } } + packagingOptions { + // 避免与宿主工程的同名 so 冲突 + pickFirsts += ['lib/*/*.so'] + } +} + +dependencies { + implementation fileTree(dir: 'libs', include: ['*.aar']) +} +``` + +在 `AndroidManifest.xml` 声明权限： + +``` + + + + + +``` + +其中 `RECORD_AUDIO` 和 `CAMERA` 为运行时权限，应用需在运行时调用 Android `ActivityCompat.requestPermissions()` 方法，主动向 Android 系统申请用户授权。 + +### **iOS（framework）** + +1. 将 `AoqClientSdk.framework` 与 `PluginOpus.framework` 拖入 Xcode 工程，在 Target > General > Frameworks, Libraries, and Embedded Content 中选择 **Embed & Sign**。 + +2. 权限声明：在 Xcode 中选中您的 Target > Info > Custom iOS Target Properties，添加以下两项权限用途描述： + + **Key** + + **Value** + + `NSMicrophoneUsageDescription` + + 用于实时语音通话 + + `NSCameraUsageDescription` + + 用于实时视频通话 + +3. Swift 工程：`import AoqClientSdk`；Objective-C 工程：`#import `。 + + +### **HarmonyOS（har）** + +1. 将 `aoq-client-sdk.har` 放入工程 `libs/` 目录，将 `libPluginOpus.so` 按 ABI 放入 `entry/libs/armeabi-v7a/` 和 `entry/libs/arm64-v8a/`；并在 `entry/oh-package.json5` 中声明。 + +2. 在 `entry/src/main/module.json5` 添加权限： + + ``` + "requestPermissions": [ + { "name": "ohos.permission.INTERNET" }, + { "name": "ohos.permission.MICROPHONE", + "reason": "$string:perm_mic_reason", + "usedScene": { "abilities": ["EntryAbility"], "when": "inuse" } }, + { "name": "ohos.permission.CAMERA", + "reason": "$string:perm_camera_reason", + "usedScene": { "abilities": ["EntryAbility"], "when": "inuse" } } + ] + ``` + +3. 在 `EntryAbility` 中通过 `abilityAccessCtrl.createAtManager().requestPermissionsFromUser` 触发运行时授权。 + + +## **AppServer获取Token** + +请按照[Token鉴权](https://help.aliyun.com/zh/model-studio/realtime-token-authentication)的 AOQ 章节搭建获取 Token 的 AppServer。每次通话前，客户端需要向业务侧 AppServer 请求一次 Token。 + +## **实现 AI 音视频通话** + +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0075914871/p1088080.svg) + +### **创建引擎并设置回调** + +调用 `createEngine` 接口创建 `AoqClientEngine` 实例。 + +**iOS：** + +``` +let config = AoqCreateConfig() +config.workDir = workDir +engine = AoqClientEngine.createEngine(config, delegate: self) +``` + +实现 `AoqEngineDelegate` 协议监听 `onConnectionStatusChange`、`onDataMsg`、`onError` 等回调。 + +**Android：** + +``` +AoqCreateConfig config = new AoqCreateConfig(); +config.workDir = appCtx.getFilesDir().getAbsolutePath(); +engine = AoqClientEngine.createEngine(appCtx, config, this); +``` + +**HarmonyOS：** + +``` +const config: AoqCreateConfig = { workDir: context.filesDir, extras: '' }; +engine = AoqClientEngine.createEngine(config, this, context); +``` + +### **启动音视频采集与播放** + +调用 `startAudioCapture` 与 `startAudioPlayer` 启动本地音频采集与播放；调用 `startVideoCapture` 启动摄像头，并通过 `setLocalView` 将 SDK 渲染目标绑定到业务侧的预览控件。 + +**iOS：** + +``` +// 音频采集 +let capCfg = AoqAudioCaptureConfig() +capCfg.channel = 1; capCfg.isExternal = false +engine.startAudioCapture(capCfg) + +// 音频播放 +let playCfg = AoqAudioPlaybackConfig() +playCfg.channel = 1; playCfg.isExternal = false +engine.startAudioPlayer(playCfg) + +// 视频采集 +let vidCfg = AoqVideoCaptureConfig() +vidCfg.width = 720; vidCfg.height = 1280; vidCfg.fps = 15 +engine.startVideoCapture(vidCfg) + +// 为本地预览画面设置渲染视图 +let canvas = AoqVideoCanvas() +canvas.view = localPreview +canvas.renderMode = .crop +engine.setLocalView(.video, canvas: canvas) +``` + +**Android：** + +``` +// 音频采集 +AoqAudioCaptureConfig capCfg = new AoqAudioCaptureConfig(); +capCfg.channel = 1; capCfg.isExternal = false; +engine.startAudioCapture(capCfg); + +// 音频播放 +AoqAudioPlaybackConfig playCfg = new AoqAudioPlaybackConfig(); +playCfg.channel = 1; playCfg.isExternal = false; +engine.startAudioPlayer(playCfg); + +// 视频采集 +AoqVideoCaptureConfig vidCfg = new AoqVideoCaptureConfig(); +vidCfg.width = 720; vidCfg.height = 1280; vidCfg.fps = 15; +engine.startVideoCapture(vidCfg); + +// 为本地预览画面设置渲染视图 +AoqVideoCanvas canvas = new AoqVideoCanvas(); +canvas.view = localPreview; +canvas.renderMode = AoqRenderMode.AoqRenderModeCrop; +engine.setLocalView(AoqTrackType.AoqTrackTypeVideo, canvas); +``` + +**HarmonyOS：** + +``` +// 音频采集 +const capCfg: AoqAudioCaptureConfig = { channel: 1, isExternal: false }; +engine.startAudioCapture(capCfg); + +// 音频播放 +const playCfg: AoqAudioPlaybackConfig = { channel: 1, isExternal: false }; +engine.startAudioPlayer(playCfg); + +// 视频采集 +const vidCfg: AoqVideoCaptureConfig = { width: 720, height: 1280, fps: 15, isExternal: false }; +engine.startVideoCapture(vidCfg); + +// 为本地预览画面设置渲染视图 +const canvas: AoqVideoCanvas = { view: localCtrl, renderMode: AoqRenderMode.AoqRenderModeCrop }; +engine.setLocalView(AoqTrackType.AoqTrackTypeVideo, canvas); +``` + +### **获取连接凭证** + +由业务 AppServer 代理百炼请求，参见[Token鉴权](https://help.aliyun.com/zh/model-studio/realtime-token-authentication)。 + +### **设置编解码及建立连接** + +设置编解码参数后调用 `connect`。 + +注意：qwen3.5-omni-plus-realtime 要求客户端在收到服务端的 `session.updated` 之后才能开始发送媒体数据。为避免 `connect` 建联成功到 `session.updated` 到达之间的空档期误推媒体，在 `connect` 之前对上行音频与视频轨道分别调用 `enableSendMediaStream(trackType, false)`，将上行推流暂时关闭。WebSocket事件说明详见[客户端事件](https://help.aliyun.com/zh/model-studio/client-events)。 + +**iOS：** + +``` +// 音频编解码配置 +let encCfg = AoqAudioCodecConfig() +encCfg.codecType = .audioPCM; encCfg.sampleRate = 16000; encCfg.channel = 1 +engine.setAudioEncoderConfig(encCfg) +engine.setAudioDecoderConfig(encCfg) + +// connect 前关闭媒体发送，待 session.updated 后再开启 +engine.enableSendMediaStream(.audio, enable: false) +engine.enableSendMediaStream(.video, enable: false) + +// 建立连接 +let conn = AoqConnectConfig() +conn.token = token; conn.sid = sid; conn.certFingerprint = cert +conn.relayEndpoints = endpoints; conn.workspaceIdHash = workspaceIdHash + +let aTrack = AoqTrackParam(); aTrack.trackType = .audio +let vTrack = AoqTrackParam(); vTrack.trackType = .video +let dTrack = AoqTrackParam(); dTrack.trackType = .data +conn.publishTracks = [aTrack, vTrack, dTrack] +conn.subscribeTracks = [aTrack, dTrack] +engine.connect(conn) +``` + +**Android：** + +``` +// 音频编解码配置 +AoqAudioCodecConfig encCfg = new AoqAudioCodecConfig(); +encCfg.codecType = AoqEncoderType.AoqEncoderTypeAudioPCM; +encCfg.sampleRate = 16000; encCfg.channel = 1; +engine.setAudioEncoderConfig(encCfg); +engine.setAudioDecoderConfig(encCfg); + +// connect 前关闭媒体发送，待 session.updated 后再开启 +engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeAudio, false); +engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeVideo, false); + +// 建立连接 +AoqConnectConfig conn = new AoqConnectConfig(); +conn.token = token; conn.sid = sid; conn.certFingerprint = cert; +conn.relayEndpoints.addAll(endpoints); conn.workspaceIdHash = workspaceIdHash; + +AoqTrackParam aTrack = new AoqTrackParam(); aTrack.trackType = AoqTrackType.AoqTrackTypeAudio; +AoqTrackParam vTrack = new AoqTrackParam(); vTrack.trackType = AoqTrackType.AoqTrackTypeVideo; +AoqTrackParam dTrack = new AoqTrackParam(); dTrack.trackType = AoqTrackType.AoqTrackTypeData; +conn.publishTracks.add(aTrack); +conn.publishTracks.add(vTrack); +conn.publishTracks.add(dTrack); +conn.subscribeTracks.add(aTrack); +conn.subscribeTracks.add(dTrack); +engine.connect(conn); +``` + +**HarmonyOS：** + +``` +// 音频编解码配置 +const encCfg: AoqAudioCodecConfig = { + codecType: AoqEncoderType.AoqEncoderTypeAudioPCM, + sampleRate: 16000, channel: 1 +}; +engine.setAudioEncoderConfig(encCfg); +engine.setAudioDecoderConfig(encCfg); + +// connect 前关闭媒体发送，待 session.updated 后再开启 +engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeAudio, false); +engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeVideo, false); + +// 建立连接 +const conn: AoqConnectConfig = { + token, sid, certFingerprint: cert, + relayEndpoints: endpoints, + workspaceIdHash, + publishTracks: [ + { trackType: AoqTrackType.AoqTrackTypeAudio }, + { trackType: AoqTrackType.AoqTrackTypeVideo }, + { trackType: AoqTrackType.AoqTrackTypeData } + ], + subscribeTracks: [ + { trackType: AoqTrackType.AoqTrackTypeAudio }, + { trackType: AoqTrackType.AoqTrackTypeData } + ] +}; +engine.connect(conn); +``` + +**重要**：AOQ SDK 在建联后会默认发送媒体数据，此示例演示了连接模型时关闭媒体发送的能力。 + +### **配置 AI 会话** + +在 `onConnectionStatusChange(Connected)` 回调中通过 `sendDataMsg` 发送 `session.update` 消息（业务自定义 JSON，包含 modalities、voice、instructions、turn\_detection 等会话参数），完成会话握手，WebSocket事件说明详见[客户端事件](https://help.aliyun.com/zh/model-studio/client-events)。 + +**iOS：** + +``` +func onConnectionStatusChange(_ status: AoqConnectionStatus) { + if status == .connected { sendSessionUpdate() } +} + +private func sendSessionUpdate() { + let json = """ + { + // 该事件的id，由客户端生成 + "event_id": "event_ToPZqeobitzUJnt3QqtWg", + // 事件类型，固定为session.update + "type": "session.update", + // 会话配置 + "session": { + // 输出模态，支持设置为["text"]（仅输出文本）或["text","audio"]（输出文本与音频）。 + "modalities": [ + "text", + "audio" + ], + // 输出音频的音色 + "voice": "Ethan", + // 输入音频格式，当前仅支持设置为pcm。输入音频为16 kHz采样率的PCM音频流。 + "input_audio_format": "pcm", + // 输出音频格式，当前仅支持设置为pcm。输出音频为24 kHz采样率的PCM音频流。 + "output_audio_format": "pcm", + // 系统消息，用于设定模型的目标或角色。 + "instructions": "你是某五星级酒店的AI客服专员，请准确且友好地解答客户关于房型、设施、价格、预订政策的咨询。请始终以专业和乐于助人的态度回应，杜绝提供未经证实或超出酒店服务范围的信息。", + // 是否开启语音活动检测。若需启用，需传入一个配置对象，服务端将据此自动检测语音起止。 + // 设置为null表示由客户端决定何时发起模型响应。 + "turn_detection": { + // VAD类型，取值为server_vad或semantic_vad。使用qwen3.5-omni-realtime模型时推荐设为semantic_vad。 + "type": "semantic_vad", + // VAD检测阈值。建议在嘈杂的环境中增加，在安静的环境中降低。 + "threshold": 0.5, + // 检测语音停止的静音持续时间，超过此值后会触发模型响应 + "silence_duration_ms": 800 + } + } + } + """ + let msg = AoqDataMsg() + msg.data = json.data(using: .utf8)! + engine.send(msg) +} +``` + +**Android：** + +``` +@Override +public void onConnectionStatusChange(AoqConnectionStatus status) { + if (status == AoqConnectionStatus.AoqConnectionStatusConnected) { + sendSessionUpdate(); + } +} + +private void sendSessionUpdate() { + String sessionUpdateJson = /* 与 Swift 示例中相同的 session.update JSON */; + AoqDataMsg msg = new AoqDataMsg(); + msg.data = sessionUpdateJson.getBytes(StandardCharsets.UTF_8); + engine.sendDataMsg(msg); +} +``` + +**HarmonyOS：** + +``` +onConnectionStatusChange(status: AoqConnectionStatus): void { + if (status === AoqConnectionStatus.AoqConnectionStatusConnected) { + this.sendSessionUpdate(); + } +} + +private sendSessionUpdate(): void { + const sessionUpdateJson = /* 与 Swift 示例中相同的 session.update JSON */; + const msg: AoqDataMsg = { data: new TextEncoder().encode(sessionUpdateJson).buffer }; + this.engine.sendDataMsg(msg); +} +``` + +### **收到 session.updated 后开启媒体发送** + +在 `onDataMsg` 回调中解析下行消息，收到模型回复 `session.updated` 的时候，对上一步禁推的每个轨道类型调用 `enableSendMediaStream(trackType, true)` 放开推流。下面为代码示例，WebSocket事件说明详见[服务端事件](https://help.aliyun.com/zh/model-studio/server-events)。 + +**iOS：** + +``` +func onDataMsg(_ msg: AoqDataMsg) { + guard let obj = try? JSONSerialization.jsonObject(with: msg.data) as? [String: Any], + let type = obj["type"] as? String else { return } + if type == "session.updated" { + engine.enableSendMediaStream(.audio, enable: true) + engine.enableSendMediaStream(.video, enable: true) + } +} +``` + +**Android：** + +``` +@Override +public void onDataMsg(AoqDataMsg msg) { + if (msg == null || msg.data == null) return; + try { + JSONObject obj = new JSONObject(new String(msg.data, StandardCharsets.UTF_8)); + if ("session.updated".equals(obj.optString("type"))) { + engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeAudio, true); + engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeVideo, true); + } + } catch (JSONException ignored) {} +} +``` + +**HarmonyOS：** + +``` +onDataMsg(msg: AoqDataMsg): void { + if (!msg?.data) return; + try { + const text = new TextDecoder('utf-8').decode(new Uint8Array(msg.data)); + const obj = JSON.parse(text) as { type?: string }; + if (obj.type === 'session.updated') { + this.engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeAudio, true); + this.engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeVideo, true); + } + } catch (_) { /* 非 JSON，忽略 */ } +} +``` + +**重要** + +1. 模型必须在收到 `session.updated` 后才开启媒体流发送，否则 AI 侧可能还未准备好接收数据。 + +2. 建连时添加的音频轨道和视频轨道（即 AOQ 媒体通道）会自动将数据传输到服务端。 + + 1. 音频：通过音频轨道直接传输，无需发送 `input_audio_buffer.append` 事件。 + + 2. 视频：通过视频轨道发送画面帧，无需发送 `input_image_buffer.append` 事件。 + + +### **断开连接与销毁引擎** + +``` +engine.disconnect() +AoqClientEngine.destroy() +``` + +## **典型场景** + +### **打断（Barge-in）** + +- SDK 与百炼深度融合，支持百炼模型的打断消息会在新一轮对话开始时打断上一轮次。 + +- SDK 提供**本地播放器打断**接口 `interruptAudioPlayer`，当用户主动需要停止时可以调用打断 API 实现此功能。 + + +``` +// iOS +engine.interruptAudioPlayer(.audio, fadeMs: 100) +``` + +### **静音 / 取消静音** + +静音后 SDK 仍在采集音频，但只推送静音帧，`session` 不会中断。 + +``` +engine.muteAudioCapture(true); // 静音麦克风（采集仍在跑，但只送静音帧） +engine.muteAudioCapture(false); // 恢复 +``` + +### **切换前后摄像头** + +``` +// 传入期望切换到的方向枚举即可 +engine.switchCamera(AoqCameraDirection.AoqCameraDirectionFront); +engine.switchCamera(AoqCameraDirection.AoqCameraDirectionBack); +``` + +### **通话字幕与ASR结果显示** + +服务端通过下行数据消息推送 ASR 结果与 AI 文本回复。业务侧在 `onDataMsg` 回调中根据 `type` 字段分流即可。WebSocket事件说明详见[服务端事件](https://help.aliyun.com/zh/model-studio/server-events)。 + +## **注意事项** + +1. **单例语义**：`createEngine` 是单例，重复调用返回同一实例；`destroy` 后才能重新创建。多页面共用建议在 Application/Ability 级管理引擎生命周期。 + +2. **本地预览 View 类型**： + + - Android：`SurfaceView` 或 `TextureView`；其它类型不支持。 + + - iOS：任意 `UIView` 子类。 + + - HarmonyOS：请参考 SDK 文档。 + +3. **音频路由变化**：耳机插拔、蓝牙连接等会触发 `onAudioDeviceRouteChanged`，业务侧通常无需处理；如果 UI 上显示"扬声器/听筒"开关，需要根据该回调同步状态。 + +4. **后台续传**：如需通话切到后台后继续传音频，`Info.plist` 必须开启 `UIBackgroundModes = audio`，并在前台时正确激活 `AVAudioSession`（SDK 会处理大部分情况，业务侧用 `setAudioSessionRestriction:` 可精细控制是否让 SDK 接管）。 + + +## **Demo 示例下载** + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5762814871/p1087917.png) + +示例源码下载： + +**iOS：**[aoqdemo.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/grpnos/aoqdemo.zip) + +## **相关文档** + +- AOQ Client SDK 详细 API：[AOQ SDK简介](https://help.aliyun.com/zh/model-studio/realtime-api-aoq-sdk-desc/) + +- qwen3.5-omni-plus-realtime 模型客户端事件：[客户端事件](https://help.aliyun.com/zh/model-studio/client-events) + +- qwen3.5-omni-plus-realtime 模型服务端事件：[服务端事件](https://help.aliyun.com/zh/model-studio/server-events) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md new file mode 100644 index 00000000..e743e6b6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md @@ -0,0 +1,512 @@ +# 通过WebRTC使用多模态交互套件实现实时通话 + +本文档说明如何在浏览器端通过 WebRTC + JavaScript 接入通义多模态交互套件（multimodal-dialog），实现与多模态 AI 应用的实时音视频交互。 + +**说明** + +多模态交互套件面向 **AI/AR 眼镜、学习机、智能机器人**等硬件场景，提供可视化应用配置、预置 Agent/插件、音色管理等完整业务能力。WebRTC 模式下音频通过 UDP 直接传输，内置回声消除和降噪，适合浏览器端低延迟交互场景。 + +## **前提条件及注意事项** + +1. 已在百炼控制台完成以下准备： + + - 已[配置 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并将其[设置到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + + - 创建多模态交互应用，获取 **Workspace ID** 和 **App ID**（详情请参见[应用创建](https://help.aliyun.com/zh/model-studio/multimodal-app-creation)）。 + + - 在应用中完成模型、音色、提示词、Agent/插件等配置（详情请参见[应用配置](https://help.aliyun.com/zh/model-studio/multimodal-app-configuration)）。 + +2. 使用支持 WebRTC 的现代浏览器（Chrome、Edge、Firefox、Safari 等）。 + +3. 浏览器需要麦克风权限；如需视频交互，还需摄像头权限。 + +4. 浏览器无法直接向服务端发起 SDP 交换请求（受 CORS 限制），Demo 中需通过终端执行 curl 命令完成连接建立；正式产品中由业务后端代理时不存在此限制。 + + +## **实现实时通话** + +以下时序图展示了整个 WebRTC 实时通话的完整流程： + +WebRTC 多模态交互套件实时通话流程时序图 + +![1111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6265914871/p1088078.svg) + +### **创建 RTCPeerConnection** + +调用浏览器原生 `RTCPeerConnection` 构造函数创建连接实例。服务端采用 ICE-lite 模式直连，无需配置 ICE 服务器。 + +``` +pc = new RTCPeerConnection(); +``` + +同时注册关键回调： + +``` +pc.onconnectionstatechange = () => { + if (pc.connectionState === 'connected') { + // 连接成功 + } else if (['failed', 'closed', 'disconnected'].includes(pc.connectionState)) { + // 连接断开，清理资源 + endSession(); + } +}; + +pc.ontrack = (e) => { + // 将远端音频流绑定到 audio 元素播放 + const remoteAudio = document.createElement('audio'); + remoteAudio.autoplay = true; + remoteAudio.srcObject = e.streams[0]; + document.body.appendChild(remoteAudio); +}; +``` + +### **获取本地媒体流** + +通过 `navigator.mediaDevices.getUserMedia` 获取麦克风权限（必须），以及摄像头权限（可选）。 + +**纯音频模式：** + +``` +const localStream = await navigator.mediaDevices.getUserMedia({ audio: true }); +``` + +**音视频模式（需要 AI 视觉理解时）：** + +``` +const localStream = await navigator.mediaDevices.getUserMedia({ + audio: true, + video: { + facingMode: { ideal: 'environment' }, // 后置摄像头 + frameRate: { ideal: 30, max: 30 }, + width: { ideal: 640 }, + height: { ideal: 480 } + } +}); +``` + +**说明** + +视频帧率根据场景调整。需要 AI 实时理解画面的场景（如物体识别、场景描述）建议 15-30 fps。 + +### **添加媒体轨道到 PeerConnection** + +将本地音频轨道添加到 PeerConnection。如果开启了视频，视频轨道一并添加。 + +``` +localStream.getTracks().forEach(track => pc.addTrack(track, localStream)); +``` + +**维持视频发送质量（可选）：** + +在弱网环境下，浏览器可能自动降低视频分辨率。可通过设置 sender 参数尽量维持： + +``` +const sender = pc.getSenders().find(s => s.track && s.track.kind === 'video'); +if (sender) { + const params = sender.getParameters(); + if (!params.encodings || params.encodings.length === 0) params.encodings = [{}]; + params.encodings[0].scaleResolutionDownBy = 1.0; + params.encodings[0].maxBitrate = 2500000; // 2.5 Mbps + params.encodings[0].maxFramerate = 30; + params.degradationPreference = 'maintain-resolution'; + await sender.setParameters(params); +} +``` + +### **创建 DataChannel** + +创建名为 `oai-events` 的 DataChannel，用于与服务端交换控制消息（run-task、事件通知等）。 + +``` +const dc = pc.createDataChannel('oai-events'); + +dc.onopen = () => { + console.log('DataChannel open'); + // DataChannel 就绪后发送 run-task + sendStartMessage(dc); +}; + +dc.onmessage = (e) => { + const evt = JSON.parse(e.data); + handleServerEvent(evt, dc); +}; +``` + +同时监听 `pc.ondatachannel` 以处理服务端主动创建的 DataChannel： + +``` +pc.ondatachannel = (event) => { + const ch = event.channel; + if (ch.label === 'txt' || ch.label === 'oai-events') { + ch.onopen = () => sendStartMessage(ch); + ch.onmessage = (e) => handleServerEvent(JSON.parse(e.data), ch); + } +}; +``` + +### **生成 Offer SDP** + +调用 `createOffer` 并 `setLocalDescription`，浏览器生成包含本地媒体能力描述的 Offer SDP。服务端采用 ICE-lite 模式，客户端无需等待 ICE candidates 收集完成，`setLocalDescription` 后即可发送。 + +``` +const offer = await pc.createOffer(); +await pc.setLocalDescription(offer); + +// offer.sdp 即为待发送的 Offer SDP 字符串 +``` + +### **交换 SDP（HTTP POST）** + +将 Offer SDP 通过 HTTP POST 发送到服务端 WebRTC 端点，服务端返回 Answer SDP。 + +**Endpoint 格式：**`{workspace_id}.{region}.maas.aliyuncs.com`，其中 `workspace_id` 为百炼工作空间 ID（如 `llm-xxxxxxxxxx`），`region` 为部署区域（如 `cn-beijing`）。创建工作空间后即可在百炼控制台获取。 + +**请求配置：** + +**配置项** + +**说明** + +请求地址 + +`POST https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/webrtc/inference?model=multimodal-dialog` + +Content-Type + +`application/sdp` + +请求头 + +`Authorization: Bearer {DASHSCOPE_API_KEY}` + +请求体 + +客户端生成的 Offer SDP 字符串 + +响应 + +成功：HTTP 200，返回服务端 Answer SDP 字符串 + +``` +const API_KEY = 'your-api-key'; // 百炼控制台获取 +const WORKSPACE_ID = '{workspace-id}'; // 百炼控制台获取 +const REGION = 'cn-beijing'; +const SIGNALING_URL = `https://${WORKSPACE_ID}.${REGION}.maas.aliyuncs.com/api/v1/webrtc/inference?model=multimodal-dialog`; + +const resp = await fetch(SIGNALING_URL, { + method: 'POST', + headers: { + 'Content-Type': 'application/sdp', + 'Authorization': `Bearer ${API_KEY}`, + }, + body: offer.sdp, +}); + +if (!resp.ok) throw new Error('SDP 交换失败: ' + resp.status); +const answerSdp = await resp.text(); +``` + +**说明** + +由于浏览器端受 CORS 限制无法直接请求服务端，Demo 中需用户手动在终端执行 curl 命令。正式产品中应通过业务后端代理此请求。 + +**curl 命令格式：** + +``` +curl -X POST 'https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/webrtc/inference?model=multimodal-dialog' \ + -H 'Content-Type: application/sdp' \ + -H 'Authorization: Bearer $DASHSCOPE_API_KEY' \ + --data-binary '' +``` + +### **设置 Answer SDP 建立连接** + +将从服务端获取的 Answer SDP 设置为远端描述，WebRTC 连接随即建立。 + +``` +await pc.setRemoteDescription({ type: 'answer', sdp: answerSdp }); +// 连接建立完成，pc.connectionState 将变为 'connected' +``` + +### **发送 run-task 启动会话** + +WebRTC 连接建立、DataChannel 就绪后，客户端发送 **run-task** 消息启动多模态会话。 + +``` +let currentTaskId = null; + +function sendStartMessage(channel) { + currentTaskId = generateTaskId(); + const msg = { + payload: { + input: { + workspace_id: '{workspace-id}', + app_id: '{app-id}', + directive: 'Start' + }, + task_group: 'aigc', + task: 'multimodal-generation', + function: 'generation', + model: 'multimodal-dialog', + parameters: { + client_info: { + user_id: '{user-id}', + device: { uuid: '{device-uuid}' }, + network: { ip: '{client-ip}' } + }, + upstream: { + mode: 'duplex', + sample_rate: '16000', + type: 'AudioAndVideo' // 'Audio' 或 'AudioAndVideo' + }, + dialog_attributes: { + vocabulary_id: '{vocabulary-id}' // 可选 + }, + downstream: { + voice: 'longanhuan', + sample_rate: 24000, + audio_format: 'pcm' + } + } + }, + header: { + streaming: 'duplex', + action: 'run-task', + task_id: currentTaskId + } + }; + channel.send(JSON.stringify(msg)); +} + +function generateTaskId() { + if (window.crypto && typeof window.crypto.randomUUID === 'function') { + return window.crypto.randomUUID().replace(/-/g, ''); + } + const r = () => Math.random().toString(16).slice(2); + return (Date.now().toString(16) + r() + r()).slice(0, 32); +} +``` + +**参数说明：** + +**参数路径** + +**类型** + +**说明** + +`payload.input.workspace_id` + +string + +百炼工作空间 ID，在控制台「应用管理」中获取 + +`payload.input.app_id` + +string + +多模态交互应用 ID，在控制台「应用管理」中获取 + +`payload.input.directive` + +string + +固定值 `Start` + +`payload.parameters.client_info.user_id` + +string + +业务系统中的用户标识，用于日志追踪 + +`payload.parameters.client_info.device.uuid` + +string + +设备唯一标识，用于设备维度的数据分析 + +`payload.parameters.upstream.mode` + +string + +交互模式：`duplex`（全双工）、`push2talk`（按住说话）、`tap2talk`（点击说话） + +`payload.parameters.upstream.type` + +string + +上行媒体类型：`Audio`（纯语音）或 `AudioAndVideo`（音视频） + +`payload.parameters.upstream.sample_rate` + +string + +上行音频采样率，通常 `16000` + +`payload.parameters.downstream.voice` + +string + +下行音色，可在百炼控制台音色列表中选取 + +`payload.parameters.downstream.sample_rate` + +number + +下行音频采样率，通常 `24000` + +`payload.parameters.dialog_attributes.vocabulary_id` + +string + +可选，热词表 ID，提升专有名词识别准确率 + +**说明** + +应用中的模型选择、提示词、Agent/插件、知识库等配置均在百炼控制台可视化完成，无需通过代码传入。run-task 只需指定 `workspace_id` 和 `app_id`，服务端会自动加载对应配置。 + +### **实时对话** + +run-task 发送成功后，进入实时对话状态： + +- **上行**：浏览器采集的音频/视频通过 RTP 协议自动发送到服务端 + +- **下行音频**：AI 语音回复通过 `ontrack` 回调接收并播放 + +- **下行事件**：通过 DataChannel 接收业务事件 + + +``` +function handleServerEvent(evt, channel) { + const type = evt.type || evt.header?.action; + + switch (type) { + case 'open_videochat': + // 服务端请求开启视频通道 + // 延迟数秒响应，确保视频处理通道就绪 + setTimeout(() => { + channel.send(JSON.stringify({ + payload: { + input: { text: '', type: 'prompt', directive: 'RequestToRespond' }, + parameters: { + biz_params: { + videos: [{ action: 'connect', type: 'voicechat_video_channel' }] + } + } + }, + header: { + streaming: 'duplex', + action: 'continue-task', + task_id: currentTaskId + } + })); + }, 3000); + break; + + default: + console.log('[服务端事件]', type, evt); + break; + } +} +``` + +**说明** + +**open\_videochat 机制**：当 `upstream.type` 设为 `AudioAndVideo` 时，客户端已通过 WebRTC 发送视频轨道。服务端在需要时（如 AI Agent 判断需要"看"画面）会推送 `open_videochat` 事件，客户端需回复 `continue-task` 确认视频通道建立。该机制允许按需开启视频处理，节省服务端资源。 + +**静音 / 取消静音：** + +``` +// 静音 +localStream.getAudioTracks().forEach(t => { t.enabled = false; }); +// 取消静音 +localStream.getAudioTracks().forEach(t => { t.enabled = true; }); +``` + +**开启/关闭视频：** + +``` +// 关闭视频 +localStream.getVideoTracks().forEach(t => { t.enabled = false; }); +// 开启视频 +localStream.getVideoTracks().forEach(t => { t.enabled = true; }); +``` + +### **结束会话与资源清理** + +通话结束后需要正确释放所有资源，避免内存泄漏和设备占用。 + +``` +function endSession() { + // 1. 关闭 DataChannel + if (dataChannel) { + dataChannel.close(); + dataChannel = null; + } + + // 2. 停止本地媒体流（释放麦克风/摄像头） + if (localStream) { + localStream.getTracks().forEach(t => t.stop()); + localStream = null; + } + + // 3. 关闭 PeerConnection + if (pc) { + pc.close(); + pc = null; + } + + // 4. 重置状态 + currentTaskId = null; +} +``` + +## **注意事项** + +1. **API Key 安全**：切勿将 API Key 硬编码在前端代码中。生产环境应通过后端服务代理 SDP 交换请求，API Key 仅存放在服务端。 + +2. **CORS 限制**：浏览器端无法直接调用百炼 API 进行 SDP 交换，正式产品中需要通过后端代理转发请求。 + +3. **HTTPS 要求**：`getUserMedia` 在非 localhost 环境下要求页面必须通过 HTTPS 提供服务。 + +4. **交互模式选择**：多模态套件支持三种交互模式——`duplex`（全双工，用户可随时打断）、`push2talk`（按住说话）、`tap2talk`（点击说话）。根据硬件形态选择合适的模式。 + +5. **视频帧率**：视频理解场景建议 15-30 fps；如果仅需偶尔拍照识别，可降低帧率节省带宽。 + +6. **浏览器兼容性**：推荐使用 Chrome 90+、Edge 90+、Firefox 85+、Safari 15+。 + +7. **单实例限制**：同一页面同时只应维护一个 `RTCPeerConnection` 实例，创建新会话前需先关闭旧连接。 + + +## **完整 Demo 示例下载** + +以下是一个完整的 HTML 页面 Demo，可直接在浏览器中运行体验多模态交互。由于浏览器 CORS 限制，SDP 交换通过 curl 命令手动完成。 + +[webrtc\_multimodel\_demo.html](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/ifirko/webrtc_multimodel_demo.html) + +**使用步骤：** + +1. 在浏览器中打开该文件。 + +2. 填写连接配置：Endpoint（格式为 `{workspace_id}.{region}.maas.aliyuncs.com`）、API Key、Workspace ID 和 App ID。 + +3. 如需视频交互，勾选“开启视频”。 + +4. 点击**开始会话**，允许浏览器访问麦克风（及摄像头）。 + +5. 如果浏览器能直接发起请求（无 CORS 限制），将自动完成连接；否则页面会显示 curl 命令，复制到终端执行后将返回的 Answer SDP 粘贴回页面即可。 + +6. 连接建立后，对着麦克风说话即可与多模态 AI 实时对话。 + +7. 通话结束后可点击“下载远端音频”保存 AI 回复的录音。 + + +## **相关文档** + +- [通义多模态交互开发套件产品概述](https://help.aliyun.com/zh/model-studio/multimodal-products-overview) + +- [多模态交互套件使用指南](https://help.aliyun.com/zh/model-studio/multimodal-guidelines/) + +- [多模态交互 SDK（Python/Java）GitHub 示例代码](https://github.com/aliyun/alibabacloud-bailian-speech-demo/tree/master/samples/conversation/multimodal_dialog) + +- [WebRTC API (MDN)](https://developer.mozilla.org/zh-CN/docs/Web/API/WebRTC_API) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md new file mode 100644 index 00000000..b5d496b5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md @@ -0,0 +1,388 @@ +# 通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话 + +本文档说明如何在浏览器端通过 WebRTC + JavaScript 接入百炼 Realtime API，实现与 qwen3.5-omni-plus-realtime 模型的实时音视频通话。 + +**说明** + +WebRTC 适合浏览器端、低延迟语音场景，音频通过 UDP 直接传输，内置回声消除和降噪。WebRTC 仅支持服务端 VAD 模式（`server_vad` 或 `semantic_vad`），不支持手动模式。 + +## **前提条件及注意事项** + +1. 已[配置 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并将其[设置到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + +2. 使用支持 WebRTC 的现代浏览器（Chrome、Edge、Firefox、Safari 等）。 + +3. 浏览器需要麦克风权限；如需视频通话，还需摄像头权限。 + +4. 浏览器无法直接向服务端发起 SDP 交换请求（受 CORS 限制），Demo 中通过终端执行 curl 命令完成连接建立；正式使用时由业务 AppServer 代理完成，无此限制。 + + +## **实现 AI 音视频通话** + +以下时序图展示了整个 WebRTC 音视频通话的完整流程： + +WebRTC 音视频通话流程时序图 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5665914871/p1088079.png) + +### **创建 RTCPeerConnection** + +调用浏览器原生 `RTCPeerConnection` 创建连接实例，无需配置 ICE 服务器（服务端会处理 NAT 穿透）。 + +``` +pc = new RTCPeerConnection({ iceServers: [ ] }); +``` + +注册关键回调： + +``` +// 连接状态监听 +pc.onconnectionstatechange = () => { + if (!pc) return; + if (pc.connectionState === 'connected') { + setStatus('已连接，请说话', 'connected'); + } else if (["failed", "closed", "disconnected"].includes(pc.connectionState)) { + endSession(true); + } +}; + +// 接收远端音频流并播放 + 启动录制 +pc.ontrack = async (e) => { + const stream = e.streams[0]; + ensureHiddenAudioEl(); + hiddenRemoteAudioEl.srcObject = stream; + try { await hiddenRemoteAudioEl.play(); } catch {} + startRecordingRemoteStream(stream); +}; +``` + +### **获取本地媒体流** + +通过一次 `getUserMedia` 调用获取所需的音频（必须）和视频（可选）。是否开启视频由用户勾选"开启视频"复选框决定。 + +``` +const wantVideo = !!sendVideoCheckbox.checked; + +const constraints = wantVideo + ? { + audio: true, + video: { + facingMode: { ideal: "user" }, + frameRate: { ideal: 30, max: 30 }, + width: { ideal: 640 }, + height: { ideal: 480 }, + } + } + : { audio: true }; + +localStream = await navigator.mediaDevices.getUserMedia(constraints); +``` + +**说明** + +音频和视频通过**同一次** `getUserMedia` 调用获取，而非分开请求。视频预览帧率为 30fps（本地流畅预览），发送帧率会通过 Canvas 降至 2fps。 + +### **添加媒体轨道到 PeerConnection** + +**添加音频轨道：** + +``` +localStream.getAudioTracks().forEach(t => { + pc.addTrack(t, localStream); + gatedAudioTracks.push(t); +}); +``` + +**添加视频轨道（可选，通过 Canvas 降帧至 2fps）：** + +Canvas 尺寸从摄像头实际分辨率动态获取，而非硬编码： + +``` +const sendFps = 2; +const settings = localStream.getVideoTracks()[0].getSettings(); +sendCanvas = document.createElement("canvas"); +sendCanvas.width = settings.width || 640; // 动态获取实际宽度 +sendCanvas.height = settings.height || 480; // 动态获取实际高度 +sendCanvasCtx = sendCanvas.getContext("2d", { alpha: false }); + +sendCanvasStream = sendCanvas.captureStream(sendFps); // 2fps +const lowFpsTrack = sendCanvasStream.getVideoTracks()[0]; +pc.addTrack(lowFpsTrack, sendCanvasStream); +gatedVideoTracks.push(lowFpsTrack); + +// requestAnimationFrame 循环：将摄像头画面绘制到 Canvas +const pump = () => { + if (!sendCanvasCtx || !sendCanvas) return; + try { sendCanvasCtx.drawImage(localVideo, 0, 0, sendCanvas.width, sendCanvas.height); } catch {} + sendRafId = requestAnimationFrame(pump); +}; +sendRafId = requestAnimationFrame(pump); +``` + +**媒体门控（关键）：** + +添加轨道后立即禁止发送，确保在收到 `session.created` 之前不推送媒体数据： + +``` +// 1. 禁用所有轨道的 enabled +gateMedia(false); // track.enabled = false + +// 2. 将 sender 的 track 替换为 null，彻底阻止发送 +audioSender = pc.getSenders().find(s => s.track?.kind === 'audio'); +videoSender = pc.getSenders().find(s => s.track?.kind === 'video'); +audioTrack = audioSender?.track; +videoTrack = videoSender?.track; +await audioSender?.replaceTrack(null); +await videoSender?.replaceTrack(videoTrack ? null : undefined); +``` + +**说明** + +等价于其他 SDK 中的 `enableSendMediaStream(false)`，必须在收到 `session.created` 后才恢复发送。 + +### **创建 DataChannel** + +创建名为 `oai-events` 的 DataChannel，用于与 AI 服务端交换会话控制事件。 + +``` +const dc = pc.createDataChannel('oai-events'); + +dc.onopen = () => console.log("DC open"); +dc.onmessage = (e) => { + handleDcMessage(e.data, dc); +}; + +// 同时监听服务端主动创建的 DataChannel +pc.ondatachannel = (event) => { + const ch = event.channel; + ch.onmessage = (e) => { + handleDcMessage(e.data, ch); + }; +}; +``` + +### **生成 Offer SDP** + +调用 `createOffer()` 并设置本地描述，等待 ICE 候选收集完成后获取完整的 Offer SDP。 + +``` +pc.onicegatheringstatechange = () => { + if (!pc) return; + if (pc.iceGatheringState === "complete" && pc.localDescription?.sdp) { + const sdp = pc.localDescription.sdp; + // ICE 收集完成，Offer SDP 可用 + // 自动生成 curl 命令供用户使用 + } +}; + +const offer = await pc.createOffer(); +await pc.setLocalDescription(offer); +``` + +**说明** + +必须等待 `iceGatheringState === "complete"` 后再使用 SDP，此时 SDP 中包含所有 ICE 候选信息。 + +### **交换 SDP（通过 curl 命令或业务 AppServer）** + +将 Offer SDP 发送到百炼服务端，获取 Answer SDP。Demo 中通过 curl 命令完成： + +``` +curl -X POST 'https://{endpoint}/api/v1/webrtc/realtime?model=qwen3.5-omni-plus-realtime' \ + -H 'Content-Type: application/sdp' \ + -H 'Authorization: Bearer $DASHSCOPE_API_KEY' \ + --data-binary '' +``` + +**说明** + +生产环境中，此步骤应由业务 AppServer 代理完成，避免前端暴露 API Key。`{endpoint}` 为 Realtime API 接入地址。 + +### **设置 Answer SDP 建立连接** + +将服务端返回的 Answer SDP 设置为远端描述，WebRTC 连接即开始建立。注意 SDP 格式需要规范化处理： + +``` +function normalizeSdpForSetRemote(sdp) { + sdp = String(sdp).trim().replace(/\r?\n/g, "\r\n"); + if (!sdp.endsWith("\r\n")) sdp += "\r\n"; + return sdp; +} + +const answerSdp = normalizeSdpForSetRemote(txt); +await pc.setRemoteDescription({ type: 'answer', sdp: answerSdp }); +``` + +**说明** + +SDP 规范要求行尾为 `\r\n`，`normalizeSdpForSetRemote` 负责处理不同来源的换行符兼容问题。 + +### **配置 AI 会话（session.update）** + +连接建立后，服务端通过 DataChannel 发送 `session.created` 事件。收到后需： + +1. 解除媒体门控，恢复音视频发送 + +2. 发送 `session.update` 配置会话参数 + + +**解除门控并恢复媒体：** + +``` +function handleDcMessage(data, channel) { + let obj; + try { obj = JSON.parse(data); } catch (err) { return; } + + if (obj?.type === "session.created") { + // 解除门控：恢复 track.enabled + gateMedia(true); + // 恢复 sender 的实际 track + if (audioSender) audioSender.replaceTrack(audioTrack); + if (videoSender && videoTrack) videoSender.replaceTrack(videoTrack); + // 发送会话配置 + sendUpdate(channel); + } +} +``` + +**session.update 消息体：** + +``` +const update = { + event_id: `event_${Date.now()}`, + type: "session.update", + session: { + input_audio_format: "pcm", + input_audio_transcription: { model: "qwen3-asr-flash-realtime" }, + instructions: "You are a helpful assistant.", + modalities: ["text", "audio"], + output_audio_format: "pcm", + smooth_output: false, + turn_detection: { + prefix_padding_ms: 500, + silence_duration_ms: 800, + threshold: 0.5, + type: "server_vad", + }, + }, +}; +if (channel && channel.readyState === "open") channel.send(JSON.stringify(update)); +``` + +**说明** + +`turn_detection.type` 可设为 `server_vad`（基于音量检测）或 `semantic_vad`（基于语义检测）。WebRTC 模式不支持手动 VAD。 + +### **实时对话** + +连接建立后，音视频通过 RTP 实时传输。远端 AI 语音通过 `ontrack` 回调接收并播放，同时使用 MediaRecorder 录制以便下载。 + +**接收远端音频并录制：** + +``` +pc.ontrack = async (e) => { + const stream = e.streams[0]; + ensureHiddenAudioEl(); + hiddenRemoteAudioEl.srcObject = stream; + try { await hiddenRemoteAudioEl.play(); } catch {} + startRecordingRemoteStream(stream); // 启动录制 +}; + +function startRecordingRemoteStream(remoteStream) { + const audioTracks = remoteStream.getAudioTracks(); + if (!audioTracks.length) return; + const audioStream = new MediaStream(audioTracks); + + recordedChunks = [ ]; + + mediaRecorder = new MediaRecorder(audioStream, { mimeType: 'audio/webm' }); + mediaRecorder.ondataavailable = (e) => { + if (e.data && e.data.size > 0) recordedChunks.push(e.data); + }; + mediaRecorder.onstop = () => { + audioBlob = new Blob(recordedChunks, { type: 'audio/webm' }); + // 录制结束后可下载 + }; + mediaRecorder.start(); +} +``` + +**DataChannel 事件统一展示：** + +所有通过 DataChannel 收发的事件（包括 `session.created`、`response.audio_transcript.done` 等）统一通过事件面板展示，支持展开查看完整 JSON： + +``` +function pushEventFromDataChannel(eventObj) { + const ts = eventObj.timestamp || nowTs(); + events.unshift({ event: eventObj, timestamp: ts }); + renderEvents(); +} +``` + +### **结束会话与资源清理** + +结束通话时需依次清理所有资源，顺序很重要： + +``` +function endSession(silent = false) { + // 1. 停止 Canvas 降帧循环 + if (sendRafId) cancelAnimationFrame(sendRafId); + sendRafId = 0; + if (sendCanvasStream) sendCanvasStream.getTracks().forEach(t => t.stop()); + sendCanvasStream = null; sendCanvasCtx = null; sendCanvas = null; + + // 2. 停止录制 + try { if (mediaRecorder && mediaRecorder.state !== "inactive") mediaRecorder.stop(); } catch {} + mediaRecorder = null; + + // 3. 停止本地媒体流 + if (localStream) { + localStream.getTracks().forEach(t => t.stop()); + localStream = null; + } + + // 4. 关闭 PeerConnection + if (pc) { try { pc.close(); } catch {} pc = null; } + + // 5. 清理远端音频元素 + if (hiddenRemoteAudioEl) { + try { hiddenRemoteAudioEl.pause(); } catch {} + hiddenRemoteAudioEl.srcObject = null; + hiddenRemoteAudioEl.remove(); + hiddenRemoteAudioEl = null; + } +} +``` + +**说明** + +结束后可通过"下载远端音频"按钮下载 AI 回复的录音（WebM 格式）。 + +## **注意事项** + +1. **媒体门控必须在 session.created 后解除**：在服务端发送 `session.created` 之前推送媒体数据会被丢弃，必须通过 `replaceTrack(null)` 彻底阻断发送。 + +2. **视频降帧通过 Canvas 实现**：本地预览 30fps，发送至服务端仅 2fps，通过 `captureStream(2)` 控制，节省带宽。 + +3. **SDP 格式规范化**：设置 Answer SDP 前必须确保行尾为 `\r\n`，否则 `setRemoteDescription` 可能失败。 + +4. **视频为可选功能**：用户未勾选视频时，仅请求音频权限，不会触发摄像头授权弹窗。 + +5. **远端音频自动录制**：通过 MediaRecorder 录制 AI 回复的音频流，会话结束后可下载 WebM 格式文件。 + +6. **WebRTC 仅支持服务端 VAD**：不支持 `manual` 模式，可选 `server_vad`（音量检测）或 `semantic_vad`（语义检测）。 + + +## **完整 demo 下载** + +完整示例代码请下载：[webrtc\_demo.html](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/ychtmj/webrtc_demo.html)。 + +## **相关文档** + +- [WebRTC API (MDN)](https://developer.mozilla.org/zh-CN/docs/Web/API/WebRTC_API) + +- [RTCPeerConnection (MDN)](https://developer.mozilla.org/zh-CN/docs/Web/API/RTCPeerConnection) + +- qwen3.5-omni-plus-realtime 模型客户端事件：[客户端事件](https://help.aliyun.com/zh/model-studio/client-events) + +- qwen3.5-omni-plus-realtime 模型服务端事件：[服务端事件](https://help.aliyun.com/zh/model-studio/server-events) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md new file mode 100644 index 00000000..217e42ed --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md @@ -0,0 +1,180 @@ +# Realtime API简介 + +Realtime API 是一系列针对性能、延迟、抗弱网、对接成本、适配性提供多种对接方式的方法，供客户灵活选择。 + +## **概述** + +Realtime API 支持 **WebSocket**、**WebRTC** 和 **AOQ（AI over QUIC）**三种传输协议，开发者可以根据业务场景灵活选择。 + +**维度** + +**WebSocket** + +**WebRTC** + +**AOQ** + +适用场景 + +服务端集成、快速原型验证 + +浏览器端互动、传统音视频通话 + +AI 多模态实时交互、弱网场景、混合数据传输 + +浏览器兼容性 + +原生支持 + +原生支持 + +不支持 + +接入难度 + +极低 + +中等 + +低 + +弱网对抗 + +差 + +良好 + +极致 + +数据类型 + +文本/音频/图像 + +音视频 + 文本 + +音视频 + 文本 + +建连速度 + +慢 + +慢 + +快 + +回声消除/降噪 + +无，需客户端自行处理 + +内置 + +内置 + +AI 场景适配 + +基础，适合纯文本或低实时性场景 + +传统设计，AI 场景需额外适配 + +原生为 AI 多模态数据特征深度定制 + +端侧平台支持 + +全平台（任何支持 WebSocket 的环境） + +浏览器、移动端 + +Android / iOS / HarmonyOS + +开发者可根据实际需求选择协议方案： + +- **WebSocket 方案**：适合服务端集成、快速原型验证、对接入门槛要求极低的场景。通过 DashScope SDK 可快速实现实时语音对话。 + +- **WebRTC 方案**：适合需要浏览器原生支持、已有 WebRTC 基础设施的传统音视频通话场景，内置回声消除和降噪能力。 + +- **AOQ 方案**：适合对延迟、弱网对抗、多模态数据传输有极致要求的 AI 实时交互场景，同时内置回声消除和降噪能力，尤其是移动端原生应用。 + + +## **模型/应用支持力度** + +不同协议对模型和应用的支持情况如下： + +**模型/应用类型** + +**模型** + +**AOQ** + +**WebRTC** + +**WebSocket** + +实时全模态 + +qwen3.5-omni-plus-realtime + +支持 + +支持 + +支持 + +qwen3.5-omni-flash-realtime + +支持 + +支持 + +支持 + +qwen3.5-livetranslate-flash-realtime + +支持 + +支持 + +支持 + +多模态开发套件 + +multimodal-dialog + +不支持 + +支持 + +支持 + +实时语音识别 + +Fun-ASR系列模型 + +不支持 + +不支持 + +支持 + +实时语音合成 + +CosyVoice系列模型 + +不支持 + +不支持 + +支持 + +实时语音对话 + +qwen-audio-3.0-realtime-plus、qwen-audio-3.0-realtime-flash + +不支持 + +不支持 + +支持 + +**说明** + +模型的名称、上下文、价格、快照版本等信息请参见[阿里云百炼控制台](https://bailian.console.aliyun.com/cn-beijing#/home)；并发限流条件请参考[限流](https://help.aliyun.com/zh/model-studio/rate-limit)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md new file mode 100644 index 00000000..147179b9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md @@ -0,0 +1,291 @@ +# 实现接通模型/应用 + +介绍如何通过 AOQ、WebRTC、WebSocket 三种协议接入 Realtime API 模型或应用，包含各协议的连接流程、时序图和代码示例。 + +## **AOQ 接入** + +AOQ 基于 QUIC 协议深度定制，适合移动端原生应用，支持音频/视频/数据混合传输，内置极致抗弱网能力。以下以 iOS Demo 为例。 + +### **整体流程时序图** + +![AOQ中文1](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5755914871/p1088073.jpg) + +### **创建引擎并设置回调** + +``` +let config = AoqCreateConfig() +config.workDir = workDir +config.enableDumpAudio = false +engine = AoqClientEngine.createEngine(config, delegate: self) +``` + +实现 `AoqEngineDelegate` 协议监听 `onConnectionStatusChange`、`onDataMsg`、`onError` 等回调。 + +### **启动音频采集与播放** + +``` +// 音频采集 +let capCfg = AoqAudioCaptureConfig() +capCfg.channel = 1; capCfg.isExternal = false +engine.startAudioCapture(capCfg) + +// 音频播放 +let playCfg = AoqAudioPlaybackConfig() +playCfg.channel = 1; playCfg.isExternal = false +engine.startAudioPlayer(playCfg) + +// 视频采集（可选） +let vidCfg = AoqVideoCaptureConfig() +vidCfg.width = 720; vidCfg.height = 1280; vidCfg.fps = 15 +engine.startVideoCapture(vidCfg) +``` + +### **获取连接凭证** + +由业务 AppServer 代理百炼请求，参考 [Token 鉴权](https://help.aliyun.com/zh/model-studio/realtime-token-authentication) 章节。 + +### **设置编解码及建立连接** + +设置编解码参数后调用 `connect`： + +``` +// 音频编解码配置 +let encCfg = AoqAudioCodecConfig() +encCfg.codecType = .audioPCM; encCfg.sampleRate = 16000; encCfg.channel = 1 +engine.setAudioEncoderConfig(encCfg) +engine.setAudioDecoderConfig(encCfg) + +// connect 前关闭媒体发送，待 session.updated 后再开启 +engine.enableSendMediaStream(.audio, enable: false) + +let config = AoqConnectConfig() +config.token = token +config.sid = sid +config.certFingerprint = certificate +config.relayEndpoints = relayEndpoints +config.workspaceIdHash = workspaceIdHash +config.publishTracks = [audioTrack, dataTrack] +config.subscribeTracks = [audioTrack, dataTrack] +engine.connect(config) +``` + +**重要** + +**重要**：AOQ SDK 在建联后会默认发送媒体数据，此示例演示了连接模型时关闭媒体发送的能力。 + +### **配置 AI 会话** + +连接成功后发送 `session.update` 的示例，详见[模型客户端事件参考](https://help.aliyun.com/zh/model-studio/client-events)： + +``` +func onConnectionStatusChange(_ status: AoqConnectionStatus) { + if status == .connected { sendSessionUpdate() } +} + +private func sendSessionUpdate() { + let json = """ + { + // 该事件的id，由客户端生成 + "event_id": "event_ToPZqeobitzUJnt3QqtWg", + // 事件类型，固定为session.update + "type": "session.update", + // 会话配置 + "session": { + // 输出模态，支持设置为["text"]（仅输出文本）或["text","audio"]（输出文本与音频）。 + "modalities": [ + "text", + "audio" + ], + // 输出音频的音色 + "voice": "Ethan", + // 输入音频格式，当前仅支持设置为pcm。输入音频为16 kHz采样率的PCM音频流。 + "input_audio_format": "pcm", + // 输出音频格式，当前仅支持设置为pcm。输出音频为24 kHz采样率的PCM音频流。 + "output_audio_format": "pcm", + // 系统消息，用于设定模型的目标或角色。 + "instructions": "你是某五星级酒店的AI客服专员，请准确且友好地解答客户关于房型、设施、价格、预订政策的咨询。请始终以专业和乐于助人的态度回应，杜绝提供未经证实或超出酒店服务范围的信息。", + // 是否开启语音活动检测。若需启用，需传入一个配置对象，服务端将据此自动检测语音起止。 + // 设置为null表示由客户端决定何时发起模型响应。 + "turn_detection": { + // VAD类型，取值为server_vad或semantic_vad。使用qwen3.5-omni-realtime模型时推荐设为semantic_vad。 + "type": "semantic_vad", + // VAD检测阈值。建议在嘈杂的环境中增加，在安静的环境中降低。 + "threshold": 0.5, + // 检测语音停止的静音持续时间，超过此值后会触发模型响应 + "silence_duration_ms": 800 + } + } + } + """ + let msg = AoqDataMsg() + msg.data = json.data(using: .utf8)! + engine.send(msg) +} +``` + +### **收到 session.updated 后开启媒体发送** + +收到模型回复 `session.updated` 的示例，详见[模型服务器事件参考](https://help.aliyun.com/zh/model-studio/server-events)： + +``` +func onDataMsg(_ msg: AoqDataMsg) { + guard let obj = try? JSONSerialization.jsonObject(with: msg.data) as? [String: Any], + let type = obj["type"] as? String else { return } + if type == "session.updated" { + engine.enableSendMediaStream(.audio, enable: true) + engine.enableSendMediaStream(.video, enable: true) + } +} +``` + +**重要** + +**重要**： + +1. 模型必须在收到 `session.updated` 后才开启媒体流发送，否则 AI 侧可能还未准备好接收数据。 + +2. 建连时添加的音频轨道和视频轨道（即 AOQ 媒体通道）会自动将数据传输到服务端。 + + 1. 音频：通过音频轨道直接传输，无需发送 `input_audio_buffer.append` 事件。 + + 2. 视频：通过视频轨道发送画面帧，无需发送 `input_image_buffer.append` 事件。 + + +### **断开连接与销毁引擎** + +``` +engine.disconnect() +AoqClientEngine.destroy() +``` + +## **WebRTC 接入** + +WebRTC 协议不提供 SDK，Web 端可以通过 JavaScript，其他端可以通过开源项目或者第三方支持标准 WebRTC 协议的 RTC 服务商进行接入。以下文档以 Web 端 JavaScript 为例进行介绍。 + +### **整体流程图** + +![AOQ中文2](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5755914871/p1088074.jpg) + +### **建立连接** + +``` +# pip install aiortc aiohttp certifi +import asyncio, aiohttp, ssl, certifi +from aiortc import RTCPeerConnection, RTCConfiguration, RTCSessionDescription +from aiortc.mediastreams import AudioStreamTrack + +API_KEY = "your-api-key" +MODEL = "目标模型" +SIGNALING_URL = f"https://{{endpoint}}/api/v1/webrtc/realtime?model={MODEL}" + +async def connect(): + pc = RTCPeerConnection(RTCConfiguration(iceServers=[])) + + # 添加音频轨道，确保 Offer SDP 包含 m=audio（服务端必需） + pc.addTrack(AudioStreamTrack()) + + # 创建 DataChannel 以触发 SDP 协商（名称可自定义，服务端会通过名为 "txt" 的通道推送事件） + pc.createDataChannel("oai-events") + + # SDP 交换：创建 Offer 并发送到服务端 + offer = await pc.createOffer() + await pc.setLocalDescription(offer) + + async with aiohttp.ClientSession() as session: + async with session.post( + SIGNALING_URL, + ssl=ssl.create_default_context(cafile=certifi.where()), + data=offer.sdp.encode("utf-8"), + headers={ + "Content-Type": "application/sdp", + "Authorization": f"Bearer {API_KEY}", + }, + ) as resp: + if not resp.ok: + raise Exception(f"SDP 交换失败: {resp.status} {await resp.text()}") + answer_sdp = await resp.text() + + print("=== Offer SDP ===") + print(offer.sdp) + print("=== Answer SDP ===") + print(answer_sdp) + + # ICE 建连自动完成 + await pc.setRemoteDescription(RTCSessionDescription(sdp=answer_sdp, type="answer")) + print("WebRTC 连接已建立") + return pc +``` + +### **配置目标模型参数** + +监听模型返回的 DataChannel 消息保证交互时序： + +``` +pc.ondatachannel = (event) => { + const ch = event.channel; + ch.onmessage = (e) => { + let obj; + try { obj = JSON.parse(e.data); } + catch (err) { + return; + } + if (obj?.type === "session.created") { + sendUpdate(event.channel); + //开始推送音视频 + audioSender?.replaceTrack(audioTrack); + videoSender?.replaceTrack(videoTrack); + } + }; +}; +``` + +### **收发媒体数据** + +建连时添加的音频轨道和视频轨道（即 RTP 媒体通道）会自动将数据传输到服务端。 + +- 音频：通过音频轨道（RTP）直接传输，无需发送 `input_audio_buffer.append` 事件。 + +- 图片：通过视频轨道（RTP）发送画面帧，不支持 `input_image_buffer.append` 事件。 + + +**说明** + +WebRTC 仅支持服务端 VAD 模式（`server_vad` 或 `semantic_vad`），不支持手动模式。 + +### **Demo 源码** + +#### **前提条件** + +- 使用支持 WebRTC 的现代浏览器（Chrome、Edge、Firefox、Safari 等）。 + +- 浏览器需要麦克风权限。 + +- 浏览器无法直接向服务端发起建立连接的请求（受浏览器跨域安全策略限制），因此需要通过终端执行 curl 命令来完成连接建立。 + + +#### **运行示例** + +新建一个 HTML 文件，命名为 `webrtc_demo.html`，并将以下代码复制到文件中： + +[webrtc\_demo.html](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/crtwmi/webrtc_demo.html)。 + +在浏览器中打开此文件，按以下步骤操作： + +1. 点击开始会话，页面会自动生成 Offer SDP 和对应的 curl 命令。 + +2. 点击复制 curl 命令，在终端中执行。命令返回的内容即为 Answer SDP。 + +3. 将 Answer SDP 粘贴到页面的 Answer SDP 文本框中，点击设置 Answer 即可建立连接并开始语音对话。 + + +## **WebSocket 接入** + +可以通过 DashScope SDK 或者模型的 API 进行接入，详见： + +- [实时全模态](https://help.aliyun.com/zh/model-studio/realtime#bdaa43cdd7hsd) + +- [多模态开发套件](https://help.aliyun.com/zh/model-studio/multimodal-interaction-protocol/) + +- [实时语音识别](https://help.aliyun.com/zh/model-studio/fun-asr-realtime-websocket-api) + +- [实时语音合成](https://help.aliyun.com/zh/model-studio/cosyvoice-websocket-api) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md new file mode 100644 index 00000000..69982edf --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md @@ -0,0 +1,60 @@ +# SDK下载 + +本文提供AOQ SDK下载链接，介绍如何集成SDK，以及SDK相关信息。 + +## **AOQ SDK下载** + +**版本** + +**平台** + +**下载** + +**更新日期** + +**更新说明** + +v1.0.1 + +Android + +[AoqClientSdk-v1.0.1.aar](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/sqsjum/AoqClientSdk-v1.0.1.aar) + +[libPluginOpus.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/oxzenn/libPluginOpus.zip) + +2026-07-09 + +1. 支持AOQ协议接入模型/应用 + +2. 支持音视频编解码参数 + +3. 支持音视频设备采集及播放 + +4. 支持自定义外部输入音频采集 + +5. 支持自定义外部输入视频采集 + +6. 支持自定义外部输入视频编码 + +7. 支持控制媒体流的发送 + + +iOS + +[AoqClientSdk-v1.0.1.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/vlyumo/AoqClientSdk-v1.0.1.zip) + +[PluginOpus.framework.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/rbvmfq/PluginOpus.framework.zip) + +Harmony + +[AoqClientSdk-v1.0.1.har](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/zwlikp/AoqClientSdk-v1.0.1.har) + +[libPluginOpus.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/wvtyrh/libPluginOpus.zip) + +**重要** + +AOQ使用音频插件的方式加载Opus编解码器，如果需要Opus编解码时需要下载opus插件并加载到工程内。 + +## **WebSocket SDK下载** + +参见[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md new file mode 100644 index 00000000..6830d3a4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md @@ -0,0 +1,358 @@ +# Token鉴权 + +介绍 Realtime API 的 Token 鉴权机制，包括 API Key 的获取方式以及 WebSocket、WebRTC、AOQ 三种协议的建连鉴权方法。 + +## **概述** + +Realtime API 使用 **API Key** 进行身份认证。无论您选择 WebSocket、WebRTC 还是 AOQ 协议接入，均通过 HTTP 请求头中的 `Authorization` 字段携带 Bearer Token 完成身份验证。 + +鉴权发生在**建连阶段**，连接建立后的音视频/数据传输无需重复鉴权。 + +三种协议的鉴权差异： + +**协议** + +**鉴权时机** + +**鉴权方式** + +**说明** + +WebSocket + +WebSocket 连接握手时 + +HTTP Header `Authorization: Bearer ` + +客户端或服务端直接携带 API Key 建连 + +WebRTC + +SDP 交换 HTTP 请求时 + +HTTP Header `Authorization: Bearer ` + +客户端或服务端携带 API Key 发起 SDP 交换 + +AOQ + +业务 AppServer 请求网关时 + +HTTP Header `Authorization: Bearer ` + +API Key 仅在服务端使用，客户端使用网关返回的 Token + +## **获取 API Key** + +### **步骤 1：开通百炼服务** + +1. 访问[阿里云百炼控制台](https://bailian.console.aliyun.com/cn-beijing#/home)并登录您的阿里云账号。 + +2. 如果是首次使用，按照页面提示完成服务开通。 + + +### **步骤 2：创建 API Key** + +1. 在控制台左侧导航栏中，选择 **API Key 管理**。 + +2. 点击 **创建 API Key**，选择关联的业务空间。 + +3. 创建完成后，请**立即复制并妥善保存** API Key。 + + +**重要** + +**安全提示**：API Key 是您访问服务的唯一凭证，请勿将其硬编码到客户端代码中或提交到代码仓库。建议通过环境变量或后端服务下发的方式管理。 + +## **建连鉴权详解** + +### **AOQ 协议鉴权** + +AOQ 采用**服务端代理鉴权**模式：API Key 仅在业务 AppServer 侧使用，客户端使用网关返回的临时 Token 建连，避免 API Key 暴露在客户端。 + +![Token鉴权](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3935914871/p1088069.jpg) + +#### **百炼网关请求 curl 示例** + +``` +curl -X POST \ + "https://{endpoint}/api/v1/webrtc/realtime?model=qwen3.5-omni-plus-realtime" \ + -H "Content-Type: application/json" \ + -H "Authorization: Bearer ${DASHSCOPE_API_KEY}" \ + -H "x-dashscope-rtc-transport: moq" \ + -d '{"clientIp": ${客户端真实IP}}' +``` + +#### **请求字段说明** + +**配置项** + +**值** + +**说明** + +endpoint + +根据业务情况选择接入域名 + +指定对应的接入域名，详情请参见[选择地域、服务部署范围和接入域名](https://help.aliyun.com/zh/model-studio/regions/) + +Content-Type + +`application/json` + +\- + +Authorization + +`Bearer ` + +必填 + +x-dashscope-rtc-transport + +`moq` + +**指定使用 AOQ 协议** + +clientIp + +选填。客户端真实公网 IP + +不填写时，使用请求百炼网关的 IP 作为客户端 IP；若填写，则以 clientIp 作为客户端 IP。Realtime API 会参考客户端 IP 提供最佳的 Relay 接入点信息 + +#### **响应示例** + +``` +{ + "sid": "1d06b55683db49bba67a407902f62d02:1782706970:69aecdc5...", + "aoqTokenForClient": "ecc1a46015d5496ca4ff7a48281eb739", + "clientRelayEndpoints": [{"endpoint": "121.199.XX.XX", "port": 8443}], + "clientRelayCertFingerprint": "sha256/99843495...", + "sidExpiresInSecs": 7200, + "extraInfo": {"workspaceIdHash": "2021b6f98cea4cff"} +} +``` + +#### **响应字段说明** + +**字段** + +**说明** + +sid + +会话唯一标识 + +aoqTokenForClient + +客户端连接令牌，传给 SDK 的 token 字段 + +clientRelayEndpoints + +Relay 接入点数组（endpoint + port） + +clientRelayCertFingerprint + +Relay TLS 证书指纹 + +sidExpiresInSecs + +会话过期时间（秒） + +extraInfo.workspaceIdHash + +工作区 ID 哈希 + +#### **AOQ Client SDK 连接示例** + +## **iOS (Swift)** + +``` +let resp = try JSONDecoder().decode(AllocateResponse.self, from: responseData) + +let config = AoqConnectConfig() +config.token = resp.aoqTokenForClient +config.sid = resp.sid +config.certFingerprint = resp.clientRelayCertFingerprint +config.relayEndpoints = resp.clientRelayEndpoints.map { item in + let ep = AoqRelayEndpoint() + ep.endpoint = item.endpoint + ep.port = item.port + return ep +} +config.workspaceIdHash = resp.extraInfo?.workspaceIdHash ?? "" + +let audioTrack = AoqTrackParam() +audioTrack.trackType = .audio +let dataTrack = AoqTrackParam() +dataTrack.trackType = .data +config.publishTracks = [audioTrack, dataTrack] +config.subscribeTracks = [audioTrack, dataTrack] + +engine.connect(config) +``` + +## **Android (Java)** + +``` +JSONObject obj = new JSONObject(responseText); +AoqClientEngine.AoqConnectConfig cfg = new AoqClientEngine.AoqConnectConfig(); +cfg.token = obj.optString("aoqTokenForClient", ""); +cfg.sid = obj.optString("sid", ""); +cfg.certFingerprint = obj.optString("clientRelayCertFingerprint", ""); + +JSONArray arr = obj.optJSONArray("clientRelayEndpoints"); +if (arr != null) { + for (int i = 0; i < arr.length(); i++) { + JSONObject o = arr.optJSONObject(i); + AoqClientEngine.AoqRelayEndpoint ep = new AoqClientEngine.AoqRelayEndpoint(); + ep.endpoint = o.optString("endpoint", ""); + ep.port = o.optInt("port", 0); + cfg.relayEndpoints.add(ep); + } +} + +JSONObject ext = obj.optJSONObject("extraInfo"); +cfg.workspaceIdHash = ext != null ? ext.optString("workspaceIdHash", "") : ""; + +AoqClientEngine.AoqTrackParam audio = new AoqClientEngine.AoqTrackParam(); +audio.trackType = AoqClientEngine.AoqTrackType.AoqTrackTypeAudio; +AoqClientEngine.AoqTrackParam data = new AoqClientEngine.AoqTrackParam(); +data.trackType = AoqClientEngine.AoqTrackType.AoqTrackTypeData; +cfg.publishTracks.add(audio); +cfg.publishTracks.add(data); +cfg.subscribeTracks.add(audio); +cfg.subscribeTracks.add(data); + +engine.connect(cfg); +``` + +## **OHOS (ArkTS)** + +``` +const obj = JSON.parse(responseText) as Record; +const cfg: AoqConnectConfig = { + token: String(obj['aoqTokenForClient'] ?? ''), + sid: String(obj['sid'] ?? ''), + certFingerprint: String(obj['clientRelayCertFingerprint'] ?? ''), + relayEndpoints: (obj['clientRelayEndpoints'] as Array).map(item => ({ + endpoint: String(item['endpoint'] ?? ''), + port: Number(item['port'] ?? 0) + })), + workspaceIdHash: String((obj['extraInfo'] as any)?.['workspaceIdHash'] ?? ''), + publishTracks: [ + { trackType: AoqTrackType.AoqTrackTypeAudio }, + { trackType: AoqTrackType.AoqTrackTypeData } + ], + subscribeTracks: [ + { trackType: AoqTrackType.AoqTrackTypeAudio }, + { trackType: AoqTrackType.AoqTrackTypeData } + ] +}; +engine.connect(cfg); +``` + +**说明** + +`clientIp` 为请求体中的非必填字段。不填写时，使用请求百炼网关的 IP 作为客户端 IP；若填写，则以 clientIp 作为客户端 IP。建议由业务 AppServer 在服务端获取客户端真实 IP 后填入，以获得最佳的 Relay 接入点。 + +## **WebRTC 协议鉴权** + +WebRTC 通过 HTTP POST 请求完成 SDP 交换，鉴权在此阶段完成。客户端将 Offer SDP 发送给服务端，服务端返回 Answer SDP。 + +**配置项** + +**值** + +**说明** + +请求方法 + +POST + +\- + +请求地址 + +`https://{endpoint}/api/v1/webrtc/realtime?model={model_name}` + +替换 endpoint 和 model\_name + +Content-Type + +`application/sdp` + +请求体为 SDP 字符串 + +Authorization + +`Bearer ` + +必填 + +响应 + +HTTP 200，返回 Answer SDP + +失败返回 4xx + +**说明** + +WebRTC 功能目前为白名单开放，请联系商务经理获取 Endpoint。 + +``` +const pc = new RTCPeerConnection(); +const stream = await navigator.mediaDevices.getUserMedia({ audio: true }); +stream.getAudioTracks().forEach(t => pc.addTrack(t, stream)); +pc.createDataChannel('oai-events'); + +const offer = await pc.createOffer(); +await pc.setLocalDescription(offer); + +// 等待 ICE 收集完成后发送 +const resp = await fetch(API_URL, { + method: 'POST', + headers: { + 'Content-Type': 'application/sdp', + 'Authorization': `Bearer ${API_KEY}`, + }, + body: pc.localDescription.sdp, +}); +const answerSdp = await resp.text(); +await pc.setRemoteDescription({ type: 'answer', sdp: answerSdp }); +``` + +## **WebSocket 协议鉴权** + +WebSocket 鉴权最为简单，客户端在建立 WebSocket 连接时直接通过 HTTP Header 携带 API Key。 + +**配置项** + +**值** + +**说明** + +连接地址 + +`wss://dashscope.aliyuncs.com/api-ws/v1/realtime?model={model_name}` + +华北2（北京） + +Authorization + +`Bearer ` + +必填 + +``` +import websocket, os +API_KEY = os.getenv("DASHSCOPE_API_KEY") +URL = "wss://dashscope.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-omni-plus-realtime" +ws = websocket.WebSocketApp(URL, header=["Authorization: Bearer " + API_KEY]) +ws.run_forever() +``` + +**说明** + +您也可以使用 [DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk) 方式接入。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md index 26d8278b..0a419a8e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md @@ -6,7 +6,7 @@ ## **工作流程** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2515653871/CAEQaxiBgIDB5qWk4BkiIDViYzQ0MWUwNTYyNDQ3NDM5NzM0ZTc4N2Y3NTU2NjA56318723_20260129171731.699.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3199194871/CAEQaxiBgIDB5qWk4BkiIDViYzQ0MWUwNTYyNDQ3NDM5NzM0ZTc4N2Y3NTU2NjA56318723_20260129171731.699.svg) ## **前提条件** @@ -18,9 +18,9 @@ - **服务端点** - - **中国内地：**`https://dashscope.aliyuncs.com/compatible-mode/v1` + - **华北2（北京）：**`https://dashscope.aliyuncs.com/compatible-mode/v1` - - **国际：**`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` + - **新加坡：**`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` **重要** @@ -39,7 +39,7 @@ - 千问 Plus：qwen3.7-plus、qwen3.6-plus、qwen3.5-plus、qwen-plus、qwen-plus-latest - - 千问 Flash：qwen3.6-flash、qwen3.5-flash、qwen-flash + - 千问 Flash：qwen3.7-flash、qwen3.7-flash-2026-07-15、qwen3.6-flash、qwen3.5-flash、qwen-flash - 千问 Long：qwen-long、qwen-long-latest @@ -47,7 +47,7 @@ - **多模态模型** - - [图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)：qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、qwen3.5-plus、qwen3.5-flash、qwen3-vl-plus、qwen3-vl-flash + - [图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)：qwen3.7-plus、qwen3.6-plus、qwen3.7-flash、qwen3.7-flash-2026-07-15、qwen3.6-flash、qwen3.5-plus、qwen3.5-flash、qwen3-vl-plus、qwen3-vl-flash - [文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr)：qwen-vl-ocr、qwen-vl-ocr-latest @@ -58,7 +58,7 @@ **重要** -- 在Batch 场景下，`qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`、`qwen3.5-plus`、`qwen3.5-flash`和`qwen3.5-omni-plus`单次请求的上下文 Token 数最大支持 256K，`qwen3.5-omni-plus`不支持语音输出。 +- 在Batch 场景下，`qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.7-flash`、`qwen3.7-flash-2026-07-15`、`qwen3.6-flash`、`qwen3.5-plus`、`qwen3.5-flash`和`qwen3.5-omni-plus`单次请求的上下文 Token 数最大支持 256K，`qwen3.5-omni-plus`不支持语音输出。 - 部分模型支持思考模式，开启后会产生思考`tokens`导致成本增加。 @@ -771,9 +771,9 @@ fi JSONL 批量生成工具 -**请选择模式：** +**请选择地域：** -中国内地 国际 +华北2（北京） 新加坡 **选择模型系列:** 文本生成模型 多模态模型 通用文本向量模型 @@ -785,9 +785,9 @@ fi 生成 -**请选择模式：** +**请选择地域：** -中国内地 国际 +华北2（北京） 新加坡 **选择模型系列:** 文本生成模型 @@ -878,7 +878,7 @@ client = OpenAI( # 若没有配置环境变量，可用阿里云百炼API Key将下行替换为：api_key="sk-xxx"。但不建议在生产环境中直接将API Key硬编码到代码中，以减少API Key泄露风险。 # 新加坡和北京地域的API Key不同。 api_key=os.getenv("DASHSCOPE_API_KEY"), - # 以下是北京地域base_url，如果使用新加坡地域的模型，需要将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 + # 以下是北京地域base_url，如果使用新加坡地域的模型，需要将base_url替换为：https://dashscope-intl.aliyuncs.com/compatible-mode/v1 # 注意：切换地域时，API Key也需要对应更换 base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", ) @@ -909,7 +909,7 @@ const fs = require('fs'); // 北京地域配置（默认） const BASE_URL = 'https://dashscope.aliyuncs.com/compatible-mode/v1'; // 如果使用新加坡地域，请将上面的 BASE_URL 替换为： -// const BASE_URL = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1'; +// const BASE_URL = 'https://dashscope-intl.aliyuncs.com/compatible-mode/v1'; // 注意：切换地域时，API Key也需要对应更换 const apiKey = process.env.DASHSCOPE_API_KEY; @@ -955,7 +955,7 @@ import java.util.regex.Matcher; * * 地域配置： * - 北京地域：https://dashscope.aliyuncs.com/compatible-mode/v1 - * - 新加坡地域：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 + * - 新加坡地域：https://dashscope-intl.aliyuncs.com/compatible-mode/v1 * 注意：切换地域时，API Key也需要对应更换 */ public class BatchAPIUploadFile { @@ -963,7 +963,7 @@ public class BatchAPIUploadFile { // 北京地域配置（默认） private static final String BASE_URL = "https://dashscope.aliyuncs.com/compatible-mode/v1"; // 如果使用新加坡地域，请将上面的 BASE_URL 替换为： - // private static final String BASE_URL = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1"; + // private static final String BASE_URL = "https://dashscope-intl.aliyuncs.com/compatible-mode/v1"; // 注意：切换地域时，API Key也需要对应更换 private static String API_KEY; @@ -1035,7 +1035,7 @@ String fileId = uploadFile("test.jsonl"); ``` # ======= 重要提示 ======= # 新加坡和北京地域的API Key不同。 -# 以下是北京地域base_url，如果使用新加坡地域的模型，需要将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/files +# 以下是北京地域base_url，如果使用新加坡地域的模型，需要将base_url替换为：https://dashscope-intl.aliyuncs.com/compatible-mode/v1/files # === 执行时请删除该注释 === curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/files \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -2469,7 +2469,7 @@ System.out.println(content); ``` # ======= 重要提示 ======= # 新加坡和北京地域的API Key不同。 -# 以下是北京地域base_url，如果使用新加坡地域的模型，需要将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/files/file-batch_output-xxx/content +# 以下是北京地域base_url，如果使用新加坡地域的模型，需要将base_url替换为：https://dashscope-intl.aliyuncs.com/compatible-mode/v1/files/file-batch_output-xxx/content # === 执行时请删除该注释 === curl -X GET https://dashscope.aliyuncs.com/compatible-mode/v1/files/file-batch_output-xxx/content \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" > result.jsonl diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md index 53fb9ef1..823f8b6c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md @@ -19,7 +19,7 @@ ## 支持的模型 -`qwen3-max`、`qwen3-max-2026-01-23`、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.6-plus-2026-04-02`、`qwen3.5-plus`、`qwen3.5-plus-2026-02-15`、`qwen3.5-plus-2026-04-20`、`qwen3.6-flash`、`qwen3.6-flash-2026-04-16`、`qwen3.5-flash`、`qwen3.5-flash-2026-02-23`、`qwen3.6-35b-a3b`、`qwen3.5-397b-a17b`、`qwen3.5-122b-a10b`、`qwen3.5-27b`、`qwen3.5-35b-a3b`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash`、`qwen3-coder-next`。 +`qwen3-max`、`qwen3-max-2026-01-23`、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.6-plus-2026-04-02`、`qwen3.5-plus`、`qwen3.5-plus-2026-02-15`、`qwen3.5-plus-2026-04-20`、`qwen3.7-flash`、`qwen3.7-flash-2026-07-15`、`qwen3.6-flash`、`qwen3.6-flash-2026-04-16`、`qwen3.5-flash`、`qwen3.5-flash-2026-02-23`、`qwen3.6-35b-a3b`、`qwen3.5-397b-a17b`、`qwen3.5-122b-a10b`、`qwen3.5-27b`、`qwen3.5-35b-a3b`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash`、`qwen3-coder-next`。 ## 服务地址 @@ -771,7 +771,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/ **使用方式**：在请求 Header 中添加 `x-dashscope-session-cache: enable` 开启，或设置为 `disable` 关闭。默认值为 `disable`。 -**支持的模型：**`qwen3-max`、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.5-plus`、`qwen3.6-flash`、`qwen3.5-flash`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash` +**支持的模型：**`qwen3-max`、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.5-plus`、`qwen3.7-flash`、`qwen3.6-flash`、`qwen3.5-flash`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash` > Session 缓存 最小可缓存提示词长度为 1024 Token，缓存有效期为 5 分钟。相关约束限制与[显式缓存](https://help.aliyun.com/zh/model-studio/context-cache)一致。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/completions.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/completions.md index a88b645e..849e2704 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/completions.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/completions.md @@ -4,7 +4,7 @@ Completions 接口专为文本补全场景设计，适合代码补全、内容 **说明** -本文档仅适用于中国内地（北京地域），需使用中国（北京）地域的[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)。 +本文档仅适用于华北2（北京）地域，需使用华北2（北京）地域的[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)。 ## **支持的模型** @@ -46,7 +46,7 @@ import os from openai import OpenAI client = OpenAI( - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", api_key=os.getenv("DASHSCOPE_API_KEY") ) @@ -67,7 +67,7 @@ const openai = new OpenAI( { // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey: "sk-xxx", apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" + baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1" } ); @@ -85,7 +85,7 @@ main(); curl ``` -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/completions \ +curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -113,7 +113,7 @@ import os from openai import OpenAI client = OpenAI( - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", api_key=os.getenv("DASHSCOPE_API_KEY") ) @@ -145,7 +145,7 @@ Node.js import OpenAI from 'openai'; const client = new OpenAI({ - baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1", apiKey: process.env.DASHSCOPE_API_KEY }); @@ -178,7 +178,7 @@ main(); curl ``` -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/completions \ +curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md index 3b3cff7e..0cdb9d23 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md @@ -19,14 +19,14 @@ ### **华北2（北京）** -- **文本生成模型：**qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、qwen3.5-plus、qwen3.5-flash、qwen3-max、qwen-plus、qwen-flash、deepseek-v3.2 +- **文本生成模型：**qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.7-flash、qwen3.7-flash-2026-07-15、qwen3.6-flash、qwen3.5-plus、qwen3.5-flash、qwen3-max、qwen-plus、qwen-flash、deepseek-v3.2 -- **图像与视频理解模型：**qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、qwen3.5-plus、qwen3.5-flash、qwen3.5-omni-plus、qwen3-vl-plus、qwen3-vl-flash +- **图像与视频理解模型：**qwen3.7-plus、qwen3.6-plus、qwen3.7-flash、qwen3.7-flash-2026-07-15、qwen3.6-flash、qwen3.5-plus、qwen3.5-flash、qwen3.5-omni-plus、qwen3-vl-plus、qwen3-vl-flash **重要** -- 在Batch 场景下，`qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`、`qwen3.5-plus`、`qwen3.5-flash`和`qwen3.5-omni-plus`单次请求的上下文 Token 数最大支持 256K，`qwen3.5-omni-plus`不支持语音输出。 +- 在Batch 场景下，`qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.7-flash`、`qwen3.7-flash-2026-07-15`、`qwen3.6-flash`、`qwen3.5-plus`、`qwen3.5-flash`和`qwen3.5-omni-plus`单次请求的上下文 Token 数最大支持 256K，`qwen3.5-omni-plus`不支持语音输出。 - 部分模型支持思考模式，开启后会产生思考`tokens`导致成本增加。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md index b7487808..37eae351 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md @@ -16,14 +16,14 @@ **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 **items** `_array_`（可选） @@ -948,3 +948,5 @@ console.log("第二轮响应:", response2.output_text); - 创建会话或添加消息项时，`items` 最多包含20条。 - `metadata` 最多16对键值对，key最大长度64字符，value最大长度512字符。 + +- 会话信息最多保存7天。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md index d94c7a68..480aa12c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md @@ -64,9 +64,7 @@ text-embedding-async-v1 通用文本向量批处理接口API支持通过HTTP和DashScope SDK进行调用。 -在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 - -如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。目前，该SDK已支持Python和Java。 +在调用前，先[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 ## HTTP调用 @@ -79,14 +77,16 @@ HTTP调用仅支持异步模式，需通过两步完成： **通过HTTP调用时需配置的endpoint：** -`POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ### 创建任务 ##### **请求参数** ``` -curl -X POST 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ +curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ -H 'X-DashScope-Async: enable' \ @@ -232,7 +232,7 @@ url `_string_` **（必选）** ### **根据任务ID查询结果** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` ##### **请求参数** @@ -240,10 +240,10 @@ url `_string_` **（必选）** 请将`86ecf553-d340-4e21-xxxxxxxxx`替换为真实的task\_id。 -> 若使用新加坡地域的模型，需将base\_url替换为https://dashscope-intl.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx +> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中{WorkspaceId}需替换为真实的业务空间ID。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -398,6 +398,9 @@ SDK与HTTP接口的参数名基本一致，参数结构根据不同语言的SDK ``` from dashscope import BatchTextEmbedding +import dashscope +# 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" result = BatchTextEmbedding.call(BatchTextEmbedding.Models.text_embedding_async_v1, url="https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241016/nigwvr/text_embedding_file.txt", @@ -416,8 +419,11 @@ import com.alibaba.dashscope.exception.NoApiKeyException; import com.alibaba.dashscope.task.AsyncTaskListParam; import com.alibaba.dashscope.task.AsyncTaskListResult; import com.alibaba.dashscope.utils.JsonUtils; +import com.alibaba.dashscope.utils.Constants; public class Main { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; public static void basicCall() throws ApiException, NoApiKeyException { BatchTextEmbeddingParam param = BatchTextEmbeddingParam.builder() .model(BatchTextEmbedding.Models.TEXT_EMBEDDING_ASYNC_V1) @@ -430,6 +436,8 @@ public class Main { } public static void main(String[] args) { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; try { basicCall(); } catch (ApiException | NoApiKeyException e) { @@ -446,7 +454,10 @@ public class Main { ``` from dashscope import BatchTextEmbedding +import dashscope from http import HTTPStatus +# 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" # 创建异步任务 def create_async_task(): @@ -508,8 +519,11 @@ import com.alibaba.dashscope.exception.NoApiKeyException; import com.alibaba.dashscope.task.AsyncTaskListParam; import com.alibaba.dashscope.task.AsyncTaskListResult; import com.alibaba.dashscope.utils.JsonUtils; +import com.alibaba.dashscope.utils.Constants; public class Main { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; /**创建批处理任务*/ public static BatchTextEmbeddingResult createTask() throws ApiException, NoApiKeyException { @@ -738,7 +752,7 @@ url `_string_` **（必选）** 任务状态 -- SUCCESSED: 任务执行成功 +- SUCCEEDED: 任务执行成功 - FAILED: 任务执行失败 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md index f65767c7..0d7a583d 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md @@ -68,7 +68,7 @@ qwen3-vl-embedding 100万Token -有效期：百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天（以较晚者为准） qwen2.5-vl-embedding @@ -178,16 +178,24 @@ qwen3-vl-embedding 支持中、英、日、韩、法、德等33种主流语言 +**所有支持语言** + +中文、日语、韩语、印尼语、越南语、泰语、英语、法语、德语、俄语、葡萄牙语、西班牙语、意大利语、瑞典语、丹麦语、捷克语、挪威语、荷兰语、芬兰语、土耳其语、波兰语、斯瓦希里语、罗马尼亚语、塞尔维亚语、希腊语、哈萨克语、乌兹别克语、宿务语、阿拉伯语、乌尔都语、波斯语、印地语 / 天城语、希伯来语。 + JPEG, PNG, WEBP, BMP, TIFF, ICO, DIB, ICNS, SGI（支持URL或Base64） MP4, AVI, MOV（仅支持URL） -一次请求中传入内容元素总数不超过 20。图片数量不超过5，视频数量不超过1。 +一次请求中传入内容元素总数不超过 20。图片数量不超过10，视频数量不超过1。 qwen2.5-vl-embedding 支持中、英、日、韩、法、德等11种主流语言 +**所有支持语言** + +中文、英语、日语、韩语、法语、德语、俄语、葡萄牙语、西班牙语、意大利语、印尼语 + 一次请求内，图片、文本、视频、融合对象每种类型最多出现 1 次。 **多模态向量模型** @@ -206,6 +214,10 @@ tongyi-embedding-vision-plus-2026-03-06 支持中、英、日、韩等超30种主流语言 +**所有支持语言** + +中文、日语、韩语、印尼语、越南语、泰语、英语、法语、德语、俄语、葡萄牙语、西班牙语、意大利语、瑞典语、丹麦语、捷克语、挪威语、荷兰语、芬兰语、土耳其语、波兰语、斯瓦希里语、罗马尼亚语、塞尔维亚语、希腊语、哈萨克语、乌兹别克语、宿务语、阿拉伯语、乌尔都语、波斯语、印地语 / 天城语、希伯来语。 + JPEG, PNG, WEBP, BMP, TIFF, ICO, DIB, ICNS, SGI（支持URL或Base64） MP4, MPEG, MOV, MPG, WEBM, AVI, FLV, MKV（仅支持URL） @@ -542,7 +554,7 @@ contents `_array_`**（必选）** ] }, "usage": { - "input_tokens": 10, + "input_tokens": 903, "input_tokens_details": { "image_tokens": 896, "text_tokens": 7 @@ -644,11 +656,11 @@ contents `_array_`**（必选）** **image\_tokens** `_int_` -输入的图片或视频的 Token 数量。 +输入内容中图片或视频等**视觉部分**消耗的 Token 数量，**不包含文本**（文本部分见 `text_tokens`）。图片消耗的 Token 数量与输入图片的分辨率有关，分辨率越高消耗的 Token 越多；若输入为视频，系统会先对视频抽帧，再基于抽帧结果计算 Token。 **text\_tokens** `_int_` -输入的文本的 Token 数量。 +输入内容中**文本部分**消耗的 Token 数量（不包含图片或视频等视觉部分）。 **output\_tokens** `_int_` @@ -660,7 +672,7 @@ contents `_array_`**（必选）** **image\_tokens** `_int_` -本次请求输入的图片或视频的 Token 数量。系统会对输入视频进行抽帧处理，帧数上限受系统配置控制，随后基于处理结果计算 Token。仅 `qwen3-vl-embedding`、`qwen2.5-vl-embedding` 和 `multimodal-embedding-v1` 返回此字段（作为顶层字段），`tongyi-embedding-vision-*` 系列模型的图片 Token 包含在 `input_tokens_details.image_tokens` 中。 +本次请求输入的图片或视频等**视觉部分**消耗的 Token 数量（**不包含文本**）。图片消耗的 Token 数量与输入图片的分辨率有关；系统会对输入视频进行抽帧处理，帧数上限受系统配置控制，随后基于处理结果计算 Token。仅 `qwen3-vl-embedding`、`qwen2.5-vl-embedding` 和 `multimodal-embedding-v1` 返回此字段（作为顶层字段），`tongyi-embedding-vision-*` 系列模型的图片 Token 包含在 `input_tokens_details.image_tokens` 中。 **image\_count** `_int_` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md index 79c7fb66..07f2c9b9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md @@ -4,13 +4,13 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 ## 适用范围 -为确保调用成功，请务必保证模型、endpoint URL 和 API Key 均属于**同一地域**。跨地域调用将会失败。 +为确保调用成功，请务必保证模型、endpoint URL和API Key 均属于**同一地域**。跨地域调用将会失败。 - [**选择模型**](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)：确认模型所属的地域。 - **选择 URL**：选择对应的地域 Endpoint URL，支持HTTP URL。 -- **配置 API Key**：获取该地域的[API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 +- **配置API Key**：获取该地域的[API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 **说明** @@ -19,14 +19,14 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -38,14 +38,10 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -54,7 +50,11 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +## **日本（东京）** + +`POST https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -68,6 +68,7 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 ## 图生视频-基于首帧 ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -169,7 +170,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi 1. 公网URL： - - 支持 HTTP 或 HTTPS 协议。 + - 支持HTTP或HTTPS协议。 - 示例值：https://xxx/xxx.png。 @@ -326,14 +327,10 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -342,7 +339,9 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +## **日本（东京）** + +`GET https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -363,7 +362,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md index b93ad55b..57e4536b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md @@ -19,14 +19,14 @@ HappyHorse-参考生视频模型支持传入**多张参考图像**，通过**文 **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -38,14 +38,10 @@ HappyHorse-参考生视频模型支持传入**多张参考图像**，通过**文 `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -54,7 +50,11 @@ HappyHorse-参考生视频模型支持传入**多张参考图像**，通过**文 `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +## **日本（东京）** + +`POST https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -68,6 +68,7 @@ HappyHorse-参考生视频模型支持传入**多张参考图像**，通过**文 ## 参考生视频（多图像） ``` +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -368,14 +369,10 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -384,7 +381,9 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +## **日本（东京）** + +`GET https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -403,7 +402,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md index 385ff8f8..373f79e0 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md @@ -19,14 +19,14 @@ HappyHorse文生视频模型，输入文本提示词生成物理真实、运动 **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -38,14 +38,10 @@ HappyHorse文生视频模型，输入文本提示词生成物理真实、运动 `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -54,7 +50,11 @@ HappyHorse文生视频模型，输入文本提示词生成物理真实、运动 `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +## **日本（东京）** + +`POST https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -68,6 +68,7 @@ HappyHorse文生视频模型，输入文本提示词生成物理真实、运动 ## 文生视频 ``` +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -266,14 +267,10 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -282,7 +279,9 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +## **日本（东京）** + +`GET https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -303,7 +302,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md index edfc94c8..58de1d49 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md @@ -19,14 +19,14 @@ HappyHorse 视频编辑模型支持输入视频与参考图，结合文本指令 **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -38,14 +38,10 @@ HappyHorse 视频编辑模型支持输入视频与参考图，结合文本指令 `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -54,7 +50,11 @@ HappyHorse 视频编辑模型支持输入视频与参考图，结合文本指令 `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +## **日本（东京）** + +`POST https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -68,6 +68,7 @@ HappyHorse 视频编辑模型支持输入视频与参考图，结合文本指令 ## 视频编辑（指令+参考图） ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -369,14 +370,10 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -385,7 +382,9 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +## **日本（东京）** + +`GET https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -404,7 +403,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ @@ -561,7 +560,7 @@ task\_id查询有效期为 24 小时，超时后将无法查询，返回以下 **属性** -**duration** `_float_` +**duration** `_float_` 生成视频的总视频时长，用于计费。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md index 4a7b656c..f2b1777b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md @@ -27,7 +27,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -43,7 +45,8 @@ 支持模型：`kling/kling-v3-omni-video-generation`、`kling/kling-v3-video-generation`。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -67,7 +70,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：`kling/kling-v3-omni-video-generation`、`kling/kling-v3-video-generation` 。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -107,7 +111,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：`kling/kling-v3-omni-video-generation`、`kling/kling-v3-video-generation` 。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -136,7 +141,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：`kling/kling-v3-omni-video-generation`、`kling/kling-v3-video-generation` 。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -169,7 +175,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：`kling/kling-v3-omni-video-generation`。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -640,7 +647,7 @@ audio直接影响费用，请前往[百炼控制台](https://bailian.console.ali ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -659,7 +666,7 @@ audio直接影响费用，请前往[百炼控制台](https://bailian.console.ali ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md index 7fea4312..7e98f28e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md @@ -17,20 +17,26 @@ 为确保调用成功，请务必保证模型、endpoint URL 和 API Key 均属于**同一地域**。跨地域调用将会失败。 -- [**选择模型**](https://help.aliyun.com/zh/model-studio/use-video-generation#183f1b6fa0lox)：确认模型所属的地域。 +- [**选择模型**](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)：确认模型所属的地域。 - **选择 URL**：选择对应的地域 Endpoint URL，支持HTTP URL或 DashScope SDK URL。 - **配置 API Key**：获取该地域的[API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 +**说明** + +本文的示例代码适用于**华北2（北京）地域**。 + ## HTTP调用 图生视频任务耗时较长（通常为1-5分钟），API采用异步调用的方式。整个流程包含 **"创建任务 -> 轮询获取"** 两个核心步骤，具体如下： ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -46,7 +52,8 @@ 支持模型：pixverse/pixverse-c1-it2v、pixverse/pixverse-v6-it2v、pixverse/pixverse-v5.6-it2v。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -79,7 +86,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在`prompt`中描述多镜头场景即可，不支持设置 `shot_type`参数。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -110,7 +118,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在`prompt`中描述多镜头场景，并设置 `shot_type` 为`multi`， 即可生成有声多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -155,7 +164,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener **model** `_string_` **（必选）** -模型名称。模型输出规格请参见[模型列表](https://help.aliyun.com/zh/model-studio/use-video-generation#183f1b6fa0lox)。 +模型名称。 可选值： @@ -383,7 +392,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -402,7 +411,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md index 99e60eb8..557607a9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md @@ -17,20 +17,26 @@ 为确保调用成功，请务必保证**模型、Endpoint URL 和 API Key 均属于同一地域**。跨地域调用将会失败。 -- [**选择模型**](https://help.aliyun.com/zh/model-studio/use-video-generation#f32f686472enw)：确认模型所属的地域。 +- [**选择模型**](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)：确认模型所属的地域。 - **选择 URL**：选择对应的地域 Endpoint URL，支持HTTP URL或 DashScope SDK URL。 - **配置 API Key**：选择地域并[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 +**说明** + +本文的示例代码适用于**华北2（北京）地域**。 + ## HTTP调用 由于首尾帧生视频任务耗时较长（通常为1-5分钟），API采用异步调用。整个流程包含 **"创建任务 -> 轮询获取"** 两个核心步骤，具体如下： ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -46,7 +52,8 @@ 支持模型：pixverse/pixverse-c1-kf2v、pixverse/pixverse-v6-kf2v、pixverse/pixverse-v5.6-kf2v。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -95,7 +102,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener **model** `_string_` **（必选）** -模型名称。模型输出规格请参见[模型列表](https://help.aliyun.com/zh/model-studio/use-video-generation#f32f686472enw)。 +模型名称。 可选值： @@ -310,7 +317,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -329,7 +336,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md index 7a9a606c..61775d7e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md @@ -34,7 +34,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -50,7 +52,8 @@ 传入视频和音频文件，生成对口型视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -79,7 +82,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 传入视频和 TTS 文本，由模型合成语音并生成对口型视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -273,7 +277,7 @@ auto ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md index f6766e71..44892823 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md @@ -34,7 +34,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -50,7 +52,8 @@ 传入角色图片和动作视频，生成角色模仿动作的视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -162,7 +165,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md index c90d87c5..1e14eac3 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md @@ -30,7 +30,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -46,7 +48,8 @@ 通过 `media` 传入参考图像，设置 `size` 和 `duration` 控制视频分辨率和时长。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -83,7 +86,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在 `media` 传入参考图像，并传入`ref_name`用于设置参考图像的主体名称。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -759,7 +763,7 @@ audio直接影响费用，请在调用前确认[爱诗-参考生视频](https:// ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -778,7 +782,7 @@ audio直接影响费用，请在调用前确认[爱诗-参考生视频](https:// ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md index 4775d1a3..320e8241 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md @@ -17,7 +17,7 @@ 为确保调用成功，请务必保证**模型、Endpoint URL 和 API Key 均属于同一地域**。跨地域调用将会失败。 -- [**选择模型**](https://help.aliyun.com/zh/model-studio/use-video-generation#3ad2d09509ldb)：确认模型所属的地域。 +- [**选择模型**](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)：确认模型所属的地域。 - **选择 URL**：选择对应的地域 Endpoint URL，支持HTTP URL或 DashScope SDK URL。 @@ -26,13 +26,19 @@ - **安装 SDK**：如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 +**说明** + +本文的示例代码适用于**华北2（北京）地域**。 + ## HTTP调用 由于文生视频任务耗时较长（通常为1-5分钟），API采用异步调用。整个流程包含 **"创建任务 -> 轮询获取"** 两个核心步骤，具体如下： ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -48,7 +54,8 @@ 支持模型：pixverse/pixverse-c1-t2v、pixverse/pixverse-v6-t2v、pixverse/pixverse-v5.6-t2v。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -74,7 +81,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在`prompt`中描述多镜头场景即可，不支持设置 `shot_type`参数。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -99,7 +107,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在`prompt`中描述多镜头场景，并设置 `shot_type` 为`multi`， 即可生成有声多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -140,7 +149,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener **model** `_string_` **（必选）** -模型名称。模型输出规格请参见[模型列表](https://help.aliyun.com/zh/model-studio/use-video-generation#3ad2d09509ldb)。 +模型名称。 可选值： @@ -574,7 +583,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -593,7 +602,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ @@ -798,7 +807,9 @@ SDK 的参数命名与[HTTP接口](#pv101a0h2http)基本一致，参数结构根 若版本过低，可能会触发 "url error, please check url!" 等错误。请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)进行更新。 -**北京地域**：`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +**北京地域**：`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -811,7 +822,7 @@ import dashscope import os # 以下为北京地域URL -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -881,7 +892,7 @@ import dashscope import os # 以下为北京地域URL -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -984,13 +995,16 @@ if __name__ == '__main__': 若版本过低，可能会触发 “url error, please check url!” 等错误。请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)进行更新。 -**北京地域**：`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +**北京地域**：`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 ##### 请求示例 ``` +// 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 // Copyright (c) Alibaba, Inc. and its affiliates. import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis; @@ -1006,7 +1020,7 @@ public class Text2Video { static { // 以下为北京地域url - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" @@ -1080,6 +1094,7 @@ public class Text2Video { ##### 请求示例 ``` +// 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 // Copyright (c) Alibaba, Inc. and its affiliates. import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis; @@ -1097,7 +1112,7 @@ public class Text2Video { static { // 以下为北京地域url - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md index 43f148d0..ac899560 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md @@ -34,7 +34,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -50,7 +52,8 @@ 传入视频文件，输出超分辨率视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -159,7 +162,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md index c12815fa..fcb9c26c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md @@ -27,7 +27,9 @@ Vidu-图生视频模型根据**输入图像**和**文本提示词**，生成一 ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -40,8 +42,10 @@ Vidu-图生视频模型根据**输入图像**和**文本提示词**，生成一 ## 图生视频 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -299,7 +303,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -318,7 +322,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md index 5547a29f..425a4bc4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md @@ -27,7 +27,9 @@ Vidu-首尾帧生视频模型基于**首帧图像**、**尾帧图像和文本提 ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -40,8 +42,10 @@ Vidu-首尾帧生视频模型基于**首帧图像**、**尾帧图像和文本提 ## 首尾帧生视频 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -292,7 +296,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -311,7 +315,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md index 833a58c2..84700aa6 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md @@ -27,7 +27,9 @@ Vidu-参考生视频模型支持传入**参考图片**和**文本提示词**， ### **步骤 1：创建任务获取任务 ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -42,8 +44,10 @@ Vidu-参考生视频模型支持传入**参考图片**和**文本提示词**， 支持模型：vidu/viduq3-ad\_reference2video。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -79,8 +83,10 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：vidu/viduq3-drama\_reference2video。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -118,8 +124,10 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 参考生视频（仅参考图像） +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -155,8 +163,10 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：vidu/viduq2-pro\_reference2video。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -595,7 +605,7 @@ vidu/viduq3-drama\_reference2video 不支持该参数，默认输出有声视频 ### **步骤 2：根据任务 ID 查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -614,7 +624,7 @@ vidu/viduq3-drama\_reference2video 不支持该参数，默认输出有声视频 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md index bde2d5df..1254f4e7 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md @@ -29,7 +29,9 @@ Vidu-文生视频模型基于**文本提示词**，生成一段流畅的视频 ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -42,8 +44,10 @@ Vidu-文生视频模型基于**文本提示词**，生成一段流畅的视频 ## 文生视频 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -322,7 +326,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -341,7 +345,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ @@ -546,7 +550,7 @@ SDK 的参数命名与[HTTP接口](#vd101a0h2http)基本一致，参数结构根 请确保 DashScope Python SDK 版本**不低于** `**1.25.8**`，再运行以下代码。 -**北京地域**：`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +**北京地域**：`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` ## 同步调用 @@ -558,8 +562,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为北京地域URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" api_key = os.getenv("DASHSCOPE_API_KEY") @@ -628,8 +632,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为北京地域URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" api_key = os.getenv("DASHSCOPE_API_KEY") @@ -728,7 +732,7 @@ if __name__ == '__main__': 请确保 DashScope Java SDK 版本**不低于** `**2.22.6**`，再运行以下代码。 -**北京地域**：`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +**北京地域**：`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` ## 同步调用 @@ -749,8 +753,8 @@ import com.alibaba.dashscope.utils.Constants; public class Text2Video { static { - // 以下为北京地域url - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为北京地域URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" @@ -836,8 +840,8 @@ import com.alibaba.dashscope.utils.Constants; public class Text2Video { static { - // 以下为北京地域url - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为北京地域URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md index 08695f0a..083f04cc 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md @@ -29,9 +29,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -45,13 +50,13 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -67,7 +72,8 @@ 基于首帧图像和音频生成视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -101,7 +107,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 传入首帧和尾帧生成视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -134,7 +141,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 基于首段视频片段，让模型生成后续内容。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -526,13 +534,13 @@ duration直接影响费用，按秒计费，请在调用前确认[模型价格]( ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -553,10 +561,10 @@ duration直接影响费用，按秒计费，请在调用前确认[模型价格]( ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -754,13 +762,13 @@ SDK 的参数命名与[HTTP接口](#9c71bffa84zm6)基本一致，参数结构根 ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -773,8 +781,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-general-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -854,8 +862,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-general-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -973,13 +981,13 @@ if __name__ == '__main__': ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"` ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -1005,29 +1013,12 @@ import java.util.List; public class Image2Video { static { - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis; -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisParam; -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisResult; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.InputRequiredException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.utils.Constants; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.util.ArrayList; -import java.util.List; - -public class Image2Video { - - static { - // 以下为新加坡地域URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" - // 各地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/en/model-studio/get-api-key + // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key static String apiKey = System.getenv("DASHSCOPE_API_KEY"); public static void syncCall() { @@ -1093,29 +1084,12 @@ import java.util.List; public class Image2Video { static { - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis; -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisParam; -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisResult; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.InputRequiredException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.utils.Constants; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.util.ArrayList; -import java.util.List; - -public class Image2Video { - - static { - // 以下为新加坡地域URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" - // 各地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/en/model-studio/get-api-key + // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key static String apiKey = System.getenv("DASHSCOPE_API_KEY"); public static void asyncCall() { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md index bf2dd8df..352341c9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md @@ -29,9 +29,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -41,14 +46,12 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **弗吉尼亚** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -57,7 +60,7 @@ `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -75,7 +78,9 @@ 可通过设置`"prompt_extend": true`和`"shot_type":"multi"`启用。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -102,7 +107,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 若不提供 `input.audio_url` ，模型将根据视频内容自动生成匹配的背景音乐或音效。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -127,7 +134,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 如需为视频指定背景音乐或配音，可通过 `input.audio_url` 参数传入自定义音频的 URL。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -156,7 +165,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -182,7 +193,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 示例：下载[img\_base64](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250722/pmcjis/img_base64.txt)文件，并将完整内容粘贴至`img_url`参数中。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -203,11 +216,13 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener - prompt 字段将被忽略，建议留空。 -- 特效的可用性与模型相关。调用前请查阅[万相-图生视频-视频特效](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 +- 特效的可用性与模型相关。调用前请查阅[万相-图生视频-视频特效列表](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -228,7 +243,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 通过 negative\_prompt 指定生成的视频避免出现“花朵”元素。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -322,7 +339,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持输入的格式： -1. 公网URL: +1. 公网URL： - 支持 HTTP 或 HTTPS 协议。 @@ -381,7 +398,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 视频特效模板的名称。若未填写，表示不使用任何视频特效。 -不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 +不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效列表](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 示例值：flying，表示使用“魔法悬浮”特效。 @@ -596,14 +613,12 @@ audio直接影响费用，有声视频与无声视频价格不同，请前往百 ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **弗吉尼亚** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -612,7 +627,7 @@ audio直接影响费用，有声视频与无声视频价格不同，请前往百 `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -633,10 +648,10 @@ audio直接影响费用，有声视频与无声视频价格不同，请前往百 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -881,18 +896,16 @@ SDK 的参数命名与[HTTP接口](#42703589880ts)基本一致，参数结构根 若版本过低，可能会触发 “url error, please check url!” 等错误。请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)进行更新。 -根据模型所在地域设置 `**base_http_api_url**`: +根据模型所在地域设置 `**base_http_api_url**`： ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **弗吉尼亚** `dashscope.base_http_api_url = 'https://dashscope-us.aliyuncs.com/api/v1'` @@ -901,7 +914,7 @@ SDK 的参数命名与[HTTP接口](#42703589880ts)基本一致，参数结构根 `dashscope.base_http_api_url = 'https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### **示例代码** @@ -919,8 +932,9 @@ from dashscope import VideoSynthesis import mimetypes import dashscope -# 以下为北京地域url，获取url：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +# 获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -1032,8 +1046,9 @@ from http import HTTPStatus from dashscope import VideoSynthesis import dashscope -# 以下为北京地域url，获取url：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +# 获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -1145,18 +1160,16 @@ if __name__ == '__main__': 若版本过低，可能会触发 “url error, please check url!” 等错误。请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)进行更新。 -根据模型所在地域设置 `**baseHttpApiUrl**`: +根据模型所在地域设置 `**baseHttpApiUrl**`： ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **弗吉尼亚** `Constants.baseHttpApiUrl = "https://dashscope-us.aliyuncs.com/api/v1";` @@ -1165,7 +1178,7 @@ if __name__ == '__main__': `Constants.baseHttpApiUrl = "https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### **示例代码** @@ -1198,8 +1211,9 @@ import java.util.Map; public class Image2Video { static { - // 以下为北京地域url，获取url：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + // 获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" @@ -1351,8 +1365,9 @@ import java.util.Map; public class Image2Video { static { - // 以下为北京地域url，获取url：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + // 获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md index 7dc1d88e..dd4f30e4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md @@ -23,9 +23,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -35,13 +40,15 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -57,7 +64,8 @@ 根据首帧、尾帧和prompt生成视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -82,7 +90,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video 格式参见[如何输入图像](https://help.aliyun.com/zh/model-studio/image-to-video-first-and-last-frames-guide#69308edfbebnx)。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -104,10 +113,11 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video 必须传入`first_frame_url`和`template`，无需传入prompt和last\_frame\_url。 -不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 +不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效列表](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -129,7 +139,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video 通过 negative\_prompt 指定生成的视频避免出现“人物”元素。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -274,7 +285,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video 视频特效模板的名称。使用此参数时，仅需传入 `first_frame_url`。 -不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 +不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效列表](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 示例值：hufu-1，表示使用“唐韵翩然”特效。 @@ -412,13 +423,15 @@ duration直接影响费用，按秒计费，调用前请确认百炼控制台。 ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -441,10 +454,10 @@ duration直接影响费用，按秒计费，调用前请确认百炼控制台。 请将`86ecf553-d340-4e21-xxxxxxxxx`替换为真实的task\_id。 -> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中WorkspaceId需替换为真实的业务空间ID。 +> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中{WorkspaceId}需替换为真实的业务空间ID。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -641,13 +654,15 @@ SDK 的参数命名与[HTTP接口](https://help.aliyun.com/zh/model-studio/text- ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### **示例代码** @@ -666,7 +681,7 @@ import mimetypes import dashscope # 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -768,7 +783,7 @@ from http import HTTPStatus from dashscope import VideoSynthesis import dashscope -# 以下为华北2（北京）地域的URL，各地域的URL不同。 +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 ``` ### Java SDK调用 @@ -816,7 +831,7 @@ import java.util.Map; public class Kf2vSync { static { - // 以下为华北2（北京）地域的URL，各地域的URL不同。 + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 ``` ## 异步调用 @@ -851,7 +866,7 @@ import java.util.Map; public class Kf2vAsync { static { - // 以下为华北2（北京）地域的URL，各地域的URL不同。 + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 ``` ## **使用限制** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md index 8477b7b4..b5562bbf 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md @@ -21,9 +21,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -37,13 +42,15 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -53,7 +60,7 @@ `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -69,7 +76,8 @@ 通过`reference_urls`传入图像和视频URL。同时设置`shot_type`为`multi`，生成多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -99,7 +107,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 通过`reference_urls`传入多个视频URL。同时设置`shot_type`为`multi`，生成多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -125,7 +134,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 通过`reference_urls`传入单个视频URL。同时设置`shot_type`为`multi`，生成多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -150,7 +160,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 当生成无声视频时，**必须显式设置** `parameters.audio = false`。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -505,13 +516,15 @@ audio直接影响费用，有声视频与无声视频价格不同，请在调用 ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -521,7 +534,7 @@ audio直接影响费用，有声视频与无声视频价格不同，请在调用 `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -542,10 +555,10 @@ audio直接影响费用，有声视频与无声视频价格不同，请在调用 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -735,13 +748,15 @@ SDK 的参数命名与[HTTP接口](#42703589880ts)基本一致，参数结构根 ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 美国 @@ -751,7 +766,7 @@ SDK 的参数命名与[HTTP接口](#42703589880ts)基本一致，参数结构根 `dashscope.base_http_api_url = 'https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -766,7 +781,7 @@ import dashscope import os # 以下为北京地域URL，各地域的URL不同 -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -814,7 +829,7 @@ from dashscope import VideoSynthesis import dashscope # 以下为北京地域URL，各地域的URL不同 -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -877,19 +892,21 @@ if __name__ == '__main__': ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **法兰克福** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -916,7 +933,7 @@ public class Ref2Video26 { static { // 以下为北京地域url，各地域的url不同 - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" @@ -986,7 +1003,7 @@ public class Ref2Video26Async { static { // 以下为北京地域url，各地域的url不同 - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md index 03862462..b5ad73b5 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md @@ -23,9 +23,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -39,13 +44,15 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -55,7 +62,7 @@ `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -74,7 +81,8 @@ ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -100,7 +108,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 可通过 `input.audio_url` 参数传入自定义音频的 URL。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -125,7 +134,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 若不提供 `input.audio_url` ，模型将根据视频内容自动生成匹配的背景音乐或音效。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -149,7 +159,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener > wan2.6 及wan2.5系列模型默认生成有声视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -172,7 +183,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 通过 negative\_prompt 排除“花朵”元素，避免其出现在视频画面中。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -477,13 +489,15 @@ duration直接影响费用。费用 = 单价（基于分辨率）× 时长（秒 ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -493,7 +507,7 @@ duration直接影响费用。费用 = 单价（基于分辨率）× 时长（秒 `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -514,10 +528,10 @@ duration直接影响费用。费用 = 单价（基于分辨率）× 时长（秒 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -743,13 +757,15 @@ SDK 的参数命名与HTTP接口基本一致，参数结构根据语言特性进 ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -759,7 +775,7 @@ SDK 的参数命名与HTTP接口基本一致，参数结构根据语言特性进 `dashscope.base_http_api_url = 'https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -771,8 +787,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -844,8 +860,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -955,13 +971,15 @@ if __name__ == '__main__': ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -971,7 +989,7 @@ if __name__ == '__main__': `Constants.baseHttpApiUrl = "https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -994,8 +1012,8 @@ import java.util.Map; public class Text2Video { static { - // 以下为北京地域url，各地域的url不同，获取url：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" @@ -1093,8 +1111,8 @@ public class Text2Video { static { - // 以下为北京地域url，各地域的url不同，获取url：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md index 0fea1012..eb4e31d9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md @@ -21,9 +21,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -33,20 +38,23 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### 请求参数 ## **多图参考** ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -71,7 +79,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 视频重绘 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -92,8 +101,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## **局部编辑** ``` -# 如果使用华北2（北京）地域的模型，需要将url替换为：https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -117,7 +126,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 视频延展 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -137,7 +147,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 视频画面扩展 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -1168,22 +1179,24 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### 请求参数 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md index 9fa72fb6..4c0b78dd 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md @@ -465,7 +465,7 @@ duration直接影响费用，按秒计费，请在调用前确认[模型价格]( ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md index a7b65d67..e6707b3c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md @@ -53,6 +53,8 @@ **新加坡地域**：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis` +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + **说明** - 创建成功后，使用接口返回的 `task_id` 查询结果，task\_id 有效期为 24 小时。**请勿重复创建任务**，轮询获取即可。 @@ -64,7 +66,7 @@ ## 视频换人 -以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 +以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 ``` curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ @@ -256,13 +258,13 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### 新加坡 `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -285,7 +287,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi 请将`0385dc79-5ff8-4d82-bcb6-xxxxxx`替换为真实的task\_id。 -> 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 +> 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/0385dc79-5ff8-4d82-bcb6-xxxxxx \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md index dbdb443a..3808f5dd 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md @@ -253,13 +253,13 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### 新加坡 `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md index 24b5e582..d6728f38 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md @@ -21,9 +21,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -33,13 +38,13 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -53,7 +58,7 @@ ## 纯指令编辑（修改视频风格） ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -79,7 +84,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## **指令+参考图编辑（局部替换）** ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -499,13 +504,13 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -526,10 +531,10 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -729,13 +734,13 @@ SDK 的参数命名与[HTTP接口](#e9e21dd3a6945)基本一致，参数结构根 ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -749,8 +754,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同 -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -855,8 +860,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同 -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -997,13 +1002,13 @@ if __name__ == '__main__': ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -1034,8 +1039,8 @@ import java.util.Base64; public class VideoEdit { static { - // 以下为北京地域url，各地域的url不同 - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" @@ -1184,8 +1189,8 @@ import java.util.Base64; public class VideoEdit { static { - // 以下为北京地域url，各地域的url不同 - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md index d4752f4b..1114b8b8 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md @@ -565,6 +565,10 @@ Body ###### 文本生成 +**说明** + +Qwen3.7-Plus-2026-05-26 调优后部署请联系商务经理。 + **模型服务** **模型代码** @@ -1602,6 +1606,10 @@ FAILED 导出的参数快照保存在云存储中，暂不支持访问或下载。 +**说明** + +URL 中的 `` 取值来自上一步查询快照列表接口返回的 `checkpoint` 字段（如 `checkpoint-20`），而非 `full_name` 或其他含 job\_id 前缀的字段值。 + ``` curl --request GET 'https://dashscope.aliyuncs.com/api/v1/fine-tunes//export/?model_name=' \ --header 'Authorization: Bearer '${DASHSCOPE_API_KEY} \ @@ -1638,7 +1646,7 @@ Path Parameter 是 -要发布的 Checkpoint ID。 +要发布的 Checkpoint ID。取值为查询快照列表接口返回的 `checkpoint` 字段值（如 `checkpoint-20`），不含 job\_id 前缀。 model\_name diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md index 732930c9..aae1bf20 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md @@ -4,7 +4,7 @@ ## **模型调优流程** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8010204871/CAEQZhiBgMDg9PGS2hkiIDNlZDFiMGRlMTJhOTQ1YzJhMmNjNDM3NzQ1ZjNiOGZk4608430_20240830103738.564.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3448225871/CAEQZhiBgMDg9PGS2hkiIDNlZDFiMGRlMTJhOTQ1YzJhMmNjNDM3NzQ1ZjNiOGZk4608430_20240830103738.564.svg) ## **步骤一：选择调优方式** @@ -776,3 +776,21 @@ Checkpoint 有保存时长限制，超过保存时长后将被自动清理，届 **说明** 您可以在模型调优任务列表中，点击训练失败任务右侧的**日志**，查看具体的训练失败原因。 + +### **微调后模型体验仍回答基座模型身份怎么办？** + +**原因说明** + +控制台**模型体验**页面不支持设置 system prompt，导致微调后的身份设定无法生效。 + +**解决方案** + +可以通过以下方式验证微调模型的身份设定效果： + +- **方案一：通过 API 调用传入 system prompt** + + 调用 API 时，在请求的 `system` 字段中填写身份设定内容，即可让模型按照微调后的设定进行回答。 + +- **方案二：在百炼控制台模型调试页面测试** + + 前往百炼控制台**模型调试**页面，在系统提示词输入框中填写身份设定内容，再进行测试。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md index ae2396d5..530186a5 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md @@ -25,7 +25,7 @@ ### **模型调优流程** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2700204871/CAEQZhiBgMDg9PGS2hkiIDNlZDFiMGRlMTJhOTQ1YzJhMmNjNDM3NzQ1ZjNiOGZk4608430_20240830103738.564.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5928225871/CAEQZhiBgMDg9PGS2hkiIDNlZDFiMGRlMTJhOTQ1YzJhMmNjNDM3NzQ1ZjNiOGZk4608430_20240830103738.564.svg) 详情参见： @@ -35,6 +35,10 @@ #### 文本生成 +**说明** + +Qwen3.7-Plus-2026-05-26 调优后部署请联系商务经理。 + **模型服务** **模型代码** @@ -679,6 +683,540 @@ qwen-plus-character-2025-11-06 ¥0.15/千Token +**点击此处查看更多千问 VL 模型使用限制** + +##### **使用限制** + +###### **输入文件限制** + +**图像限制** + +- **图像分辨率：** + + - 最小尺寸：图像的宽度和高度均须大于`10`像素。 + + - 宽高比：原图及缩放后的图像，长边与短边的比值不得超过 `200:1`。 + + > 图像缩放逻辑请参见 [计算图像的Token](https://help.aliyun.com/zh/model-studio/vision#ea146aaca47ng) 中 `smart_resize` 函数。 + + - 像素上限： + + - 推荐将图像分辨率控制在`8K(7680x4320)`以内。超过此分辨率的图像可能因文件过大、网络传输耗时过长而导致API调用超时。 + + - 自动缩放机制：模型可通过`max_pixels`和`min_pixels`调整图像大小；因此，提供超高分辨率的图像并不会提升识别精度，反而会增加调用失败的风险，建议在客户端提前将图像缩放至合理大小。 + +- **支持的图像格式** + + - 分辨率在4K `(3840x2160)`以下，支持的图像格式如下： + + **图像格式** + + **常见扩展名** + + **MIME Type** + + BMP + + .bmp + + image/bmp + + JPEG + + .jpe, .jpeg, .jpg + + image/jpeg + + PNG + + .png + + image/png + + TIFF + + .tif, .tiff + + image/tiff + + WEBP + + .webp + + image/webp + + HEIC + + .heic + + image/heic + + - 分辨率处于`4K(3840x2160)`到`8K(7680x4320)`范围，仅支持 JPEG、JPG 、PNG 格式 + +- **图像大小：** + + - 以公网URL传入时：Qwen3.7系列、Qwen3.6系列、Qwen3.5系列单个图像不超过 `20MB`，其他模型单个图像不超过`10MB` + + - 以本地路径传入时：单个图像不超过`10MB` + + - 以 Base64 编码传入时：编码后的字符串不超过`10MB` + + + > 如需压缩文件体积请参见 [如何将图像或视频压缩到满足要求的大小](https://help.aliyun.com/zh/model-studio/vision#ec8e0a8e03moe) 。 + +- **图片数量限制：**多图输入时根据传入方式不同，支持的图片数量上限有所区别： + + - 以公网URL或本地路径传入时： + + - Qwen3.7-Plus：最多 2048 张 + + - Qwen3.7-Flash、Qwen3.6-Plus、Qwen3.6-Flash、Qwen3.5-Plus、Qwen3.5-Flash、Qwen3-VL、Qwen-VL、QVQ系列：最多 256 张 + + - 以 Base64 编码传入时：最多 250 张 + + + Qwen-Omni系列模型请参考[全模态](https://help.aliyun.com/zh/model-studio/omni/)。 + + +> 同时受模型图文总 Token 上限（即最大输入）的限制，所有图片的总 Token 数必须小于模型的最大输入。 + +**视频限制** + +- **以图像列表传入，图像列表的数量有如下限制：** + + - `qwen3.7`系列、`qwen3.6`系列、`qwen3.5` 系列：最少传入 4 张图片，最多 8000 张图片 + + - `qwen3-vl-plus` 系列、`qwen3-vl-flash` 系列、`qwen3-vl-235b-a22b-thinking`、`qwen3-vl-235b-a22b-instruct`：最少传入 4 张图片，最多 2000 张图片 + + - 其他`Qwen3-VL`开源、`Qwen2.5-VL`（包括商业版和开源版）和`QVQ`系列模型：最少传入 4 张图片，最多 512 张图片 + + - 其他模型：最少传入 4 张图片，最多 80 张图片 + +- **以视频文件传入时：** + + - **视频大小：** + + - 以公网URL传入时： + + - `qwen3.7`系列、`qwen3.6`系列、`qwen3.5` 系列、`Qwen3-VL`系列、`qwen-vl-max` ：不超过 2GB； + + - `qwen-vl-plus` 系列、其他`qwen-vl-max`模型、`Qwen2.5-VL`开源系列及`QVQ`系列模型：不超过 1GB； + + - 其他模型不超过 150MB + + - 以 Base64 编码传入时：编码后的字符串小于 10MB； + + - 以本地文件路径传入时：视频本身不超过 100MB。 + + + > 如需压缩文件体积请参见 [如何将图像或视频压缩到满足要求的大小](https://help.aliyun.com/zh/model-studio/vision#ec8e0a8e03moe) 。 + + - **视频时长：** + + - `qwen3.7`系列、`qwen3.6`系列、`qwen3.5` 系列：2秒至2小时； + + - `qwen3-vl-plus`系列、`qwen3-vl-flash`系列、`qwen3-vl-235b-a22b-thinking`、`qwen3-vl-235b-a22b-instruct`：2 秒至 1 小时； + + - 其他`Qwen3-VL`开源系列、`qwen-vl-max` ：2 秒至 20 分钟； + + - `qwen-vl-plus`系列、 其他`qwen-vl-max`模型、`Qwen2.5-VL`开源系列及`QVQ`系列模型：2 秒至 10 分钟； + + - 其他模型：2 秒至 40 秒。 + + - **视频格式：** MP4、AVI、MKV、MOV、FLV、WMV 等。 + + - **视频尺寸：**无特定限制，模型可通过`max_pixels`和`min_pixels`自动调整视频尺寸，更大尺寸的视频文件不会有更好的理解效果。 + + - **视频数量限制：**最多可传入 64 个视频。 + + - **音频理解：**不支持对视频文件的音频进行理解。 + + +###### **文件传入方式** + +- **公网URL**：提供一个公网可访问的文件地址，支持HTTP或HTTPS协议。为获得最佳稳定性和性能，可将文件[上传至OSS](https://help.aliyun.com/zh/oss/user-guide/console-quick-start)或[上传文件获取临时URL](https://help.aliyun.com/zh/model-studio/get-temporary-file-url)，获取公网 URL。 + + **重要** + + 为确保模型能成功下载文件，提供的公网URL的响应头中**必须**包含 Content-Length（文件大小）和 Content-Type（媒体类型，如 image/jpeg）。任一字段缺失或者错误将会导致文件下载失败。 + +- **Base64编码传入：**将文件转换为 Base64 编码字符串再传入。 + +- **本地文件路径传入（仅限 DashScope SDK）：**传入本地文件的路径。 + + +> 关于文件传入方式的建议，请参见 [如何选择文件上传方式？](https://help.aliyun.com/zh/model-studio/vision#dc4e7260aauuo) + +**计算图像与视频的Token** + +## **图像** + +计算公式：`图像 Token = h_bar * w_bar / token_pixels + 2` + +- `h_bar、w_bar`：缩放后的图像长宽，模型在处理图像前会进行预处理，会将图像缩小至特定像素上限内，像素上限与`max_pixels`和`vl_high_resolution_images`参数的取值有关。 + +- `token_pixels`：每视觉`Token`对应的像素值，不同模型情况不同： + + - `qwen3.7系列`、`qwen3.6系列`、`qwen3.5系列`、`Qwen3-VL`、`qwen-vl-max`、`qwen-vl-plus`**：**每个`Token`对应 `32x32`像素 + + - `QVQ`及其他`Qwen2.5-VL`模型**：**每个Token对应`28x28`像素 + + +以下代码演示了模型内部对图像的大致缩放逻辑，可用于估算一张图像的Token，实际计费请以API响应为准。 + +**点此查看图像 token 估算示例** + +``` +import math +from PIL import Image # pip install Pillow + +def smart_size(image_path, max_pixels, vl_high_resolution_images): + """根据模型参数，计算图像缩放后的尺寸，用于估算图像 Token。""" + image = Image.open(image_path) + height, width = image.height, image.width + + # Qwen3.6、Qwen3.5、Qwen3-VL 等模型的缩放因子为 32；其他模型为 28 + factor = 32 + h_bar = round(height / factor) * factor + w_bar = round(width / factor) * factor + + # Token 下限：4 个 Token + min_pixels = 4 * factor * factor + + # vl_high_resolution_images=True 时，Token 上限固定为 16384，忽略 max_pixels + if vl_high_resolution_images: + max_pixels = 16384 * factor * factor + + # 将总像素数约束在 [min_pixels, max_pixels] 范围内 + if h_bar * w_bar > max_pixels: + beta = math.sqrt((height * width) / max_pixels) + h_bar = math.floor(height / beta / factor) * factor + w_bar = math.floor(width / beta / factor) * factor + elif h_bar * w_bar < min_pixels: + beta = math.sqrt(min_pixels / (height * width)) + h_bar = math.ceil(height * beta / factor) * factor + w_bar = math.ceil(width * beta / factor) * factor + + return h_bar, w_bar + +if __name__ == "__main__": + # 注意：max_pixels 和 vl_high_resolution_images 的值需要与调用模型时传入的参数保持一致 + h_bar, w_bar = smart_size("xxx/test.jpg", max_pixels=2560 * 32 * 32, vl_high_resolution_images=False) + print(f"缩放后的图像尺寸：高度 {h_bar}，宽度 {w_bar}") + + # 每张图像额外包含 和 各 1 个 Token + token = int(h_bar * w_bar / (32 * 32)) + 2 + print(f"图像的 Token 数：{token}") +``` + +## **视频** + +- **视频文件：** + + 模型处理视频文件时，会先进行抽帧，然后计算所有视频帧的总 Token 数。由于该计算过程较为复杂，可使用以下代码，通过传入视频路径来估算视频消耗的总 Token 数： + + ``` + # 使用前安装：pip install opencv-python + import math + import os + import logging + import cv2 + + logger = logging.getLogger(__name__) + + FRAME_FACTOR = 2 + + # Qwen3.6、Qwen3.5、Qwen3-VL、qwen-vl-max-0813、qwen-vl-plus-0815、qwen-vl-plus-0710等模型，图像缩放因子为32 + IMAGE_FACTOR = 32 + + # 其他模型，图像缩放因子为28 + # IMAGE_FACTOR = 28 + + # 视频帧的最大长宽比 + MAX_RATIO = 200 + # 视频帧的像素下限 + VIDEO_MIN_PIXELS = 4 * 32 * 32 + # 视频帧的像素上限，使用Qwen3-VL-Plus模型，VIDEO_MAX_PIXELS为640 * 32 * 32，其他模型为768 * 32 * 32 + VIDEO_MAX_PIXELS = 640 * 32 * 32 + + # 用户未传入FPS参数，则fps使用默认值 + FPS = 2.0 + # 最少抽取帧数 + FPS_MIN_FRAMES = 4 + # 最大抽取帧数（根据模型选择设置值） + FPS_MAX_FRAMES = 2000 + + # 视频输入的最大像素值，使用Qwen3-VL-Plus模型，请将VIDEO_TOTAL_PIXELS设置为131072 * 32 * 32，其他模型设置为65536 * 32 * 32 + VIDEO_TOTAL_PIXELS = int(float(os.environ.get('VIDEO_MAX_PIXELS', 131072 * 32 * 32))) + + def round_by_factor(number: int, factor: int) -> int: + """返回与”number“最接近的整数，该整数可被”factor“整除。""" + return round(number / factor) * factor + + def ceil_by_factor(number: int, factor: int) -> int: + """返回大于或等于“number”且可被“factor”整除的最小整数。""" + return math.ceil(number / factor) * factor + + def floor_by_factor(number: int, factor: int) -> int: + """返回小于或等于“number”且可被“factor”整除的最大整数。""" + return math.floor(number / factor) * factor + + def extract_vision_info(conversations): + vision_infos = [] + if isinstance(conversations[0], dict): + conversations = [conversations] + for conversation in conversations: + for message in conversation: + if isinstance(message["content"], list): + for ele in message["content"]: + if ( + "image" in ele + or "image_url" in ele + or "video" in ele + or ele.get("type","") in ("image", "image_url", "video") + ): + vision_infos.append(ele) + return vision_infos + + def smart_nframes(ele,total_frames,video_fps): + """用于计算抽取的视频帧数。 + + Args: + ele (dict): 包含视频配置的字典格式 + - fps: fps用于控制提取模型输入帧的数量。 + total_frames (int): 视频的原始总帧数。 + video_fps (int | float): 视频的原始帧率 + + Raises: + nframes应该在[FRAME_FACTOR，total_frames]间隔内，否则会报错 + + Returns: + 用于模型输入的视频帧数。 + """ + assert not ("fps" in ele and "nframes" in ele), "Only accept either `fps` or `nframes`" + fps = ele.get("fps", FPS) + min_frames = ceil_by_factor(ele.get("min_frames", FPS_MIN_FRAMES), FRAME_FACTOR) + max_frames = floor_by_factor(ele.get("max_frames", min(FPS_MAX_FRAMES, total_frames)), FRAME_FACTOR) + duration = total_frames / video_fps if video_fps != 0 else 0 + if duration-int(duration)>(1/fps): + total_frames = math.ceil(duration * video_fps) + else: + total_frames = math.ceil(int(duration)*video_fps) + nframes = total_frames / video_fps * fps + if nframes > total_frames: + logger.warning(f"smart_nframes: nframes[{nframes}] > total_frames[{total_frames}]") + nframes = int(min(min(max(nframes, min_frames), max_frames), total_frames)) + if not (FRAME_FACTOR <= nframes and nframes <= total_frames): + raise ValueError(f"nframes should in interval [{FRAME_FACTOR}, {total_frames}], but got {nframes}.") + + return nframes + + def get_video(video_path): + # 获取视频信息 + cap = cv2.VideoCapture(video_path) + + frame_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH)) + # 获取视频高度 + frame_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT)) + total_frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT)) + + video_fps = cap.get(cv2.CAP_PROP_FPS) + return frame_height, frame_width, total_frames, video_fps + + def smart_resize(ele, path, factor=IMAGE_FACTOR): + # 获取原视频的宽和高 + height, width, total_frames, video_fps = get_video(path) + # 视频帧的Token下限 + min_pixels = VIDEO_MIN_PIXELS + total_pixels = VIDEO_TOTAL_PIXELS + # 抽取的视频帧数 + nframes = smart_nframes(ele, total_frames, video_fps) + max_pixels = max(min(VIDEO_MAX_PIXELS, total_pixels / nframes * FRAME_FACTOR),int(min_pixels * 1.05)) + + # 视频的长宽比不应超过200:1或1:200 + if max(height, width) / min(height, width) > MAX_RATIO: + raise ValueError( + f"absolute aspect ratio must be smaller than {MAX_RATIO}, got {max(height, width) / min(height, width)}" + ) + + h_bar = max(factor, round_by_factor(height, factor)) + w_bar = max(factor, round_by_factor(width, factor)) + if h_bar * w_bar > max_pixels: + beta = math.sqrt((height * width) / max_pixels) + h_bar = floor_by_factor(height / beta, factor) + w_bar = floor_by_factor(width / beta, factor) + elif h_bar * w_bar < min_pixels: + beta = math.sqrt(min_pixels / (height * width)) + h_bar = ceil_by_factor(height * beta, factor) + w_bar = ceil_by_factor(width * beta, factor) + return h_bar, w_bar + + def token_calculate(video_path, fps): + # 传入视频路径和fps抽帧参数 + messages = [{"content": [{"video": video_path, "fps": fps}]}] + vision_infos = extract_vision_info(messages)[0] + + resized_height, resized_width = smart_resize(vision_infos, video_path) + + height, width, total_frames, video_fps = get_video(video_path) + num_frames = smart_nframes(vision_infos, total_frames, video_fps) + print(f"原视频尺寸：{height}*{width}， 输入模型的尺寸：{resized_height}*{resized_width}，视频总帧数:{total_frames}，fps等于{fps}时，抽取的总帧数：{num_frames}", end="，") + video_token = int(math.ceil(num_frames / 2) * resized_height / 32 * resized_width / 32) + video_token += 2 # 系统会自动添加<|vision_bos|>和<|vision_eos|>视觉标记（各计1个Token） + return video_token + + video_token = token_calculate("xxx/test.mp4", 1) + print("视频tokens:", video_token) + ``` + +- **图像列表：** + + 当以图像列表形式传入视频时，表示已预先完成视频抽帧，可使用以下代码，通过传入图像的路径和数量来计算传入图像列表时消耗的Token数： + + ``` + # 使用前安装：pip install Pillow + import math + import os + import logging + from typing import Tuple + from PIL import Image + + logger = logging.getLogger(__name__) + + # ==================== 常量定义 ==================== + FRAME_FACTOR = 2 + # Qwen3-VL、qwen-vl-max-0813、qwen-vl-plus-0815、qwen-vl-plus-0710模型，缩放因子为32 + IMAGE_FACTOR = 32 + + # 其他模型，缩放因子为28 + # IMAGE_FACTOR = 28 + + # Token计算相关常量 + TOKEN_DIVISOR = 32 # token计算时的除数 + VISION_SPECIAL_TOKENS = 2 # <|vision_bos|>和<|vision_eos|>标记 + + # 视频帧的最大长宽比 + MAX_RATIO = 200 + # 视频帧的像素下限 + VIDEO_MIN_PIXELS = 4 * 32 * 32 + # 视频帧的像素上限，使用Qwen3-VL-Plus模型，VIDEO_MAX_PIXELS为640 * 32 * 32，其他模型为768 * 32 * 32 + VIDEO_MAX_PIXELS = 640 * 32 * 32 + + # 视频输入的最大像素值，使用Qwen3-VL-Plus模型，请将VIDEO_TOTAL_PIXELS设置为131072 * 32 * 32，其他模型设置为65536 * 32 * 32 + VIDEO_TOTAL_PIXELS = int(float(os.environ.get('VIDEO_MAX_PIXELS', 131072 * 32 * 32))) + + def round_by_factor(number: int, factor: int) -> int: + """返回与”number“最接近的整数，该整数可被”factor“整除。""" + return round(number / factor) * factor + + def ceil_by_factor(number: int, factor: int) -> int: + """返回大于或等于“number”且可被“factor”整除的最小整数。""" + return math.ceil(number / factor) * factor + + def floor_by_factor(number: int, factor: int) -> int: + """返回小于或等于“number”且可被“factor”整除的最大整数。""" + return math.floor(number / factor) * factor + + def get_image_size(image_path: str) -> Tuple[int, int]: + if not os.path.exists(image_path): + raise FileNotFoundError(f"图像文件不存在: {image_path}") + + try: + image = Image.open(image_path) + height = image.height + width = image.width + image.close() # 及时关闭文件 + return height, width + except Exception as e: + raise ValueError(f"无法读取图像文件 {image_path}: {str(e)}") + + def smart_resize(height: int, width: int, nframes: int, factor: int = IMAGE_FACTOR) -> Tuple[int, int]: + """ + 计算图像缩放后的尺寸 + + Args: + height: 原始图像高度 + width: 原始图像宽度 + nframes: 视频帧数 + factor: 缩放因子，默认为IMAGE_FACTOR + + Returns: + (resized_height, resized_width) 缩放后的高度和宽度 + + Raises: + ValueError: 长宽比超过限制 + """ + # 视频帧的Token下限 + min_pixels = VIDEO_MIN_PIXELS + total_pixels = VIDEO_TOTAL_PIXELS + # 抽取的视频帧数 + max_pixels = max(min(VIDEO_MAX_PIXELS, total_pixels / nframes * FRAME_FACTOR), int(min_pixels * 1.05)) + + # 视频的长宽比不应超过200:1或1:200 + aspect_ratio = max(height, width) / min(height, width) + if aspect_ratio > MAX_RATIO: + raise ValueError( + f"图像长宽比必须小于 {MAX_RATIO}:1，当前为 {aspect_ratio:.2f}:1" + ) + + h_bar = max(factor, round_by_factor(height, factor)) + w_bar = max(factor, round_by_factor(width, factor)) + if h_bar * w_bar > max_pixels: + beta = math.sqrt((height * width) / max_pixels) + h_bar = floor_by_factor(height / beta, factor) + w_bar = floor_by_factor(width / beta, factor) + elif h_bar * w_bar < min_pixels: + beta = math.sqrt(min_pixels / (height * width)) + h_bar = ceil_by_factor(height * beta, factor) + w_bar = ceil_by_factor(width * beta, factor) + return h_bar, w_bar + + def calculate_video_tokens(image_path: str, nframes: int = 1, factor: int = IMAGE_FACTOR, verbose: bool = True) -> int: + """ + + Args: + image_path: 视频帧文件路径 + nframes: 视频帧数， + factor: 缩放因子，默认为IMAGE_FACTOR + verbose: 是否打印详细信息 + + Returns: + 所消耗的token数量 + + Raises: + FileNotFoundError: 文件不存在 + ValueError: 文件格式无效或长宽比超限 + """ + # 获取原始图像尺寸（只读取一次） + height, width = get_image_size(image_path) + + # 计算缩放后的尺寸 + resized_height, resized_width = smart_resize(height, width, nframes, factor) + + # 计算token数量 + # 公式：⌈帧数/2⌉ × (高度/TOKEN_DIVISOR) × (宽度/TOKEN_DIVISOR) + VISION_SPECIAL_TOKENS + video_token = int( + math.ceil(nframes / 2) * + (resized_height / TOKEN_DIVISOR) * + (resized_width / TOKEN_DIVISOR) + ) + # 添加视觉标记token（<|vision_bos|>和<|vision_eos|>） + video_token += VISION_SPECIAL_TOKENS + + if verbose: + print(f"原视频帧尺寸：{height}×{width}，输入模型的尺寸：{resized_height}×{resized_width}，", end="") + + return video_token + + if __name__ == "__main__": + try: + video_token = calculate_video_tokens("xxx/test.jpg", nframes=30) + print(f"视频tokens: {video_token}\n") + except Exception as e: + print(f"错误: {str(e)}\n") + ``` + + ## **模型调优前必读** - 文本生成模型调优虽然能在特定业务/场景取得非常好的效果，但有以下限制： diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/rl-training-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/rl-training-overview.md new file mode 100644 index 00000000..d747c187 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/rl-training-overview.md @@ -0,0 +1,448 @@ +# 强化学习训练概述 + +本文通过端到端示例帮助您快速完成第一个强化学习（RL） 训练任务的提交与监控。 + +**如需了解 RL 训练的基本概念和适用场景，请点击这里。** + +百炼提供的多种调优方式并不互斥，而是递进的、相辅相成的。 + +`CPT（可选）→ SFT → DPO（可选）→ RL（可选）` + +1. CPT (持续预训练）- 补知识 （通用模型知识的“广度”和“浅度”，无法满足专业领域的“深度”和“精度”要求） + + - 金融模型： `学金融术语` + + - 医疗模型： `记药品病理` + + - 法律模型： `懂法条判例` + +2. SFT (监督微调）- 学做事 + + - 客服机器人： `学客服流程` + + - 代码助手： `学编程范式` + + - 工具调用 (Agent)： `学使用 MCP` + +3. DPO (直接偏好优化）- 做得更好 + + - 安全与责任感： `拒有害建议` + + - 简洁与有效性： `答干脆利落` + + - 客观与中立： `评公正客观` + +4. RL（强化学习）- 学推理（通过 Reward 信号驱动模型自主探索最优策略，无需提供标准答案） + + - 数学推理： `解题更准` + + - 工具调用 (Agent)： `调得更稳` + + - 代码生成： `写得能跑` + + +## RL 训练全流程概览 + +RL 训练从环境准备到模型部署，共 5 步： + +1. **环境准备** — 安装离线 SDK、获取 API Key、完成 RL 授权 + +2. **准备数据与代码** — 下载 Demo 包，了解数据格式与项目结构 + +3. **提交训练任务** — 运行 Demo 一步完成函数部署、数据上传、任务提交 + +4. **训练状态观测** — 查看训练状态、日志和 Reward 趋势 + +5. **发布与部署模型** — 发布 Checkpoint 至"我的模型"，部署后通过 API 调用 + + +### RL 训练原理 + +RL 训练通过"生成-评分-优化"循环不断提升模型能力。每轮训练的数据流如下： + +``` +训练数据（用户问题 + 参考答案） + │ + ▼ +策略采样（Rollout）：模型按当前策略生成候选结果 + │ + ▼ +奖励评分（Reward）：对比参考答案，给出 0~1 分（可以合并到 Rollout） + │ + ▼ +策略更新：根据奖励信号，强化高分策略、抑制低分策略 + │ + ▼ +循环迭代：用更新后的模型重新采样，逐轮提升质量 +``` + +## 支持的模型 + +**说明** + +请联系**商务经理**开启强化学习训练功能。 + +**模型名称** + +**模型 code** + +**是否为 MoE（混合专家）模型** + +**推荐的模型训练单元数量** + +千问3.5-9B + +qwen3.5-9b + +否 + +IV型模型单元 \* 24 + +千问3.6-flash-2026-04-16 + +qwen3.6-flash-2026-04-16 + +是 + +IV型模型单元 \* 72 + +千问3.5-flash-2026-02-23 + +qwen3.5-flash-2026-02-23 + +是 + +IV型模型单元 \* 72 + +## 训练单元价格 + +**计费方式** + +**说明** + +**计费公式** + +训练单元·预付费 + +按月购买模型训练单元，使用专属训练资源，**训练速度更快，无需排队等待**。 最小计费粒度：**小时**。提前退订按小时单价的 **1.2** 倍计费。 如需购买或管理预付费训练单元，可在[模型调优控制台](https://bailian.console.aliyun.com/cn-beijing/efm/model_manager?tab=model#/efm/model_manager)点击**管理训练资源**。 + +`单价 × 实例数 × 购买时长` + +训练单元·后付费 + +使用专属训练资源，**训练速度更快，无需排队等待**。按实际使用时长计费，创建任务时直接开通，无需预先购买。 + +`小时单价 × 实例数 × 使用小时数` + +**训练单元类型** + +**计费方式** + +**单价** + +**最小计费粒度** + +IV 型（MTU4） + +预付费（包月） + +19,914.00 元/月/实例 + +1 小时 + +后付费（按分钟） + +41.00 元/小时/实例 + +1 分钟 + +**说明** + +预付费扩缩容、续费、退订等运营规则与具体 MTU 容量测算请联系商务经理。 + +## 环境准备 + +**警告** + +RL 训练**仅支持**通过模型训练单元（MTU）计费，不支持按 Token 计费方式。 + +按照以下步骤完成环境配置： + +1. **获取 API Key**：在百炼控制台获取 DashScope API Key。 + +2. **完成 RL 服务授权**：在百炼控制台的**模型调优**页面完成一键授权。首次使用 RL 训练时，控制台会提示授权**阿里云 OpenTelemetry**、**函数计算（FC）**和**日志服务（SLS）**三项云服务。 + + ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7125558771/p1073237.png) + +3. **设置环境变量**：以下变量由本地 SDK 读取，控制打包、上传和部署行为： + + - **DASHSCOPE\_API\_KEY**（必填）：API 密钥。 + + - **FC\_PYPI\_LIB**（必填）：指定 FC 容器启动时安装的 dashscope 离线包。SDK 打包部署时会自动将该变量注入为 `FC_SDK_PACKAGE`，无需在 FC 侧手动安装。whl 文件名需与此变量**完全一致**，且放在项目根目录下。 + + - **LOG\_LEVEL**（可选，默认 `info`）：设为 `debug` 可输出完整请求/响应信息，便于排查提交失败等问题。 + + + ``` + export DASHSCOPE_API_KEY="sk-your-api-key" + export LOG_LEVEL="info" + ``` + + +## 准备数据与代码 + +1. 下载 Demo 包[agentic-rl-example.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260721/jfslkz/agentic-rl-example-for-customer-9bae5b71d53759bc7d051cda8f4671349d6164a5.zip)。（该 Demo 运行最低要求 24 个 IV 型模型训练单元） + +2. **下载完成后需要安装 DashScope SDK**（要求 Python >= 3.10）：请按照 Demo 包中附带的指引安装相关依赖： + + ``` + # 使用PyPI源： + pip install dashscope + ``` + + +您将获得以下项目结构： + +``` +agentic-rl-example/ +├── submit_job.py ← 主入口，提交训练任务 +├── test_functions.py ← 函数注册与远程测试 +├── config.yaml ← 可选的 YAML 配置文件 +├── job.yaml ← YAML 提交配置（CLI run / SDK init） +├── cli.sh ← 本地测试与查询脚本集合 +├── dashscope-1.25.23-py3-none-any.whl ← 离线 SDK 安装包 +├── requirements.txt ← 依赖清单 +├── data/ +│ ├── calc_train.jsonl ← 训练数据（完整版） +│ ├── calc_train_min.jsonl ← 训练数据（Demo 默认使用精简版） +│ ├── calc_validation.jsonl ← 验证数据（完整版） +│ └── calc_validation_min.jsonl ← 验证数据（Demo 默认使用精简版） +├── functions/ +│ ├── __init__.py +│ ├── rollout/ +│ │ ├── rollout.py ← Rollout 函数（LangGraph + MCP 工具） +│ │ └── rollout_only.py ← Rollout 函数（简单版，含追踪示例） +│ └── reward/ +│ ├── reward.py ← Reward 函数（基础评分） +│ ├── reward_decorator.py ← Reward 函数（装饰器风格） +│ └── group_reward.py ← Reward 函数（分组评分） +├── resources/ +│ ├── rollout_input.json ← Rollout 测试输入 +│ ├── reward_input.json ← Reward 测试输入 +│ ├── reward_decorator_input.json ← 装饰器 Reward 测试输入 +│ └── group_reward_input.json ← 分组 Reward 测试输入 +└── scripts/ ← 本地启动与查询辅助脚本（build/install/start_local_*/query_local_*/query_online_tuning/convert_parquet_to_jsonl） +``` + +### Demo 简介 + +本 Demo 实现了一个数学计算 Agent： + +- **Rollout 函数**：调用训练中的模型，配合计算器工具回答数学问题 + +- **Reward 函数**：将模型输出与参考答案对比，判定回答是否正确并评分 + + +**说明** + +Demo 中的 Rollout 函数使用 LangGraph 编排 Agent 流程，通过 **MCP**（Model Context Protocol，一种 LLM 工具调用协议）调用计算器工具。 + +如需编写自己的 Rollout 和 Reward 函数，请参见[强化学习开发指南](https://help.aliyun.com/zh/model-studio/rl-function-development-guide)。 + +### 数据格式 + +训练数据采用 **JSONL** 格式，每行一个 JSON 对象，包含 `messages` 和 `rollout_extra` 两个字段： + +``` +{"messages": [{"content": "Output the answer when you are ready. The answer should be surrounded by three sharps (`###`), in the form of ### ANSWER: ###. 6.6 minus x (3/2) times equals 5.6.", "role": "user"}], "rollout_extra": {"solution": "2/3"}} +``` + +- `messages`：用户问题，格式与 ChatML 一致 + +- `rollout_extra`：存放参考答案或其他业务数据，会传递给 Reward 函数用于评分（rollout\_extra 支持自定义字段） + + +`rollout_extra` 中的数据会透传到 Rollout 函数的 `RolloutInput.rollout_extra` 和 Reward 函数的 `RewardInput.agent_output.rollout_extra`，方便在评分逻辑中使用。该字段支持自定义 key，不限于示例中的 `solution`。 + +### 数据量建议 + +- **起步**：几十到几百条即可验证方案有效性（数据量至少大于 `Batch_size`）。 + +- **正式训练**：数据量越大效果越优，大规模数据集还支持更大的 batch\_size，有助于提升训练稳定性 + + +## 提交训练任务 + +Demo 中已包含 Rollout 和 Reward 函数，运行以下命令即可一步完成函数部署、数据上传、训练任务提交： + +``` +python submit_job.py +``` + +**完整代码及参数说明** + +以下为训练脚本的完整代码，注释中说明了各参数的含义和取值建议： + +``` +import asyncio +from dashscope.finetune.reinforcement import ( + RolloutFunctionComponent, RewardFunctionComponent, + FunctionComponentModel, FunctionComponentRuntime, + TrainingDataset, ValidationDataset, DataSourceType, +) +from dashscope.finetune.agentic_rl import AgenticRL + +async def main(): + client = AgenticRL() + + # client.run() 一步完成：注册函数到 FC → 上传训练数据 → 提交训练任务 + result = await client.run( + model="qwen3.5-9b", # 基座模型 + training_datasets=[TrainingDataset( # 训练集 + data_source_type=DataSourceType.FILE_ID, + file_name="./data/calc_train_min.jsonl", + )], + validation_datasets=[ValidationDataset( # 验证集（可选） + data_source_type=DataSourceType.FILE_ID, + file_name="./data/calc_validation_min.jsonl", + )], + functions=[ + # Rollout 函数：定义模型如何执行策略采样 + RolloutFunctionComponent( + name="rollout-1", + timeout=600, + fcmodel=FunctionComponentModel( + classpath="functions.rollout.rollout.CalcXRolloutProcessor"), + runtime=FunctionComponentRuntime( + cpu=2, memory_size=4096, disk_size=512, + concurrency=30, capacity=30, + min_capacity=30, max_capacity=60, + memory_scale_threshold=0.6, + concurrency_scale_threshold=0.6, + env={})), # FC 函数环境变量 + RewardFunctionComponent( + name="reward-1", + weight=1.0, + timeout=120, + reward_metric_weight={"reward_metric_weightA": 0.3, "reward_metric_weightB": 0.7}, + fcmodel=FunctionComponentModel( + classpath="functions.reward.reward.DemoRewardProcessor"), + runtime=FunctionComponentRuntime( + cpu=2, memory_size=4096, disk_size=512, + concurrency=30, capacity=30, + min_capacity=30, max_capacity=60, + memory_scale_threshold=0.6, + concurrency_scale_threshold=0.6, + env={})), + ], + # MTU 计费配置 + resources={ + "charge_type": "mtu_postpaid", # 后付费 + "mtu_spec_code": "MTU4", # 训练单元规格 + "mtu_capacity": 24, # MTU 数量 + }, + # GSPO 算法超参数（以下为 qwen3.5-9b 非 MoE 模型的全部必选超参） + hyper_parameters={ + "algorithm": "gspo", # 训练算法 + "batch_size": 64, # 批大小 + "eval_steps": 1, # 每 N 步执行一次验证 + "kl_loss_coef": 0.002, # KL 散度损失系数 + "learning_rate": 2e-6, # 学习率 + "lr_scheduler_type": "linear", # 学习率调度策略 + "max_length": 8192, # 最大序列长度（prompt + response） + "n_epochs": 1, # 训练轮数 + "n_rollouts": 8, # 每个样本的 Rollout 次数 + "ppo_mini_batch_size": 8, # PPO 小批量大小 + "save_strategy": "epoch", # 保存策略（steps 或 epoch） + }) + + if result.status_code == 200: + print(f"训练任务已提交，Job ID: {result.output.job_id}") + else: + print(f"提交失败: {result}") + +asyncio.run(main()) +``` + +**说明** + +也可以通过配置文件或 CLI 提交训练任务（`dashscope rl run`），三种提交方式与字段逐项说明详见 [《训练配置 — 提交与配置》](https://help.aliyun.com/zh/model-studio/rl-training-config-monitoring)。 + +### 起步技巧 + +第一次跑通后，建议参考以下 4 项实践： + +- **先用精简数据**：Demo 默认 `calc_train_min.jsonl` + `n_epochs=1` 跑一次，确认链路通了再换完整数据集，避免拿 24 个 MTU 烧大数据集才发现配置错。 + +- **超参先不动**：上面 11 项超参是 qwen3.5-9b 的推荐起点，第一次训练别调。要调时一次只动一个变量（lr 或 batch\_size），跑满 `eval_steps × 3` 步再判断趋势。 + +- **盯一个核心指标**：在**指标**页签看 `critic/rewards/mean`——稳步上升说明在学；停滞或下降就停下来排查，别空跑。 + +- **FAILED 先看日志**：任务失败 → **日志**页签看末尾报错，或 SDK 拉 `AgenticRL.logs(job_id="ft-xxx", lines=100)`。配置/依赖类报错通常一眼能定位。 + + +**说明** + +算法选型、超参起点表、调参决策详见《强化学习训练配置 — 提交与配置》；Reward 函数设计哲学与 Hacking 防御详见《强化学习开发指南》；看到指标异常如何归因详见《强化学习的可观测配置与指标参考》。 + +## 训练过程观测与分析 + +任务提交后，`run()` 返回的 `job_id` 是后续查询训练状态的唯一标识。训练通常需要数十分钟到数小时，取决于数据量和模型大小。 + +在百炼控制台的**模型调优**页面点击任务进入详情页，包含以下 5 个页签： + +**页签** + +**定位** + +**主要内容** + +**详情** + +任务元数据 + +任务 ID、状态、基座模型、训练方法、起止时间、数据配置 + +**轨迹** + +Agent 行为回放 + +完整对话过程、Reward 分析（Step / Sample / Trajectory 三维度）、工具调用详情 + +**指标** + +训练健康度 + +13 组指标图表（actor / critic / trajectory / trace / timing / perf / fully\_async 等），含用户自定义指标 + +**日志** + +运行排查 + +训练过程 stdout / stderr，任务失败时用于定位报错原因 + +**产出** + +模型资产 + +Checkpoint 列表、发布状态，可将 Checkpoint 发布至"我的模型"后部署调用 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5224658771/p1073270.png) + +各页签的详细说明请参见[强化学习的可观测配置](https://help.aliyun.com/zh/model-studio/observable-configuration-for-reinforcement-learning)。 + +## 发布与部署模型 + +训练完成后，最后一个 Checkpoint 会自动发布至**我的模型**。如需发布中间 Checkpoint，前往百炼控制台**模型调优** > **产出**页面手动操作，详见_模型调优控制台（百炼文档，需在线查看）_。 + +发布后的模型可在**我的模型**页面进行部署。部署完成后即可通过 API 调用模型。详见_模型部署（百炼文档，需在线查看）_。 + +## 后续步骤 + +- 了解函数开发的完整细节 → [强化学习开发指南](https://help.aliyun.com/zh/model-studio/rl-function-development-guide) + +- 了解训练配置和监控 → [强化学习训练配置 — 提交与配置](https://help.aliyun.com/zh/model-studio/rl-training-config-monitoring) + +- 训练指标参考字典 → [强化学习的可观测配置与指标参考](https://help.aliyun.com/zh/model-studio/observable-configuration-for-reinforcement-learning) + +- 了解可观测性配置与轨迹查看 → [强化学习的可观测配置与指标参考](https://help.aliyun.com/zh/model-studio/observable-configuration-for-reinforcement-learning) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md index e688c82e..5c1b36a4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md @@ -115,6 +115,144 @@ curl --location --request POST 'https://dashscope.aliyuncs.com/api/v1/files' \ 请将`<替换为训练数据集的文件id>`完整替换为上一步获取的`file_id`。完整参数说明与格式约束请参见[超参数](https://help.aliyun.com/zh/model-studio/wan-generation-finetune-api-reference#5f391e4b3cezf)。 +**超参数** + +**字段** + +**类型** + +**必选** + +**描述** + +**推荐值** + +max\_steps + +int + +是 + +**训练总步数**。控制训练时长的核心参数。max\_steps 决定训练迭代次数，max\_token\_length 决定每步处理的数据量。建议不少于 500 步以确保模型充分收敛；大数据集可适当增加步数。 + +800 + +eval\_steps + +int + +是 + +**验证间隔**。取值需≥0。训练期间每隔多少个 steps 进行一次验证评估，用于阶段性评估模型训练效果。同时保存当前 step 的模型文件。 + +200 + +learning\_rate + +float + +是 + +**学习率**。控制模型权重更新的幅度。过高可能导致模型变差，过低则变化不明显。推荐使用默认值。 + +3e-5 + +generation\_type + +string + +是 + +**生成模式**。`"t2i"`：文生图模式；`"i2i"`：图生图模式。决定训练数据格式和推理方式。 + +t2i + +max\_pixels + +string + +是 + +**训练图片的最大分辨率**。例如 "1k"、"2k"（1K 即 1024×1024，2K 即 2048×2048）。设置训练集中图片分辨率的像素总数（宽×高）上限，系统仅对超过该值的图片进行缩放处理，未超限的图片保持原样。建议三个分辨率参数（max\_pixels、max\_token\_length、val\_img\_size）保持一致。 + +文生图："2k" +图生图："1k" + + +val\_img\_size + +string + +是 + +**验证图生成分辨率**。例如 "1k"、"2k"（1K 即 1024×1024，2K 即 2048×2048）。训练过程中验证评估时生成图片的目标分辨率。 + +文生图："2k" +图生图："1k" + + +max\_token\_length + +string + +是 + +**每步训练的最大 Token 长度**。例如 "1k"、"2k"。与 max\_steps 共同控制训练过程：max\_steps 决定迭代次数，max\_token\_length 决定每步处理的数据量。 + +文生图："2k" +图生图："1k" + + +gradient\_clip + +float + +是 + +**梯度裁剪**。对所有可训练参数做全局梯度范数裁剪的阈值，防止梯度爆炸。设为 -1 表示不裁剪。 + +0.5 + +weight\_decay + +float + +是 + +**权重衰减**。AdamW 解耦式权重衰减系数，对所有可训练参数生效，用于正则化防止过拟合。 + +0.02 + +lora\_rank + +int + +是 + +**LoRA 低秩矩阵的维数**。该值决定了微调参数量的大小。数值越大，模型拟合能力越强，但训练速度会变慢。取值必须为 2n（如 16、32、64）。 + +32 + +save\_total\_limit + +int + +否 + +**Checkpoint 保存数量上限**。限制最多保存的模型数量。系统将始终只保存训练生成的最后 N 个 Checkpoint（N 为该参数值）。 + +10 + +split + +float + +否 + +**训练集划分比例**。取值范围为 (0, 1)。仅在未指定 `validation_datasets` 时生效。此参数用于从训练集中自动按比例拆分出验证集。例如，0.9 表示 90% 训练集，10% 验证集。 + +0.9 + ``` curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md index 34eb3507..955a59f1 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md @@ -109,10 +109,6 @@ curl --location --request POST 'https://dashscope.aliyuncs.com/api/v1/files' \ 使用步骤1中的文件ID启动训练任务。 -**说明** - -不同模型的微调参数的值有所差异，超参数设置请参见[超参数](https://help.aliyun.com/zh/model-studio/wan-generation-finetune-api-reference#5f391e4b3cezf)，更多调用示例请参见[请求示例](https://help.aliyun.com/zh/model-studio/wan-generation-finetune-api-reference#1a9196bd16o9h)。 - **请求示例** 请将`<替换为训练数据集的文件id>`完整替换为上一步获取的`file_id`。 @@ -130,12 +126,12 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes' \ ], "training_type": "efficient_sft", "hyper_parameters": { - "n_epochs": 400, + "n_epochs": 50, "batch_size": 1, "learning_rate": 2e-5, "split": 0.9, "max_split_val_dataset_sample": 5, - "eval_epochs": 50, + "eval_epochs": 20, "max_pixels": 102400, "save_total_limit": 10, "lora_rank": 32, @@ -157,12 +153,12 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes' \ ], "training_type": "efficient_sft", "hyper_parameters": { - "n_epochs": 400, + "n_epochs": 50, "batch_size": 4, "learning_rate": 2e-5, "split": 0.9, "max_split_val_dataset_sample": 5, - "eval_epochs": 50, + "eval_epochs": 20, "max_pixels": 262144, "save_total_limit": 10, "lora_rank": 32, @@ -171,6 +167,140 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes' \ }' ``` +**超参数（hyper\_parameters）** + +**字段** + +**类型** + +**必选** + +**描述** + +**推荐值** + +batch\_size + +int + +是 + +**批次大小**。一次性送入模型进行训练的数据条数。 + +- wan2.7-i2v：推荐为 1。 + +- wan2.5-i2v-preview：推荐为 4。 + +- wan2.2-i2v-flash：推荐为 4。 + +- wan2.2-kf2v-flash：推荐为 4。 + + +以模型为准 + +n\_epochs + +int + +是 + +**训练循环次数**。steps = n\_epochs × ⌈数据集大小 / batch\_size⌉。建议总步数 ≥ 800。 + +> 例如：数据集 5 条，batch\_size=2，每轮步数=⌈5/2⌉=3，最小 n\_epochs = 800/3 ≈ 267。 + +> 推荐训练轮数会根据数据量自动调整。数据越少，需要更多轮数来充分学习；数据越多，每轮包含的样本越多，因此所需轮数会减少。50 epochs 主要适用于 2 条左右的小数据集；当数据量达到 50-60 条视频时，通常建议训练约 3000-5000 steps 即可。 + +50 + +learning\_rate + +float + +是 + +**学习率**。控制模型权重更新幅度。过高可能导致模型变差，过低则变化不明显。 + +2e-5 + +eval\_epochs + +int + +是 + +**验证间隔**。取值需 ≥ `n_epochs/10`。每隔多少个 epoch 进行一次验证评估并保存 Checkpoint。 + +20 + +max\_pixels + +int + +是 + +**训练视频的最大分辨率**（像素总数 = 宽×高）。系统仅对超过该值的视频进行缩放处理。 + +- wan2.7-i2v：推荐 102400。范围 36864～123904。 + +- wan2.5-i2v-preview：推荐 36864。范围 16384～36864。 + +- wan2.2-i2v-flash：推荐 262144。范围 65536～262144。 + +- wan2.2-kf2v-flash：推荐 262144。范围 65536～262144。 + + +以模型为准 + +split + +float + +否 + +**训练集划分比例**。取值 (0,1)，仅在未指定 validation\_datasets 时生效。 + +0.9 + +max\_split\_val\_dataset\_sample + +int + +否 + +**自动划分验证集的最大样本数**。验证集数量 = min(总数×(1−split), 此值)。 + +5 + +save\_total\_limit + +int + +否 + +**Checkpoint 保存数量上限**。系统只保存最后 N 个 Checkpoint。 + +10 + +lora\_rank + +int + +否 + +**LoRA 低秩矩阵维数**。取值须为 2n（16/32/64）。 + +32 + +lora\_alpha + +int + +否 + +**LoRA 权重缩放系数**。取值须为 2n（16/32/64）。 + +32 + **响应示例** 关注 `output` 中的三个关键参数： @@ -1276,7 +1406,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes/<替换为微 - **n\_epochs (训练轮数)** - - 默认值：**400**，推荐使用默认值。若需调整，请遵循 **“总训练步数 (Steps) ≥ 800”** 的原则。 + - 默认值：**50**，推荐使用默认值。若需调整，请遵循 **”总训练步数 (Steps) ≥ 800”** 的原则。 - 总步数计算公式： `steps = n_epochs × 向上取整（训练集大小 / batch_size）。` diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md index bb37de23..49deecfe 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md @@ -12,6 +12,16 @@ [ +qwen3.8-max-preview + +](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +仅 Token Plan 可用 + +qwen3.8-max-preview 目前仅面向 Token Plan 订阅用户提供，[前往开通 Token Plan →](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +[ + qwen3.7-max ](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-max) @@ -228,27 +238,27 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/a [ -qwen3.6-flash +qwen3.7-flash -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.6-flash) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-flash) 华北2（北京）新加坡日本（东京）德国（法兰克福）美国（弗吉尼亚） OpenAI 兼容Anthropic 兼容DashScope -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/apps/anthropic` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1` @@ -256,19 +266,19 @@ API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api- OpenAI 兼容Anthropic 兼容DashScope -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/apps/anthropic` API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1` @@ -276,19 +286,19 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=mod OpenAI 兼容Anthropic 兼容DashScope -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-northeast-1?tab=globalset#/efm/business_management).ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://modelstudio.console.aliyun.com/ap-northeast-1?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-northeast-1?tab=globalset#/efm/business_management).ap-northeast-1.maas.aliyuncs.com/apps/anthropic` API Key[获取↗](https://modelstudio.console.aliyun.com/ap-northeast-1?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-northeast-1?tab=globalset#/efm/business_management).ap-northeast-1.maas.aliyuncs.com/api/v1` @@ -296,19 +306,19 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/ap-northeast-1?tab=mod OpenAI 兼容Anthropic 兼容DashScope -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/apps/anthropic` API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/api/v1` @@ -316,19 +326,19 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model OpenAI 兼容Anthropic 兼容DashScope -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://dashscope-us.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://dashscope-us.aliyuncs.com/apps/anthropic` API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) -模型 ID`qwen3.6-flash` +模型 ID`qwen3.7-flash` Base URL`https://dashscope-us.aliyuncs.com/api/v1` @@ -558,92 +568,20 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/a [ -kimi-k2.7-code +kimi/kimi-k3 -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/kimi-k2.7-code) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/kimi%2Fkimi-k3?serviceSite=asia-pacific-china) -华北2（北京）新加坡德国（法兰克福）美国（弗吉尼亚） +华北2（北京） -OpenAI 兼容Anthropic 兼容DashScope +OpenAI 兼容 -模型 ID`kimi-k2.7-code` +模型 ID`kimi/kimi-k3` Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - [ glm-5.2 @@ -888,7 +826,7 @@ qwen3.5-omni-plus 华北2（北京）新加坡 -OpenAI 兼容Anthropic 兼容DashScope +OpenAI 兼容 模型 ID`qwen3.5-omni-plus` @@ -896,19 +834,7 @@ Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?t API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`qwen3.5-omni-plus` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -模型 ID`qwen3.5-omni-plus` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope +OpenAI 兼容 模型 ID`qwen3.5-omni-plus` @@ -916,106 +842,22 @@ Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-south API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) -模型 ID`qwen3.5-omni-plus` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`qwen3.5-omni-plus` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - [ -kimi-k2.7-code +kimi/kimi-k3 -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/kimi-k2.7-code) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/kimi%2Fkimi-k3?serviceSite=asia-pacific-china) -华北2（北京）新加坡德国（法兰克福）美国（弗吉尼亚） +华北2（北京） -OpenAI 兼容Anthropic 兼容DashScope +OpenAI 兼容 -模型 ID`kimi-k2.7-code` +模型 ID`kimi/kimi-k3` Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - ### 生成 通过文本或图片生成图像与视频，支持编辑、参考与高分辨率输出 @@ -1024,41 +866,41 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/a [ -wan2.7-image-pro +qwen-image-3.0-pro -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/wan2.7-image-pro) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen-image-3.0-pro) 华北2（北京）新加坡 -模型 ID`wan2.7-image-pro` +模型 ID`qwen-image-3.0-pro` -Request URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` +Request URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`wan2.7-image-pro` +模型 ID`qwen-image-3.0-pro` -Request URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` +Request URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) [ -qwen-image-2.0-pro +wan2.7-image-pro -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen-image-2.0-pro) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/wan2.7-image-pro) 华北2（北京）新加坡 -模型 ID`qwen-image-2.0-pro` +模型 ID`wan2.7-image-pro` -Request URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` +Request URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`qwen-image-2.0-pro` +模型 ID`wan2.7-image-pro` -Request URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` +Request URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) @@ -1258,18 +1100,24 @@ API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api- [ -cosyvoice-v3.5-plus +qwen-audio-3.0-tts-plus -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/cosyvoice-v3.5-plus) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen-audio-3.0-tts-plus) -华北2（北京） +华北2（北京）新加坡 -模型 ID`cosyvoice-v3.5-plus` +模型 ID`qwen-audio-3.0-tts-plus` Request URL`wss://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api-ws/v1/inference` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) +模型 ID`qwen-audio-3.0-tts-plus` + +Request URL`wss://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api-ws/v1/inference` + +API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) + [ MiniMax/speech-2.8-hd @@ -1426,24 +1274,18 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=mod [ -qwen3.5-omni-plus-realtime +qwen-audio-3.0-realtime-plus -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.5-omni-plus-realtime) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen-audio-3.0-realtime-plus) -华北2（北京）新加坡 +华北2（北京） -模型 ID`qwen3.5-omni-plus-realtime` +模型 ID`qwen-audio-3.0-realtime-plus` Request URL`wss://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`qwen3.5-omni-plus-realtime` - -Request URL`wss://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - [ qwen3.5-omni-plus diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md index 8e375ee1..cbad05cc 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md @@ -171,8 +171,6 @@ **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -187,48 +185,36 @@ qwen3.7-max > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 30,000 5,000,000 qwen3.7-max-2026-06-08 -中国内地 - 600 1,000,000 qwen3.7-max-2026-05-20 -中国内地 - 600 1,000,000 qwen3.7-max-preview -中国内地 - 60 500,000 qwen3.7-max-2026-05-17 -中国内地 - 60 500,000 qwen3.6-max-preview -中国内地 - 600 1,000,000 @@ -237,32 +223,24 @@ qwen3-max > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 30,000 5,000,000 qwen3-max-2026-01-23 -中国内地 - 600 1,000,000 qwen3-max-2025-09-23 -中国内地 - 60 100,000 qwen3-max-preview -中国内地 - 600 1,000,000 @@ -271,24 +249,18 @@ qwen-max > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,200 1,000,000 qwen3.7-plus -中国内地 - 30,000 5,000,000 qwen3.7-plus-2026-05-26 -中国内地 - 600 1,000,000 @@ -297,15 +269,25 @@ qwen3.6-plus > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 30,000 5,000,000 qwen3.6-plus-2026-04-02 -中国内地 +600 + +1,000,000 + +qwen3.7-flash + +> 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 + +30,000 + +5,000,000 + +qwen3.7-flash-2026-07-15 600 @@ -315,16 +297,12 @@ qwen3.6-flash > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 30,000 10,000,000 qwen3.6-flash-2026-04-16 -中国内地 - 600 1,000,000 @@ -333,24 +311,18 @@ qwen3.5-plus > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 30,000 5,000,000 qwen3.5-plus-2026-04-20 -中国内地 - 600 1,000,000 qwen3.5-plus-2026-02-15 -中国内地 - 600 1,000,000 @@ -359,8 +331,6 @@ qwen-plus > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 30,000 5,000,000 @@ -369,24 +339,18 @@ qwen-plus-latest > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 15,000 1,200,000 qwen-plus-2025-12-01 -中国内地 - 120 1,000,000 qwen-plus-2025-09-11 -中国内地 - 60 1,000,000 @@ -395,8 +359,6 @@ qwen-plus-2025-07-28 （qwen-plus-0728） -中国内地 - 60 1,000,000 @@ -405,8 +367,6 @@ qwen-plus-2025-07-14 （qwen-plus-0714） -中国内地 - 60 100,000 @@ -415,8 +375,6 @@ qwen-plus-2025-04-28 （qwen-plus-0428） -中国内地 - 60 1,000,000 @@ -425,8 +383,6 @@ qwen-plus-2025-01-25 （qwen-plus-0125） -中国内地 - 60 150,000 @@ -435,8 +391,6 @@ qwen-plus-2025-01-12 （qwen-plus-0112） -中国内地 - 60 150,000 @@ -445,8 +399,6 @@ qwen-plus-2024-12-20 （qwen-plus-1220） -中国内地 - 60 150,000 @@ -455,16 +407,12 @@ qwen3.5-flash > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 30,000 10,000,000 qwen3.5-flash-2026-02-23 -中国内地 - 600 1,000,000 @@ -473,16 +421,12 @@ qwen-flash > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 30,000 10,000,000 qwen-flash-2025-07-28 -中国内地 - 60 1,000,000 @@ -491,8 +435,6 @@ qwen-turbo > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,200 5,000,000 @@ -501,8 +443,6 @@ qwq-plus > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 600 1,000,000 @@ -511,8 +451,6 @@ qwen-long > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,200 3,000,000 @@ -521,8 +459,6 @@ qwen-long-latest > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,200 60,000 @@ -531,8 +467,6 @@ qwen-long-2025-01-25 (qwen-long-0125) -中国内地 - 3 7,500 @@ -665,6 +599,14 @@ qwen3.6-flash-2026-04-16 1,000,000 +qwen3.6-flash-us + +美国 + +15,000 + +5,000,000 + qwen3.5-plus 全球 @@ -899,6 +841,22 @@ qwen3.6-plus-2026-04-02 1,000,000 +qwen3.7-flash + +国际 + +15,000 + +5,000,000 + +qwen3.7-flash-2026-07-15 + +国际 + +60 + +1,000,000 + qwen3.6-flash 国际 @@ -1393,8 +1351,6 @@ qwen3.6-flash-2026-04-16 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -1409,24 +1365,18 @@ qwen3-vl-plus > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 3,000 5,000,000 qwen3-vl-plus-2025-12-19 -中国内地 - 60 100,000 qwen3-vl-plus-2025-09-23 -中国内地 - 60 100,000 @@ -1435,24 +1385,18 @@ qwen3-vl-flash > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 3,000 5,000,000 qwen3-vl-flash-2026-01-22 -中国内地 - 60 100,000 qwen3-vl-flash-2025-10-15 -中国内地 - 60 100,000 @@ -1461,8 +1405,6 @@ qwen-vl-max > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,200 1,000,000 @@ -1471,24 +1413,18 @@ qwen-vl-plus > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,200 1,000,000 qvq-max -中国内地 - 60 100,000 qvq-plus -中国内地 - 60 100,000 @@ -1731,8 +1667,6 @@ qwen3-vl-flash-2025-10-15 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -1745,56 +1679,42 @@ qwen3-vl-flash-2025-10-15 qwen3.5-omni-plus -中国内地 - 60 100,000 qwen3.5-omni-plus-2026-03-15 -中国内地 - 60 100,000 qwen3.5-omni-flash -中国内地 - 60 100,000 qwen3.5-omni-flash-2026-03-15 -中国内地 - 60 100,000 qwen3-omni-flash -中国内地 - 60 100,000 qwen3-omni-flash-2025-12-01 -中国内地 - 60 100,000 qwen3-omni-flash-2025-09-15 -中国内地 - 60 100,000 @@ -1803,16 +1723,12 @@ qwen-omni-turbo > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 60 100,000 qwen-omni-turbo-latest -中国内地 - 60 100,000 @@ -1821,8 +1737,6 @@ qwen-omni-turbo-2025-03-26 （qwen-omni-turbo-0326） -中国内地 - 60 100,000 @@ -1831,8 +1745,6 @@ qwen-omni-turbo-2025-01-19 （qwen-omni-turbo-0119） -中国内地 - 60 100,000 @@ -1939,8 +1851,6 @@ qwen-omni-turbo-2025-03-26 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -1953,72 +1863,54 @@ qwen-omni-turbo-2025-03-26 qwen3.5-omni-plus-realtime -中国内地 - 60 100,000 qwen3.5-omni-plus-realtime-2026-03-15 -中国内地 - 60 100,000 qwen3.5-omni-flash-realtime -中国内地 - 60 100,000 qwen3.5-omni-flash-realtime-2026-03-15 -中国内地 - 60 100,000 qwen3-omni-flash-realtime -中国内地 - 60 100,000 qwen3-omni-flash-realtime-2025-12-01 -中国内地 - 60 100,000 qwen3-omni-flash-realtime-2025-09-15 -中国内地 - 60 100,000 qwen-omni-turbo-realtime-latest -中国内地 - 60 100,000 qwen-omni-turbo-realtime-2025-05-08 -中国内地 - 60 100,000 @@ -2125,8 +2017,6 @@ qwen-omni-turbo-realtime**\-**2025-05-08 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2139,8 +2029,6 @@ qwen-omni-turbo-realtime**\-**2025-05-08 qwen3.5-ocr -中国内地 - 6,000 30,000,000 @@ -2149,8 +2037,6 @@ qwen-vl-ocr > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 600 6,000,000 @@ -2159,40 +2045,30 @@ qwen-vl-ocr-latest > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - -1,200 +6,000 -6,000,000 +30,000,000 qwen-vl-ocr-2025-11-20 -中国内地 - -1,200 +6,000 -6,000,000 +30,000,000 qwen-vl-ocr-2025-08-28 -中国内地 - 600 6,000,000 qwen-vl-ocr-2025-04-13 -中国内地 - 600 6,000,000 qwen-vl-ocr-2024-10-28 -中国内地 - 600 6,000,000 @@ -2299,8 +2175,6 @@ qwen-vl-ocr-2025-11-20 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2313,16 +2187,12 @@ qwen-vl-ocr-2025-11-20 qwen-audio-turbo -中国内地 - 120 100,000 qwen-audio-turbo-latest -中国内地 - 60 100,000 @@ -2333,8 +2203,6 @@ qwen-audio-turbo-latest **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2347,16 +2215,12 @@ qwen-audio-turbo-latest qwen-math-plus -中国内地 - 1,200 1,000,000 qwen-math-plus-latest -中国内地 - 1,200 1,000,000 @@ -2365,8 +2229,6 @@ qwen-math-plus-2024-09-19 （qwen-math-plus-0919） -中国内地 - 60 100,000 @@ -2375,16 +2237,12 @@ qwen-math-plus-2024-08-16 （qwen-math-plus-0816） -中国内地 - 10 20,000 qwen-math-turbo -中国内地 - 1200 1,000,000 @@ -2395,8 +2253,6 @@ qwen-math-turbo **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2409,56 +2265,42 @@ qwen-math-turbo qwen3-coder-plus -中国内地 - 5,000 5,000,000 qwen3-coder-plus-2025-09-23 -中国内地 - 60 1,000,000 qwen3-coder-plus-2025-07-22 -中国内地 - 60 1,000,000 qwen3-coder-flash -中国内地 - 5,000 5,000,000 qwen3-coder-flash-2025-07-28 -中国内地 - 60 1,000,000 qwen-coder-plus -中国内地 - 1,200 1,000,000 qwen-coder-turbo -中国内地 - 1,200 1,000,000 @@ -2637,8 +2479,6 @@ qwen3-coder-flash-2025-07-28 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2651,32 +2491,24 @@ qwen3-coder-flash-2025-07-28 qwen-mt-plus -中国内地 - 60 25,000 qwen-mt-flash -中国内地 - 60 35,000 qwen-mt-lite -中国内地 - 60 100,000 qwen-mt-turbo -中国内地 - 60 35,000 @@ -2823,8 +2655,6 @@ qwen-mt-lite **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2837,8 +2667,6 @@ qwen-mt-lite qwen-doc-turbo -中国内地 - 600 3,000,000 @@ -2849,8 +2677,6 @@ qwen-doc-turbo **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2863,8 +2689,6 @@ qwen-doc-turbo qwen-deep-research -中国内地 - 120 1,200,000 @@ -2875,8 +2699,6 @@ qwen-deep-research **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2889,16 +2711,12 @@ qwen-deep-research tongyi-xiaomi-analysis-flash -中国内地 - 600 1,000,000 tongyi-xiaomi-analysis-pro -中国内地 - 600 1,000,000 @@ -2911,8 +2729,6 @@ tongyi-xiaomi-analysis-pro **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -2925,136 +2741,102 @@ tongyi-xiaomi-analysis-pro qwen3.6-35b-a3b -中国内地 - 600 1,000,000 qwen3.6-27b -中国内地 - 600 1,000,000 qwen3.5-397b-a17b -中国内地 - 600 1,000,000 qwen3.5-122b-a10b -中国内地 - 600 1,000,000 qwen3.5-27b -中国内地 - 600 1,000,000 qwen3.5-35b-a3b -中国内地 - 600 1,000,000 qwen3-next-80b-a3b-thinking -中国内地 - 600 1,000,000 qwen3-next-80b-a3b-instruct -中国内地 - 600 1,000,000 qwen3-235b-a22b-thinking-2507 -中国内地 - 600 1,000,000 qwen3-235b-a22b-instruct-2507 -中国内地 - 600 1,000,000 qwen3-30b-a3b-thinking-2507 -中国内地 - 600 1,000,000 qwen3-30b-a3b-instruct-2507 -中国内地 - 600 1,000,000 qwen3-235b-a22b -中国内地 - 600 1,000,000 qwen3-30b-a3b -中国内地 - 600 1,000,000 qwen3-32b -中国内地 - 2400 1,000,000 qwen3-14b -中国内地 - 600 1,000,000 qwen3-8b -中国内地 - 600 1,000,000 @@ -3505,8 +3287,6 @@ qwen3-8b **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -3519,64 +3299,48 @@ qwen3-8b qwen3-vl-32b-thinking -中国内地 - 600 1,000,000 qwen3-vl-32b-instruct -中国内地 - 600 1,000,000 qwen3-vl-30b-a3b-thinking -中国内地 - 600 1,000,000 qwen3-vl-30b-a3b-instruct -中国内地 - 600 1,000,000 qwen3-vl-8b-thinking -中国内地 - 600 1,000,000 qwen3-vl-8b-instruct -中国内地 - 600 1,000,000 qwen3-vl-235b-a22b-thinking -中国内地 - 60 100,000 qwen3-vl-235b-a22b-instruct -中国内地 - 60 100,000 @@ -3827,8 +3591,6 @@ qwen3-vl-8b-instruct **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -3841,8 +3603,6 @@ qwen3-vl-8b-instruct qwen2.5-omni-7b -中国内地 - 60 100,000 @@ -3877,8 +3637,6 @@ qwen2.5-omni-7b **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -3891,8 +3649,6 @@ qwen2.5-omni-7b qwen3-omni-30b-a3b-captioner -中国内地 - 60 100,000 @@ -3927,8 +3683,6 @@ qwen3-omni-30b-a3b-captioner **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -3945,8 +3699,6 @@ qwen3-omni-30b-a3b-captioner **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -3959,24 +3711,18 @@ qwen3-omni-30b-a3b-captioner qwen3-coder-next -中国内地 - 600 1,000,000 qwen3-coder-480b-a35b-instruct -中国内地 - 600 1,000,000 qwen3-coder-30b-a3b-instruct -中国内地 - 600 1,000,000 @@ -4101,8 +3847,6 @@ qwen3-coder-next **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -4115,16 +3859,12 @@ qwen3-coder-next deepseek-v4-pro -中国内地 - 15,000 1,200,000 deepseek-v4-flash -中国内地 - 15,000 1,200,000 @@ -4133,32 +3873,24 @@ deepseek-v3.2 > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 15,000 1,200,000 deepseek-v3.2-exp -中国内地 - 15,000 1,200,000 deepseek-v3.1 -中国内地 - 15,000 1,200,000 deepseek-r1-0528 -中国内地 - 60 100,000 @@ -4167,8 +3899,6 @@ deepseek-r1 > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 15,000 1,200,000 @@ -4177,56 +3907,42 @@ deepseek-v3 > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 15,000 1,200,000 deepseek-r1-distill-qwen-7b -中国内地 - 15,000 1,200,000 deepseek-r1-distill-qwen-14b -中国内地 - 15,000 1,200,000 deepseek-r1-distill-qwen-32b -中国内地 - 15,000 1,200,000 deepseek-r1-distill-qwen-1.5b -中国内地 - 60 100,000 deepseek-r1-distill-llama-8b -中国内地 - 60 100,000 deepseek-r1-distill-llama-70b -中国内地 - 60 100,000 @@ -4405,8 +4121,6 @@ deepseek-v4-flash **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -4419,32 +4133,24 @@ deepseek-v4-flash siliconflow/deepseek-v3.2 -中国内地 - 500 500,000 siliconflow/deepseek-v3.1-terminus -中国内地 - 500 500,000 siliconflow/deepseek-r1-0528 -中国内地 - 500 500,000 siliconflow/deepseek-v3-0324 -中国内地 - 500 500,000 @@ -4455,8 +4161,6 @@ siliconflow/deepseek-v3-0324 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -4469,40 +4173,30 @@ siliconflow/deepseek-v3-0324 vanchin/deepseek-v3.2-think -中国内地 - 30 600,000 vanchin/deepseek-v3.1-terminus -中国内地 - 500 1,000,000 vanchin/deepseek-r1 -中国内地 - 500 1,000,000 vanchin/deepseek-v3 -中国内地 - 500 1,000,000 vanchin/deepseek-ocr -中国内地 - 500 1,000,000 @@ -4513,8 +4207,6 @@ vanchin/deepseek-ocr **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -4527,40 +4219,30 @@ vanchin/deepseek-ocr kimi-k2.7-code -中国内地 - 500 1,000,000 kimi-k2.6 -中国内地 - 500 1,000,000 kimi-k2.5 -中国内地 - 500 1,000,000 kimi-k2-thinking -中国内地 - 500 1,000,000 Moonshot-Kimi-K2-Instruct -中国内地 - 500 1,000,000 @@ -4683,8 +4365,6 @@ kimi-k2.7-code **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -4695,38 +4375,30 @@ kimi-k2.7-code > **含输入与输出Token** -kimi/kimi-k2.7-code-highspeed - -中国内地 +kimi/kimi-k3 500 -> 同一个阿里云百炼API Key 下，在 4 个模型中共享 500 RPM 限流配额。即这 4 个模型的每分钟请求总数加起来不能超过 500。 +> 同一个阿里云百炼API Key 下，在 5 个模型中共享 500 RPM 限流配额。即这 5 个模型的每分钟请求总数加起来不能超过 500。 3,000,000 -> 同一个阿里云百炼API Key 下，在 4 个模型中共享 3000000 TPM 限流配额。即这 4 个模型的每分钟 Token 消耗总数加起来不能超过 3000000。 +> 同一个阿里云百炼API Key 下，在 5 个模型中共享 3000000 TPM 限流配额。即这 5 个模型的每分钟 Token 消耗总数加起来不能超过 3000000。 -kimi/kimi-k2.7-code +kimi/kimi-k2.7-code-highspeed -中国内地 +kimi/kimi-k2.7-code kimi/kimi-k2.6 -中国内地 - kimi/kimi-k2.5 -中国内地 - ### **GLM** ## **华北2（北京）** **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -4739,56 +4411,42 @@ kimi/kimi-k2.5 glm-5.2 -中国内地 - 500 2,000,000 glm-5.1 -中国内地 - 500 1,000,000 glm-5 -中国内地 - 500 1,000,000 glm-4.7 -中国内地 - 500 1,000,000 glm-4.6 -中国内地 - 60 1,000,000 glm-4.5 -中国内地 - 60 1,000,000 glm-4.5-air -中国内地 - 60 1,000,000 @@ -4907,7 +4565,15 @@ glm-5.1 glm-5.2 -全球 +国际 + +500 + +1,000,000 + +glm-5.1 + +国际 500 @@ -4919,8 +4585,6 @@ glm-5.2 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -4933,24 +4597,18 @@ glm-5.2 ZHIPU/GLM-5.2 -中国内地 - 200 3,000,000 ZHIPU/GLM-5.1 -中国内地 - 200 -10,000,000 +3,000,000 ZHIPU/GLM-5 -中国内地 - 200 3,000,000 @@ -4961,8 +4619,6 @@ ZHIPU/GLM-5 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -4975,16 +4631,12 @@ ZHIPU/GLM-5 MiniMax-M2.5 -中国内地 - 500 1,000,000 MiniMax-M2.1 -中国内地 - 500 1,000,000 @@ -4995,8 +4647,6 @@ MiniMax-M2.1 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -5009,32 +4659,24 @@ MiniMax-M2.1 MiniMax/MiniMax-M3 -中国内地 - 500 20,000,000 MiniMax/MiniMax-M2.7 -中国内地 - 500 20,000,000 MiniMax/MiniMax-M2.5 -中国内地 - 500 20,000,000 MiniMax/MiniMax-M2.1 -中国内地 - 500 20,000,000 @@ -5045,8 +4687,6 @@ MiniMax/MiniMax-M2.1 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -5059,8 +4699,6 @@ MiniMax/MiniMax-M2.1 xiaomi/mimo-v2.5-pro -中国内地 - 100 10,000,000 @@ -5071,8 +4709,6 @@ xiaomi/mimo-v2.5-pro **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -5085,8 +4721,6 @@ xiaomi/mimo-v2.5-pro stepfun/step-3.7-flash -中国内地 - 500 20,000,000 @@ -5099,17 +4733,19 @@ stepfun/step-3.7-flash **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **任务下发接口调用限制** **同时处理中任务数量（并发数）** -qwen-image-2.0-pro +qwen-image-3.0-pro + +1 次/分钟 + +同步接口无限制 -中国内地 +qwen-image-2.0-pro 2 次/分钟 @@ -5117,136 +4753,102 @@ qwen-image-2.0-pro qwen-image-2.0-pro-2026-06-22 -中国内地 - 2 次/分钟 同步接口无限制 qwen-image-2.0-pro-2026-04-22 -中国内地 - 2 次/分钟 同步接口无限制 qwen-image-2.0-pro-2026-03-03 -中国内地 - 2 次/分钟 同步接口无限制 qwen-image-2.0 -中国内地 - 2 次/秒 同步接口无限制 qwen-image-2.0-2026-03-03 -中国内地 - 2 次/秒 同步接口无限制 qwen-image-max -中国内地 - 2 次/分钟 同步接口无限制 qwen-image-max-2025-12-30 -中国内地 - 2 次/分钟 同步接口无限制 qwen-image-plus -中国内地 - 2 次/秒 同步接口无限制 / 异步接口 2 qwen-image-plus-2026-01-09 -中国内地 - 2 次/秒 同步接口无限制 qwen-image -中国内地 - 2 次/秒 同步接口无限制 / 异步接口 2 qwen-image-edit-max -中国内地 - 2 次/分钟 同步接口无限制 qwen-image-edit-max-2026-01-16 -中国内地 - 2 次/分钟 同步接口无限制 qwen-image-edit-plus -中国内地 - 2 次/秒 同步接口无限制 qwen-image-edit-plus-2025-12-15 -中国内地 - 2 次/秒 同步接口无限制 qwen-image-edit-plus-2025-10-30 -中国内地 - 2 次/秒 同步接口无限制 qwen-image-edit -中国内地 - 2 次/秒 同步接口无限制 qwen-mt-image -中国内地 - 1 次/秒 2 @@ -5263,6 +4865,14 @@ qwen-mt-image **同时处理中任务数量（并发数）** +qwen-image-3.0-pro + +国际 + +1 次/分钟 + +同步接口无限制 + qwen-image-2.0-pro 国际 @@ -5273,7 +4883,7 @@ qwen-image-2.0-pro qwen-image-2.0-pro-2026-06-22 -中国内地 +国际 2 次/分钟 @@ -5405,8 +5015,6 @@ qwen-image-edit **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -5415,8 +5023,6 @@ qwen-image-edit z-image-turbo -中国内地 - 2 同步接口无限制 @@ -5447,8 +5053,6 @@ z-image-turbo **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -5457,120 +5061,90 @@ z-image-turbo wan2.7-image-pro -中国内地 - 5 5 wan2.7-image -中国内地 - 5 5 wan2.6-image -中国内地 - 5 5 wan2.6-t2i -中国内地 - 1 5 wan2.5-t2i-preview -中国内地 - 5 5 wan2.2-t2i-plus -中国内地 - 2 2 wan2.2-t2i-flash -中国内地 - 2 2 wanx2.1-t2i-plus -中国内地 - 2 2 wanx2.1-t2i-turbo -中国内地 - 2 2 wanx2.0-t2i-turbo -中国内地 - 2 2 wan2.5-i2i-preview -中国内地 - 5 5 wanx2.1-imageedit -中国内地 - 2 2 wanx-v1 -中国内地 - 2 1 wanx-x-painting -中国内地 - 2 1 wanx-sketch-to-image-lite -中国内地 - 2 1 @@ -5729,8 +5303,6 @@ wan2.6-image **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -5739,72 +5311,54 @@ wan2.6-image shoemodel-v1 -中国内地 - 2 1 wanx-virtualmodel -中国内地 - 2 1 wanx-style-repaint-v1 -中国内地 - 2 2 wanx-poster-generation-v1 -中国内地 - 2 1 virtualmodel-v2 -中国内地 - 2 1 wanx-background-generation-v2 -中国内地 - 2 1 image-instance-segmentation -中国内地 - 2 1 image-erase-completion -中国内地 - 2 1 image-out-painting -中国内地 - 2 10 @@ -5815,8 +5369,6 @@ image-out-painting **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **作业提交接口RPS限制** @@ -5825,24 +5377,18 @@ image-out-painting facechain-facedetect -中国内地 - 5 同步接口无限制 facechain-finetune -中国内地 - 1 1 facechain-generation -中国内地 - 2 1 @@ -5853,8 +5399,6 @@ facechain-generation **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **作业提交接口RPS限制** @@ -5863,16 +5407,12 @@ facechain-generation wordart-texture -中国内地 - 2 1 wordart-semantic -中国内地 - 2 1 @@ -5883,8 +5423,6 @@ wordart-semantic **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **作业提交接口RPS限制** @@ -5893,32 +5431,24 @@ wordart-semantic aitryon -中国内地 - 10 5 aitryon-plus -中国内地 - 10 5 aitryon-parsing-v1 -中国内地 - 10 同步接口无限制 aitryon-refiner -中国内地 - 10 5 @@ -5931,8 +5461,6 @@ aitryon-refiner **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -5941,8 +5469,6 @@ aitryon-refiner kling/kling-v3-omni-image-generation -中国内地 - 5 10 @@ -5951,16 +5477,12 @@ kling/kling-v3-omni-image-generation kling/kling-v3-image-generation -中国内地 - ### **Vidu系列** ## **华北2（北京）** **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每分钟请求数RPM限制** @@ -5969,8 +5491,6 @@ kling/kling-v3-image-generation vidu/vidu-image\_reference2image -中国内地 - 300 5 @@ -5979,36 +5499,24 @@ vidu/vidu-image\_reference2image vidu/viduq3-fast\_reference2image -中国内地 - vidu/viduq2-pro\_reference2image -中国内地 - vidu/viduq2-fast\_reference2image -中国内地 - ## **音乐生成** ## **华北2（北京）** **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** fun-music-preview -中国内地 - 180 fun-music-v1 -中国内地 - 180 ## **语音对话** @@ -6019,8 +5527,6 @@ fun-music-v1 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -6033,16 +5539,12 @@ fun-music-v1 qwen-audio-3.0-realtime-plus -中国内地 - 60 100,000 qwen-audio-3.0-realtime-flash -中国内地 - 60 100,000 @@ -6055,20 +5557,14 @@ qwen-audio-3.0-realtime-flash **模型名称** -**服务部署范围** - **提交作业接口RPS限制** qwen-audio-3.0-tts-plus -中国内地 - 3 qwen-audio-3.0-tts-flash -中国内地 - 3 #### 新加坡 @@ -6099,82 +5595,58 @@ qwen-audio-3.0-tts-flash **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-tts-instruct-flash -中国内地 - 180 qwen3-tts-instruct-flash-2026-01-26 -中国内地 - 180 ##### **千问3-TTS-VD** **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-tts-vd-2026-01-26 -中国内地 - 180 ##### **千问3-TTS-VC** **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-tts-vc-2026-01-22 -中国内地 - 180 ##### 千问3-TTS-Flash **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-tts-flash -中国内地 - 180 qwen3-tts-flash-2025-11-27 -中国内地 - 180 qwen3-tts-flash-2025-09-18 -中国内地 - 10 ##### 千问-TTS **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -6187,24 +5659,16 @@ qwen3-tts-flash-2025-09-18 qwen-tts -中国内地 - 10 100,000 qwen-tts-latest -中国内地 - qwen-tts-2025-05-22 -中国内地 - qwen-tts-2025-04-10 -中国内地 - #### 新加坡 ##### **千问3-TTS-Instruct-Flash** @@ -6289,90 +5753,62 @@ qwen3-tts-flash-2025-09-18 **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-tts-instruct-flash-realtime -中国内地 - 180 qwen3-tts-instruct-flash-realtime-2026-01-22 -中国内地 - 180 ##### 千问3-TTS-VD-Realtime **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-tts-vd-realtime-2026-01-15 -中国内地 - 180 qwen3-tts-vd-realtime-2025-12-16 -中国内地 - ##### 千问3-TTS-VC-Realtime **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-tts-vc-realtime-2026-01-15 -中国内地 - 180 qwen3-tts-vc-realtime-2025-11-27 -中国内地 - ##### 千问3-TTS-Flash-Realtime **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-tts-flash-realtime -中国内地 - 180 qwen3-tts-flash-realtime-2025-11-27 -中国内地 - 180 qwen3-tts-flash-realtime-2025-09-18 -中国内地 - 10 ##### 千问-TTS-Realtime **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -6385,20 +5821,14 @@ qwen3-tts-flash-realtime-2025-09-18 qwen-tts-realtime -中国内地 - 10 100,000 qwen-tts-realtime-latest -中国内地 - qwen-tts-realtime-2025-07-15 -中国内地 - #### 新加坡 ##### **千问3-TTS-Instruct-Flash-Realtime** @@ -6489,14 +5919,10 @@ qwen3-tts-flash-realtime-2025-09-18 **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen-voice-enrollment -中国内地 - 180 #### 新加坡 @@ -6519,14 +5945,10 @@ qwen-voice-enrollment **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen-voice-design -中国内地 - 180 #### 新加坡 @@ -6549,36 +5971,22 @@ qwen-voice-design **模型名称** -**服务部署范围** - **提交作业接口RPS限制** cosyvoice-v3.5-plus -中国内地 - 3 cosyvoice-v3.5-flash -中国内地 - cosyvoice-v3-plus -中国内地 - cosyvoice-v3-flash -中国内地 - cosyvoice-v2 -中国内地 - cosyvoice-v1 -中国内地 - #### 新加坡 **模型名称** @@ -6605,14 +6013,10 @@ Qwen-Audio-TTS/CosyVoice声音复刻/设计共用一个模型，共用限流额 **模型名称** -**服务部署范围** - **提交作业接口RPS限制** voice-enrollment -中国内地 - 10 #### 新加坡 @@ -6635,14 +6039,10 @@ voice-enrollment **模型服务** -**服务部署范围** - **提交作业接口RPS限制** Sambert系列模型 -中国内地 - 20 ## **语音合成（文本转语音）-第三方模型** @@ -6653,8 +6053,6 @@ Sambert系列模型 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每分钟调用次数（RPM）** @@ -6665,32 +6063,24 @@ Sambert系列模型 MiniMax/speech-2.8-hd -中国内地 - 20 20,000 MiniMax/speech-02-hd -中国内地 - 20 20,000 MiniMax/speech-2.8-turbo -中国内地 - 20 20,000 MiniMax/speech-02-turbo -中国内地 - 20 20,000 @@ -6703,8 +6093,6 @@ MiniMax/speech-02-turbo **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -6717,16 +6105,12 @@ MiniMax/speech-02-turbo qwen3-livetranslate-flash -中国内地 - 100 100,000 qwen3-livetranslate-flash-2025-12-01 -中国内地 - #### 新加坡 **模型名称** @@ -6761,8 +6145,6 @@ qwen3-livetranslate-flash-2025-12-01 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -6775,24 +6157,16 @@ qwen3-livetranslate-flash-2025-12-01 qwen3.5-livetranslate-flash-realtime -中国内地 - 10 100,000 qwen3.5-livetranslate-flash-realtime-2026-05-19 -中国内地 - qwen3-livetranslate-flash-realtime -中国内地 - qwen3-livetranslate-flash-realtime-2025-09-22 -中国内地 - #### 新加坡 **模型名称** @@ -6837,42 +6211,28 @@ qwen3-livetranslate-flash-realtime-2025-09-22 **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-asr-flash-filetrans -中国内地 - 100 qwen3-asr-flash-filetrans-2025-11-17 -中国内地 - ##### **千问3-ASR-Flash** **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** qwen3-asr-flash -中国内地 - 100 qwen3-asr-flash-2026-02-10 -中国内地 - qwen3-asr-flash-2025-09-08 -中国内地 - #### 新加坡 ##### 千问3-ASR-Flash-Filetrans @@ -6939,24 +6299,16 @@ qwen3-asr-flash-2025-09-08-us **模型名称** -**服务部署范围** - **每秒钟调用次数（RPS）** qwen3-asr-flash-realtime -中国内地 - 20 qwen3-asr-flash-realtime-2026-02-10 -中国内地 - qwen3-asr-flash-realtime-2025-10-27 -中国内地 - #### 新加坡 **模型名称** @@ -6985,36 +6337,22 @@ qwen3-asr-flash-realtime-2025-10-27 **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** fun-asr -中国内地 - 600 fun-asr-2025-11-07 -中国内地 - fun-asr-2025-08-25 -中国内地 - fun-asr-mtl -中国内地 - fun-asr-mtl-2025-08-25 -中国内地 - fun-asr-flash-2026-06-15 -中国内地 - #### 新加坡 **模型名称** @@ -7065,36 +6403,22 @@ fun-asr-flash-2026-06-15 **模型名称** -**服务部署范围** - **提交作业接口RPS限制** fun-asr-realtime -中国内地 - 20 fun-asr-realtime-2026-02-28 -中国内地 - fun-asr-realtime-2025-11-07 -中国内地 - fun-asr-realtime-2025-09-15 -中国内地 - fun-asr-flash-8k-realtime -中国内地 - fun-asr-flash-8k-realtime-2026-01-28 -中国内地 - #### 新加坡 **模型名称** @@ -7119,44 +6443,28 @@ fun-asr-realtime-2025-11-07 **模型名称** -**服务部署范围** - **提交作业接口RPS限制** paraformer-realtime-v2 -中国内地 - 20 paraformer-realtime-v1 -中国内地 - paraformer-realtime-8k-v2 -中国内地 - paraformer-realtime-8k-v1 -中国内地 - **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** paraformer-v2 -中国内地 - 1,200 **模型名称** -**服务部署范围** - **每分钟调用次数（RPM）** **每分钟消耗Token数（TPM）** @@ -7165,40 +6473,30 @@ paraformer-v2 paraformer-v1 -中国内地 - 600 6,000,000 paraformer-mtl-v1 -中国内地 - 600 6,000,000 **模型名称** -**服务部署范围** - **提交作业接口RPS限制** **同时处理中任务数量（并发数）** paraformer-8k-v2 -中国内地 - 20 100 paraformer-8k-v1 -中国内地 - 10 500 @@ -7211,8 +6509,6 @@ paraformer-8k-v1 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -7221,57 +6517,43 @@ paraformer-8k-v1 happyhorse-1.1-t2v -中国内地 - -10 +5 5 happyhorse-1.1-i2v -中国内地 - -10 +5 5 happyhorse-1.1-r2v -中国内地 - -10 +5 5 happyhorse-1.0-t2v -中国内地 - -10 +5 5 happyhorse-1.0-i2v -中国内地 - -10 +5 5 happyhorse-1.0-r2v -中国内地 - -10 +5 5 happyhorse-1.0-video-edit -中国内地 - -10 +5 5 @@ -7291,7 +6573,7 @@ happyhorse-1.1-t2v 全球 -10 +5 5 @@ -7299,7 +6581,7 @@ happyhorse-1.1-i2v 全球 -10 +5 5 @@ -7307,7 +6589,7 @@ happyhorse-1.1-r2v 全球 -10 +5 5 @@ -7315,7 +6597,7 @@ happyhorse-1.0-t2v 全球 -10 +5 5 @@ -7323,7 +6605,7 @@ happyhorse-1.0-i2v 全球 -10 +5 5 @@ -7331,7 +6613,7 @@ happyhorse-1.0-r2v 全球 -10 +5 5 @@ -7339,7 +6621,7 @@ happyhorse-1.0-video-edit 全球 -10 +5 5 @@ -7359,7 +6641,7 @@ happyhorse-1.1-t2v 国际 -10 +5 5 @@ -7367,7 +6649,7 @@ happyhorse-1.1-i2v 国际 -10 +5 5 @@ -7375,7 +6657,7 @@ happyhorse-1.1-r2v 国际 -10 +5 5 @@ -7383,7 +6665,7 @@ happyhorse-1.0-t2v 国际 -10 +5 5 @@ -7391,7 +6673,7 @@ happyhorse-1.0-i2v 国际 -10 +5 5 @@ -7399,7 +6681,7 @@ happyhorse-1.0-r2v 国际 -10 +5 5 @@ -7407,7 +6689,7 @@ happyhorse-1.0-video-edit 国际 -10 +5 5 @@ -7427,7 +6709,7 @@ happyhorse-1.1-t2v 全球 -10 +5 5 @@ -7435,7 +6717,7 @@ happyhorse-1.1-i2v 全球 -10 +5 5 @@ -7443,7 +6725,7 @@ happyhorse-1.1-r2v 全球 -10 +5 5 @@ -7451,7 +6733,7 @@ happyhorse-1.0-t2v 全球 -10 +5 5 @@ -7459,7 +6741,7 @@ happyhorse-1.0-i2v 全球 -10 +5 5 @@ -7467,7 +6749,7 @@ happyhorse-1.0-r2v 全球 -10 +5 5 @@ -7475,7 +6757,51 @@ happyhorse-1.0-video-edit 全球 -10 +5 + +5 + +## **日本（东京）** + +**模型名称** + +**服务部署范围** + +**限流值（任何一个值超出即触发限流）** + +**任务提交接口RPS限制** + +**正在处理中的任务数（并发）** + +happyhorse-1.1-t2v + +全球 + +5 + +5 + +happyhorse-1.1-i2v + +全球 + +5 + +5 + +happyhorse-1.1-r2v + +全球 + +5 + +5 + +happyhorse-1.0-video-edit + +全球 + +5 5 @@ -7485,8 +6811,6 @@ happyhorse-1.0-video-edit **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -7495,232 +6819,174 @@ happyhorse-1.0-video-edit wan2.7-r2v-2026-06-12 -中国内地 - 5 5 wan2.7-t2v-2026-06-12 -中国内地 - 5 5 wan2.7-t2v-2026-04-25 -中国内地 - 5 5 wan2.7-t2v -中国内地 - 5 5 wan2.6-t2v -中国内地 - 5 5 wan2.5-t2v-preview -中国内地 - 5 5 wan2.2-t2v-plus -中国内地 - 2 2 wanx2.1-t2v-turbo -中国内地 - 2 2 wanx2.1-t2v-plus -中国内地 - 2 2 wan2.7-i2v-2026-04-25 -中国内地 - 5 5 wan2.7-i2v -中国内地 - 5 5 wan2.6-i2v-flash -中国内地 - 5 5 wan2.6-i2v -中国内地 - 5 5 wan2.5-i2v-preview -中国内地 - 5 5 wan2.2-i2v-flash -中国内地 - 2 2 wan2.2-i2v-plus -中国内地 - 2 2 wanx2.1-i2v-turbo -中国内地 - 2 2 wanx2.1-i2v-plus -中国内地 - 2 2 wan2.2-kf2v-flash -中国内地 - 2 2 wanx2.1-kf2v-plus -中国内地 - 2 2 wanx2.1-vace-plus -中国内地 - 2 2 wan2.7-videoedit -中国内地 - 5 5 wan2.7-r2v -中国内地 - 5 5 wan2.6-r2v-flash -中国内地 - 5 5 wan2.6-r2v -中国内地 - 5 5 wan2.2-s2v-detect -中国内地 - 5 同步接口无限制 wan2.2-s2v -中国内地 - 5 1 wan2.2-animate-move -中国内地 - 5 1 wan2.2-animate-mix -中国内地 - 5 1 @@ -8031,24 +7297,18 @@ wan2.6-r2v **模型名称** -**服务部署范围** - **任务下发接口RPS限制** **同时处理中任务数量** animate-anyone-detect-gen2 -中国内地 - 5 同步接口无限制 animate-anyone-template-gen2 -中国内地 - 5 1 @@ -8057,8 +7317,6 @@ animate-anyone-template-gen2 animate-anyone-gen2 -中国内地 - 5 1 @@ -8067,16 +7325,12 @@ animate-anyone-gen2 animate-anyone-detect -中国内地 - 5 1算力单元支持2并发 animate-anyone -中国内地 - 5 1算力单元支持1并发 @@ -8087,24 +7341,18 @@ animate-anyone **模型名称** -**服务部署范围** - **任务下发接口RPS限制** **同时处理中任务数量** emo-detect-v1 -中国内地 - 5 同步接口无限制 emo-v1 -中国内地 - 5 1 @@ -8117,24 +7365,18 @@ emo-v1 **模型名称** -**服务部署范围** - **任务下发接口RPS限制** **同时处理中任务数量** liveportrait-detect -中国内地 - 5 同步接口无限制 liveportrait -中国内地 - 5 1 @@ -8147,16 +7389,12 @@ liveportrait **模型名称** -**服务部署范围** - **任务下发接口RPS限制** **同时处理中任务数量** videoretalk -中国内地 - 1 1 @@ -8169,24 +7407,18 @@ videoretalk **模型名称** -**服务部署范围** - **任务下发接口RPS限制** **同时处理中任务数量** emoji-detect-v1 -中国内地 - 1 同步接口无限制 emoji-v1 -中国内地 - 1 1 @@ -8199,16 +7431,12 @@ emoji-v1 **模型名称** -**服务部署范围** - **任务下发接口RPS限制** **同时处理中任务数量** video-style-transform -中国内地 - 20 2 @@ -8223,8 +7451,6 @@ video-style-transform **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每分钟任务下发接口RPM限制** @@ -8233,8 +7459,6 @@ video-style-transform pixverse/pixverse-lipsync -中国内地 - 300 5 @@ -8243,20 +7467,14 @@ pixverse/pixverse-lipsync pixverse/pixverse-motioncontrol -中国内地 - 300 pixverse/pixverse-upscale -中国内地 - 300 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -8265,8 +7483,6 @@ pixverse/pixverse-upscale pixverse/pixverse-c1-t2v -中国内地 - 5 5 @@ -8275,20 +7491,12 @@ pixverse/pixverse-c1-t2v pixverse/pixverse-c1-it2v -中国内地 - pixverse/pixverse-c1-kf2v -中国内地 - pixverse/pixverse-c1-r2v -中国内地 - pixverse/pixverse-v6-t2v -中国内地 - 5 5 @@ -8297,20 +7505,12 @@ pixverse/pixverse-v6-t2v pixverse/pixverse-v6-it2v -中国内地 - pixverse/pixverse-v6-kf2v -中国内地 - pixverse/pixverse-v6-r2v -中国内地 - pixverse/pixverse-v5.6-t2v -中国内地 - 5 5 @@ -8319,24 +7519,16 @@ pixverse/pixverse-v5.6-t2v pixverse/pixverse-v5.6-it2v -中国内地 - pixverse/pixverse-v5.6-kf2v -中国内地 - pixverse/pixverse-v5.6-r2v -中国内地 - ### **可灵系列** ## **华北2（北京）** **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -8345,8 +7537,6 @@ pixverse/pixverse-v5.6-r2v kling/kling-v3-omni-video-generation -中国内地 - 5 10 @@ -8355,16 +7545,12 @@ kling/kling-v3-omni-video-generation kling/kling-v3-video-generation -中国内地 - ### **Vidu系列** ## **华北2（北京）** **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每秒钟任务下发接口RPS限制** @@ -8373,8 +7559,6 @@ kling/kling-v3-video-generation vidu/viduq3-ad\_reference2video -中国内地 - 5 5 @@ -8383,116 +7567,78 @@ vidu/viduq3-ad\_reference2video vidu/viduq3-drama\_reference2video -中国内地 - 5 vidu/viduq3-pro-fast\_img2video -中国内地 - 5 vidu/viduq3-turbo\_text2video -中国内地 - 5 vidu/viduq3-pro\_text2video -中国内地 - 5 vidu/viduq2\_text2video -中国内地 - 5 vidu/viduq3-turbo\_img2video -中国内地 - 5 vidu/viduq3-pro\_img2video -中国内地 - 5 vidu/viduq2-turbo\_img2video -中国内地 - 5 vidu/viduq2-pro\_img2video -中国内地 - 5 vidu/viduq2-pro-fast\_img2video -中国内地 - 5 vidu/viduq3-turbo\_start-end2video -中国内地 - 5 vidu/viduq3-pro\_start-end2video -中国内地 - 5 vidu/viduq2-turbo\_start-end2video -中国内地 - 5 vidu/viduq2-pro\_start-end2video -中国内地 - 5 vidu/viduq3-mix\_reference2video -中国内地 - 5 vidu/viduq3\_reference2video -中国内地 - 5 vidu/viduq3-turbo\_reference2video -中国内地 - 5 vidu/viduq2-pro\_reference2video -中国内地 - 5 vidu/viduq2\_reference2video -中国内地 - 5 ## **3D模型生成-第三方模型** @@ -8503,18 +7649,14 @@ vidu/viduq2\_reference2video **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** -**每秒钟任务下发接口RPS限制** +**每分钟任务下发接口RPM限制** **同时处理中任务数量（并发数）** Tripo/Tripo-H3.1 -中国内地 - 5 10 @@ -8523,8 +7665,6 @@ Tripo/Tripo-H3.1 Tripo/Tripo-P1.0 -中国内地 - 5 ## **向量模型** @@ -8535,8 +7675,6 @@ Tripo/Tripo-P1.0 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** **每分钟调用次数（RPM）** @@ -8547,8 +7685,6 @@ Tripo/Tripo-P1.0 qwen3.7-text-embedding -中国内地 - 1,800 1,000,000 @@ -8557,8 +7693,6 @@ text-embedding-v1 > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,800 1,200,000 @@ -8567,8 +7701,6 @@ text-embedding-v2 > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,800 1,200,000 @@ -8577,8 +7709,6 @@ text-embedding-v3 > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,800 1,200,000 @@ -8587,16 +7717,12 @@ text-embedding-v4 > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 -中国内地 - 1,800 1,200,000 text-embedding-async-v1 -中国内地 - 60 当前用户在系统通用文本向量异步作业排队中和运行中的作业数量不超过50个。 @@ -8605,8 +7731,6 @@ text-embedding-async-v1 text-embedding-async-v2 -中国内地 - 60 当前用户在系统通用文本向量异步作业排队中和运行中的作业数量不超过50个。 @@ -8651,8 +7775,6 @@ text-embedding-v3 **模型名称** -**服务部署范围** - **限流条件** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -8665,56 +7787,42 @@ text-embedding-v3 qwen3-vl-embedding -中国内地 - 2,400 1,200,000 qwen2.5-vl-embedding -中国内地 - 1,200 600,000 tongyi-embedding-vision-plus -中国内地 - 600 200,000 tongyi-embedding-vision-flash -中国内地 - 600 200,000 tongyi-embedding-vision-flash-2026-03-06 -中国内地 - 1,200 9,600,000 tongyi-embedding-vision-plus-2026-03-06 -中国内地 - 1,200 9,600,000 multimodal-embedding-v1 -中国内地 - 120 1,000,000 @@ -8727,8 +7835,6 @@ multimodal-embedding-v1 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -8741,24 +7847,18 @@ multimodal-embedding-v1 qwen3-rerank -中国内地 - 5,400 5,000,000,000 qwen3-vl-rerank -中国内地 - 600 9,000,000 gte-rerank-v2 -中国内地 - 5,040 4,980,000,000 @@ -8803,8 +7903,6 @@ gte-rerank-v2 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -8817,8 +7915,6 @@ gte-rerank-v2 farui-plus -中国内地 - 240 1,000,000 @@ -8829,8 +7925,6 @@ farui-plus **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -8843,8 +7937,6 @@ farui-plus tongyi-intent-detect-v3 -中国内地 - 1,200 1,000,000 @@ -8855,8 +7947,6 @@ tongyi-intent-detect-v3 **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -8869,24 +7959,18 @@ tongyi-intent-detect-v3 qwen-plus-character -中国内地 - 120 500,000 qwen-flash-character -中国内地 - 120 500,000 qwen-flash-character-2026-02-26 -中国内地 - 120 500,000 @@ -9009,8 +8093,6 @@ qwen-plus-character **模型名称** -**服务部署范围** - **限流条件（超出任一数值时触发限流）** > **以下为每分钟限流条件，服务可能按 RPS（RPM/60）与 TPS（TPM/60）限制** @@ -9023,16 +8105,12 @@ qwen-plus-character gui-plus -中国内地 - 80 540,000 gui-plus-2026-02-26 -中国内地 - 100 540,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/regions.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/regions.md index d9c3ff39..c6b3ce76 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/regions.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/regions.md @@ -1,9 +1,11 @@ -# 选择地域、服务部署范围和接入域名 +# 地域及接入域名 -调用百炼前先选择**地域**、**服务部署范围、接入域名：** +阿里云百炼提供多个地域的服务，各地域提供业务空间专属、Dashscope 等多种接入域名，并支持多种服务部署范围，以满足不同场景的接入需求。 + +## 地域概述 + +地域决定**接入点和数据存储位置**，就近选择可降低延迟。 -- 地域：决定**接入点和数据存储位置**，就近选择可降低延迟； - - 服务部署范围：决定**推理执行位置**，有数据合规需求选择特定地理边界的部署范围，无合规需求选择全球部署范围（推理资源池更大）； - 接入域名：影响**并发上限、超时等服务保障**，各地域具有独立的接入域名。 @@ -11,74 +13,48 @@ 一次完整的模型调用流程如下： -1. 应用经接入域名将请求发送到所选**地域**（如华北2-北京），请求数据存于该地域； +1. 应用经接入域名将请求发送到所选**地域**，请求数据存于该地域； 2. 接入地域将请求转发至**服务部署范围**内的推理节点完成计算（过程数据不持久化，传输全程加密）； 3. 推理结果回到接入地域存储，再响应给应用（用户静态数据始终存于所选地域）。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7478182871/CAEQchiBgIDPq_PR9xkiIDJhZDdiNzAxMGFiODRhNmRiMDYxYjNjNGU2NTJkMDYw7466796_20260515102254.505.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7931094871/CAEQchiBgIDh6dzL_BkiIDk2MmRhMjdmMGQyNTQ5N2ZiNzNlMGM4NjllMDBjMTUy7466796_20260515102254.505.svg) -## 选择地域和服务部署范围 - -按场景查表选地域和服务部署范围： - -**使用场景** +各地域支持的服务部署范围如下： **地域** **服务部署范围** -无数据驻留限制，追求更大推理资源池（跨多地域调度推理，自行确保跨境合法） - 美国（弗吉尼亚） -全球（任意可用节点，含中国境内及海外） +全球 -无数据驻留限制，追求更大推理资源池（跨多地域调度推理，自行确保跨境合法） +美国 德国（法兰克福） -全球（任意可用节点，含中国境内及海外） +全球 -无数据驻留限制，追求更大推理资源池（跨多地域调度推理，自行确保跨境合法） +欧盟 日本（东京） -全球（任意可用节点，含中国境内及海外） +全球 -要求数据不出中国内地 +日本 华北2（北京） -中国内地（限境内推理） - -要求数据不经过中国内地（会跨多地域调度推理，自行确保跨境合法） +中国内地 新加坡 -国际（除中国内地以外的全球节点） - -要求数据不出美国 - -美国（弗吉尼亚） - -美国（限境内推理） +国际 -要求数据不出欧盟 - -德国（法兰克福） - -欧盟（限境内推理） - -要求数据不出日本 - -日本（东京） - -日本（限境内推理） - -## 选择接入域名 +## 接入域名 百炼为模型推理 API 提供业务空间专属、Dashscope 和试用三种接入域名，适用于从试用体验到企业级生产的不同场景。推荐使用**业务空间专属域名**，各域名的核心差异如下： @@ -104,7 +80,7 @@ 推荐在生产环境中使用，具备更高并发承载能力与网络隔离性，保障大流量场景下的稳定、低延迟访问体验。 -存量业务兼容，建议[迁移至业务空间专属域名](#section-migrate-domain)。 +存量业务兼容，建议迁移至[业务空间专属域名](#section-migrate-domain)。 快速体验、功能验证，不建议用于生产环境。 @@ -247,13 +223,9 @@ HTTP、SSE 从 Dashscope 域名或试用域名迁移到业务空间专属域名只需两步，无需修改业务逻辑代码： -1. **获取业务空间专属域名**： +1. **获取业务空间专属域名**：在[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)页面，复制 **API Host** 列的内容。 - - 方式一：在[API Key 创建](https://bailian.console.aliyun.com/cn-beijing#/api-key)后的弹窗中，复制 **API Host** 。 - - - 方式二：在[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)页面，复制 **API Host** 列的内容。 - -2. **替换 Base URL 中的域名**：将原域名替换为业务空间专属域名。以华北2（北京）地域为例，`llm-xxx` 为业务空间 ID： +2. **替换请求地址中的域名**：将复制的 API Host（如 `llm-xxx.cn-beijing.maas.aliyuncs.com`）替换代码中原有的域名部分，以华北2（北京）地域为例，`llm-xxx` 为业务空间 ID： - OpenAI 兼容接口：从 `https://dashscope.aliyuncs.com/compatible-mode/v1` 替换为 `https://llm-xxx.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/what-is-model-studio.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/what-is-model-studio.md index 86dd07a9..9a83af9b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/what-is-model-studio.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/what-is-model-studio.md @@ -223,7 +223,7 @@ A：百炼采用按量付费，本身**没有"自动扣费"开关**。以下措 - **清理计费资源：**删除不再使用的知识库；前往[模型部署](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)页面，下线按算力时长计费的部署实例。 -- **开启"**[免费额度用完即停](https://help.aliyun.com/zh/model-studio/new-free-quota#d1cb80ac11i92)**"（仅限新用户且在免费额度有效期内）：**在模型详情页开启此开关，免费额度耗尽后服务自动停止，不会转为付费。仅适用于华北2（北京）地域（[中国内地服务部署范围](https://help.aliyun.com/zh/model-studio/regions/#080da663a75xh)），且须在免费额度有效期内。 +- **开启"**[免费额度用完即停](https://help.aliyun.com/zh/model-studio/new-free-quota#d1cb80ac11i92)**"（仅限新用户且在免费额度有效期内）：**在模型详情页开启此开关，免费额度耗尽后服务自动停止，不会转为付费。仅适用于华北2（北京）地域，且须在免费额度有效期内。 - **设置费用监控和预警：**查看 [账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)和[模型监控](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)，并设置[高额消费预警](https://usercenter2.aliyun.com/home/alarm-threshold)，及时发现异常消费。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-data-overview/model-log-backflow.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-data-overview/model-log-backflow.md new file mode 100644 index 00000000..ab920367 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-data-overview/model-log-backflow.md @@ -0,0 +1,330 @@ +# 日志回流 + +日志回流将 SLS（日志服务）推理日志转化为可用于模型微调或评测的结构化数据集。 + +## 功能概述 + +日志回流功能将 SLS（日志服务）中的推理日志数据回流到百炼平台，经过格式化处理后生成结构化数据集（JSONL 格式），可用于模型微调或模型评测。回流产出的是结构化数据，而非原始日志的直接副本。 + +### 支持范围 + +日志回流支持创建以下两类数据集： + +- **训练集**：训练场景为文本生成，训练方式支持 SFT（监督微调）、DPO（直接偏好优化）和 CPT（持续预训练）。 + +- **评测集**：支持文本生成场景。 + + +日志回流目前仅在**华北2（北京）**和**新加坡** Region 可用，其他 Region 不显示日志回流入口。单次回流上限为 10 万条，可多次回流到同一数据集的不同版本以积累更多数据。 + +日志回流支持**平台存储**（默认）和对象存储 **OSS 挂载**（需额外授权）两种存储方式。存储方式的差异和选择指引，请参见[创建日志回流数据集](#sec-create)。 + +回流生成的数据集可直接用于下游任务：训练集可用于[模型调优](https://help.aliyun.com/zh/model-studio/model-training-on-console#topic-2529531)，评测集用于模型评测。数据也支持后续进行[数据清洗](https://help.aliyun.com/zh/model-studio/data-processing#topic-2761062)。 + +## 开通并授权相关服务 + +使用日志回流前，请确认当前 Region 为华北2（北京）或新加坡（其他 Region 不显示日志回流入口），并在[模型监控](https://bailian.console.aliyun.com/#/model-telemetry)页面完成以下服务开通和权限授权。全部完成后授权配置抽屉自动关闭，进入日志回流表单。 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3463064871/p1086527.png) + +### 开启审计日志和推理日志 + +审计日志和推理日志各需完成三个步骤，共六个条件全部满足后才能使用日志回流。审计日志是推理日志的前置依赖，必须先完成审计日志的全部步骤。 + +**审计日志** + +1. 授权 SLS 服务关联角色：单击**立即授权**，授权 AliyunServiceRoleForSFMAccessSLS 角色。未授权时显示红色**未授权**标签。 + +2. 开通 SLS 日志服务：未开通时显示**未开通**状态和跳转链接，单击后前往 SLS 控制台完成开通。 + +3. 开启审计日志：单击**创建并开启审计日志**，系统创建 LogStore（日志库）实例并轮询等待就绪（最多 60 秒）。 + + +**推理日志** + +1. 授权推理日志的 SLS 服务关联角色。 + +2. 确认 SLS 日志服务已开通。 + +3. 开启推理日志：审计日志未开启时该按钮置灰，需先完成审计日志开启。 + + +**重要** + +开启必须按序：先审计日志，再推理日志。关闭必须反序：先关闭推理日志，再关闭审计日志。关闭日志后已有数据不可复原，操作前请确认不再需要日志数据。推理日志开启后 SLS 持续产生存储和读写费用，不再需要时应及时关闭。 + +### OSS 多角色授权（仅 OSS 挂载模式） + +选择 OSS 挂载存储方式时，需额外完成以下授权： + +1. 在弹出的授权弹窗中，勾选数据访问授权协议。 + +2. 单击**一键授权**，系统自动授权以下两个服务关联角色： + + - AliyunServiceRoleForAccessCusOss（OSS 写入角色） + + - AliyunServiceRoleForSFMDataHubOSSImport（DataHub 导入角色） + + +下表汇总了日志回流涉及的服务关联角色。 + +**角色名称** + +**用途** + +**授权时机** + +AliyunServiceRoleForSFMAccessSLS + +百炼访问 SLS 日志数据 + +审计日志和推理日志各授权一次 + +AliyunServiceRoleForAccessCusOss + +OSS 数据写入 + +选择 OSS 挂载存储时 + +AliyunServiceRoleForSFMDataHubOSSImport + +DataHub 数据导入 + +选择 OSS 挂载存储时 + +## 创建日志回流数据集 + +日志回流提供三个入口，均可进入配置表单创建数据集： + +### 模型监控列表页 + +在[**模型监控**列表页](https://bailian.console.aliyun.com/#/model-telemetry)顶部，单击**日志回流**。首次使用时先完成授权配置，授权通过后自动展示日志回流表单。 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3463064871/p1086529.png) + +### 模型监控详情页 + +在模型监控详情页的时间选择器区域，单击**日志回流**。从此入口进入时，表单自动预填当前页面的时间范围、API Key 和模型，且模型不可修改。 + +### 数据管理页 + +在**数据管理**页面新建数据集时，导入方式选择**日志回流**。该选项仅在训练集（文本生成 + SFT）或评测集（文本生成）时可见。 + +### 配置回流参数 + +进入日志回流表单后，按从上到下的顺序配置以下参数。部分参数有前置依赖：API Key 过滤需先选时间范围，模型选择需先选时间范围和 API Key。修改时间范围、数据类型、训练场景或训练方式会联动重置其他参数，建议严格按顺序填写。各参数取值说明见下表。 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3463064871/p1086595.png) + +**预估日志回流数据**：系统根据筛选条件显示预估的回流数据条数。超过 10 万条时红色警告，超出部分不会被回流。查询结果过多时，**确定**按钮禁用，需缩小筛选范围。 + +**说明** + +表单联动重置规则：修改时间范围会重置 API Key（回到**全部**）和模型选择（清空）；修改数据类型、训练场景或训练方式会重置存储位置和导入方式。 + +**重要** + +存储方式、数据类型和训练方式在创建后均不可更改，选择前请仔细确认。 + +**参数** + +**说明** + +**是否必填** + +**取值说明** + +回流位置 + +数据集的存储方式 + +是 + +平台存储（默认）或 OSS 挂载。评测集时 OSS 挂载禁用。创建后不可更改 + +数据集名称 + +数据集在列表中的显示名称 + +是 + +中文、英文、数字、下划线、斜杠、连字符，最大 50 字符。建议采用 功能场景\_模型名\_时间 格式命名。创建后不可修改 + +数据集描述 + +数据集用途补充说明 + +否 + +最大 200 字符 + +类型与格式 + +数据集用途类型 + +是 + +训练集或评测集。选择评测集时训练场景和训练方式隐藏。创建后不可更改 + +训练场景 + +训练场景类型（仅训练集显示） + +是 + +当前仅支持文本生成 + +训练方式 + +微调方法（仅训练集显示） + +是 + +SFT、DPO、CPT，选项由系统动态展示。创建后锁定 + +时间范围 + +回流日志的时间段 + +是 + +最近 30 天（含当天），精确到时分秒。修改会重置 API Key 和模型选择 + +API Key 过滤 + +按 API Key 筛选日志数据 + +是 + +全部（不过滤）、其他（排除已列出 Key）、或选择具体 Key（多选） + +模型选择 + +回流目标模型 + +是 + +最多 10 个，按能力类型过滤不匹配模型置灰 + +OSS 数据路径 + +数据存储的目标目录（仅 OSS 挂载模式显示） + +是 + +Bucket 需在同一 Region + +## 查看回流结果 + +提交日志回流任务后，在[**数据管理**列表页](https://bailian.console.aliyun.com/#/efm/data_ass)查看数据集和导入进度。 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3463064871/p1088579.png) + +### 列表页展示 + +日志回流创建的数据集在列表页中的导入方式显示为**日志回流**，存储位置根据创建时的选择显示为**平台存储**或**OSS 挂载**。 + +可在列表页查看任务的导入状态。任务失败时，可查看系统返回的具体失败原因。 + +平台存储模式下，导入完成后系统自动发布数据集版本，无需手动操作。 + +### 详情页信息 + +数据集详情页展示的信息根据存储方式有所不同： + +- **OSS 挂载**：展示发布状态、数据量、创建时间、FileID、数据类型、导入状态和 OSS 挂载地址。 + +- **平台存储（OSS 导入）**：展示发布状态、数据量、创建时间、FileID、数据类型、导入状态和 OSS 导入地址。 + +- 其他情况：展示发布状态、数据量、创建时间、FileID、数据类型和导入状态。 + + +## 追加日志回流数据 + +在[数据管理](https://bailian.console.aliyun.com/#/efm/data_ass)页面，向已有数据集追加新一批日志回流数据，有以下两种方式： + +### 导入数据页 + +进入已有数据集的**导入数据**页面，选择**日志回流**作为导入方式。表单参数与创建流程一致，详见[创建日志回流数据集](#sec-create)。此方式额外支持按工作空间过滤。 + +此方式适用于所有存储类型的数据集，包括 OSS 挂载数据集。 + +### 新增版本弹窗 + +在数据集详情页，单击**新增版本**，在弹窗中选择日志回流导入方式。仅平台存储数据集可用此方式。 + +**说明** + +OSS 挂载数据集不支持**新增版本**操作，按钮置灰。需通过**导入数据**页追加。 + +### 增量回流最佳实践 + +推荐分批次回流，针对不同时段或不同模型分别执行回流任务，逐步积累高质量训练集。每批次可精准选择表现良好的模型和业务高峰时段的数据，确保数据质量优于一次性大量回流。 + +## 常见问题 + +**API Key 过滤的「其他」包含哪些请求？** + +**问题**：API Key 过滤中**其他**包含哪些请求？ + +**其他**过滤掉已列出 Key 的请求，包含已删除的 Key 或其他工作空间产生的日志。**全部**则不按 Key 过滤。三种模式（全部/其他/逐个选择）的完整说明见[创建日志回流数据集](#sec-create)的 API Key 过滤参数。 + +**10 万条上限是数据集总量限制吗？** + +**问题**：单次回流上限 10 万条，是否意味着数据集总量也被限制在 10 万条？ + +不是。10 万条是单次回流的上限，并非数据集的总量上限。可以多次回流到同一数据集的不同版本，累计数据量不受此限制。例如分批次针对不同时段回流，每次 10 万条以内，最终数据集可以积累远超 10 万条的数据。 + +**平台存储和 OSS 挂载有什么区别？** + +**问题**：两种存储方式除了数据存储位置不同，还有哪些差异？ + +两种存储方式的核心差异对比如下。 + +**维度** + +**平台存储** + +**OSS 挂载** + +额外授权 + +无需额外授权 + +需授权两个服务关联角色并勾选数据访问协议 + +新增版本 + +支持 + +不支持（按钮置灰），需通过导入数据页追加 + +数据访问 + +通过控制台访问 + +可在 OSS Bucket 直接查看和管理 JSONL 文件 + +评测集 + +支持 + +不支持（选项禁用） + +自动发布 + +导入完成后自动发布版本 + +不自动发布 + +**审计日志和推理日志是同一个操作吗？** + +**问题**：开启日志需要分别操作审计日志和推理日志，两者有什么关系？ + +两者需分别开启，且必须先完成审计日志才能开启推理日志。操作步骤详见[开通并授权相关服务](#sec-authorize)。 + +**预估数据量为什么和实际回流结果不一致？** + +**问题**：提交前显示的预估数据量与回流完成后的实际条数有差异。 + +预估数据量是基于筛选条件的近似估算值，实际回流完成后的数据条数可能略有差异，这是正常现象。预估值用于帮助判断是否需要调整筛选条件（如超过 10 万条时提示缩小范围），不代表精确计数。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-deployment-introduction.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-deployment-introduction.md index fbf3abc6..d8fb82b2 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-deployment-introduction.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-deployment-introduction.md @@ -108,7 +108,7 @@ 1. 预付费按天计费。无法提前退费 -2. 如果单位时间内使用超出购买的吞吐量，将自动切换成百炼提供的[模型调用](https://help.aliyun.com/zh/model-studio/model-pricing)服务。 +2. 如果单位时间内使用超出购买的吞吐量，按创建时选择的溢出策略处理：自动溢出则切换为该模型的[模型调用](https://help.aliyun.com/zh/model-studio/model-pricing)按量付费，仅使用 PTU 容量则返回 429。 预付费购买后，若在首月内提前退订，日单价（≈ 月单价 / 30）将按 **1.2** 倍计费 @@ -137,9 +137,9 @@ - 后付费时，如果账户欠费，部署的资源将继续保留并计费 24 小时，在这 24 小时内服务仍可正常使用。超过 24 小时后系统停止计费，模型部署进入欠费状态，底层资源将被删除，但模型部署任务仍会保留。补足欠费后，系统将重新分配资源并恢复使用（恢复后继续产生费用）。如果您不希望继续产生费用，可删除模型部署任务，删除成功后将不再计费。 -当模型输入超过最长输入 Token 或 超出购买的 TPM 量时，相关调用将自动切换为当前模型的按量付费模式。此时，推理性能可能下降，将受业务空间中当前快照模型的公共流量的管控，[费用](https://help.aliyun.com/zh/model-studio/model-pricing)按模型调用（按量付费）标准计收。 +当模型输入超过最长输入 Token 时，相关调用将自动切换为当前模型的按量付费模式；超出购买的 TPM 量时，按创建时选择的溢出策略处理（「自动溢出」切换为按量付费，「仅使用 PTU 容量」返回 429）。此时，推理性能可能下降，将受业务空间中当前快照模型的公共流量的管控，[费用](https://help.aliyun.com/zh/model-studio/model-pricing)按模型调用（按量付费）标准计收。 -- 此时，调用 API 返回 Header 将包含：`x-dashscope-ptu-overflow:true`。 +- 此时（仅「自动溢出」策略下），调用 API 返回 Header 将包含：`x-dashscope-ptu-overflow:true`。 - TPM 统计请前往：[模型监控（北京）](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)。 @@ -483,15 +483,33 @@ MU2 x 8 ¥240,288 +MU3 x 8 + +¥1,096 + +¥527,752 + 千问3.6-35B-A3B qwen3.6-35b-a3b -MU8 x 1 +MU1 x 8 -¥47 +¥432 -¥22,400 +¥208,944 + +MU2 x 8 + +¥504 + +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 MU9 x 1 @@ -509,6 +527,12 @@ MU1 x 2 ¥52,236 +MU3 x 8 + +¥1,096 + +¥527,752 + 千问3.6-Plus-2026-04-02 qwen3.6-plus-2026-04-02 @@ -529,12 +553,6 @@ PD分离模式：¥417,888 qwen3.5-397b-a17b -MU2 x 8 - -¥504 - -¥240,288 - MU3 x 8 MU3 x 16（PD分离模式） @@ -547,6 +565,12 @@ PD分离模式：¥2,192 PD分离模式：¥1,055,504 +MU6 x 16 + +¥400 + +¥193,424 + 千问3.5-122B-A10B qwen3.5-122b-a10b @@ -557,11 +581,11 @@ MU1 x 4 ¥104,472 -MU2 x 8 +MU6 x 16 -¥504 +¥400 -¥240,288 +¥193,424 千问3.5-35B-A3B @@ -579,10 +603,40 @@ MU2 x 8 ¥240,288 +MU3 x 8 + +¥1,096 + +¥527,752 + +MU9 x 1 + +¥51 + +¥24,600 + 千问3.5-27B qwen3.5-27b +MU2 x 8 + +¥504 + +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 + +MU8 x 1 + +¥47 + +¥22,400 + MU9 x 1 ¥51 @@ -599,6 +653,12 @@ MU1 x 2 ¥52,236 +MU2 x 2 + +¥126 + +¥60,072 + MU8 x 1 ¥47 @@ -625,12 +685,24 @@ MU1 x 2 qwen3.5-plus-2026-02-15 +MU1 x 8 + MU1 x 16（PD分离模式） +¥432 + PD分离模式：¥864 +¥208,944 + PD分离模式：¥417,888 +MU2 x 8 + +¥504 + +¥240,288 + MU3 x 8 MU3 x 16（PD分离模式） @@ -659,57 +731,25 @@ MU2 x 8 ¥240,288 -千问3-Next-80B-A3B-Instruct - -qwen3-next-80b-a3b-instruct - -MU1 x 2 - -¥108 - -¥52,236 - 千问3-32B qwen3-32b -MU1 x 4 - -¥216 - -¥104,472 - -MU6 x 4 - -¥100 - -¥48,356 - -千问3-30B-A3B - -qwen3-30b-a3b - -MU9 x 2 - -¥102 - -¥49,200 - -千问3-30B-A3B-Instruct-2507 +MU6 x 16 -qwen3-30b-a3b-instruct-2507 +¥400 -MU1 x 4 +¥193,424 -¥216 +千问3-30B-A3B-Thinking-2507 -¥104,472 +qwen3-30b-a3b-thinking-2507 -MU2 x 8 +MU1 x 2 -¥504 +¥108 -¥240,288 +¥52,236 千问3-8B @@ -759,12 +799,6 @@ MU1 x 2 ¥52,236 -MU5 x 1 - -¥21 - -¥10,139 - 千问3-Embedding-0.6B qwen3-embedding-0.6b @@ -837,11 +871,11 @@ MU5 x 1 qwen2.5-72b-instruct -MU1 x 4 +MU1 x 8 -¥216 +¥432 -¥104,472 +¥208,944 千问2.5-开源版-32B @@ -879,26 +913,6 @@ MU5 x 1 ¥10,139 -千问2.5-开源版-3B - -qwen2.5-3b-instruct - -MU5 x 1 - -¥21 - -¥10,139 - -千问-Flash-2025-07-28 - -qwen-flash-2025-07-28 - -MU1 x 4 - -¥216 - -¥104,472 - 千问-Plus-2025-07-28 qwen-plus-2025-07-28 @@ -955,6 +969,12 @@ GLM-5.1 glm-5.1 +MU2 x 8 + +¥504 + +¥240,288 + MU3 x 16（PD分离模式） PD分离模式：¥2,192 @@ -987,6 +1007,16 @@ PD分离模式：¥800 PD分离模式：¥386,848 +GLM-4.7-Flash + +glm-4.7-flash + +MU3 x 16（PD分离模式） + +PD分离模式：¥2,192 + +PD分离模式：¥1,055,504 + #### DeepSeek **模型名称** @@ -1007,11 +1037,11 @@ DeepSeek-v4-Flash deepseek-v4-flash -MU1 x 8 +MU3 x 8 -¥432 +¥1,096 -¥208,944 +¥527,752 DeepSeek-v3.2 @@ -1039,16 +1069,6 @@ PD分离模式：¥480,576 **最小计费：天** -MiniMax-M2.5 - -MiniMax-M2.5 - -MU1 x 16（PD分离模式） - -PD分离模式：¥864 - -PD分离模式：¥417,888 - Kimi-K2.5 kimi-k2.5 @@ -1091,15 +1111,21 @@ MU2 x 8 **最小计费：天** -千问3-VL-235B-A22B-Instruct +千问3-VL-32B-Instruct -qwen3-vl-235b-a22b-instruct +qwen3-vl-32b-instruct -MU1 x 4 +MU2 x 8 -¥216 +¥504 -¥104,472 +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 千问3-VL-8B-Instruct @@ -1111,6 +1137,12 @@ MU1 x 2 ¥52,236 +MU5 x 1 + +¥21 + +¥10,139 + 千问3-VL-4B-Instruct qwen3-vl-4b-instruct @@ -1131,6 +1163,16 @@ MU5 x 1 ¥10,139 +千问3-VL-Embedding-2B + +qwen3-vl-embedding-2b + +MU5 x 1 + +¥21 + +¥10,139 + 千问3-VL-Flash-2025-10-15 qwen3-vl-flash-2025-10-15 @@ -1161,16 +1203,6 @@ MU6 x 4 ¥48,356 -千问VL-OCR-2025-11-20 - -qwen-vl-ocr-2025-11-20 - -MU6 x 4 - -¥100 - -¥48,356 - #### 千问 Omni **模型名称** @@ -1251,6 +1283,14 @@ MU5 **元/千Token** +千问3.5-27B 邀测中 + +qwen3.5-27b + +¥0.0006 + +¥0.0048 + 千问3-32B qwen3-32b @@ -1281,6 +1321,16 @@ qwen3-8b 思考模式：¥0.005 +千问3-4B-Instruct-2507 + +qwen3-4b-instruct-2507 + +¥0.0003 + +非思考模式：¥0.0012 + +思考模式：¥0.003 + 千问2.5-开源版-72B qwen2.5-72b-instruct @@ -1313,6 +1363,14 @@ qwen2.5-7b-instruct ¥0.001 +千问2-开源版-7B + +qwen2-7b-instruct + +¥0.001 + +¥0.002 + #### 千问VL **基础模型** @@ -1359,6 +1417,14 @@ qwen2.5-vl-7b-instruct ¥0.005 +千问2.5-VL-3B-Instruct + +qwen2.5-vl-3b-instruct + +¥0.0012 + +¥0.0036 + 如果需要部署更多模型，请参考此[解决方案](https://www.aliyun.com/solution/tech-solution/deepseek-r1-for-platforms)并结合具体业务需求选择最适合的部署方案。 ## 部署方法 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-import.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-import.md index ec850a53..8e696bf6 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-import.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-import.md @@ -1,41 +1,35 @@ # 模型导入 -**[我的模型](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/model_center)**页面用于管理您创建和导入的模型。通过该页面，您可以将本地训练的 LoRA 模型从阿里云对象存储 OSS 导入到百炼平台。 +介绍百炼平台从阿里云 OSS 导入 LoRA 模型的全流程操作，涵盖首次 OSS 授权、模型文件准备、导入表单填写、导入模型管理与常见问题排查。 -## 使用前提 +## 模型导入概述 -导入前，请确保满足以下条件： +本篇介绍将本地训练的 LoRA 模型从阿里云对象存储 OSS 导入到百炼平台的全流程，覆盖首次 OSS 授权、模型文件准备、导入表单填写、模型查看管理与删除，以及导入失败、已失效等常见问题排查。导入成功后即可部署服务，部署、扩缩容与下线操作详见[部署运维](#)，通过 API 完成部署与调用详见[使用 API 进行模型部署](#)，部署前置概念与计费方案对比详见[模型部署简介](#)，本篇与三者构成概念、导入、部署、调用的完整链路。 -- **OSS Bucket 准备**：已创建 OSS Bucket，并为目标 Bucket 添加标签。 - - **说明** - - - 支持的 OSS Bucket 存储类型不包括归档、冷归档或深度冷归档。支持内容加密的 Bucket。支持私有的Bucket。 - - - 不支持访问OSS Bucket根目录下的文件，请您在OSS Bucket下选择已有的子目录或新建一个子目录供阿里云百炼访问。 - - - 支持导入任意大小的模型文件，导入后将使用阿里云百炼提供的免费存储空间。 - - -- **模型文件准备**：模型文件需符合[导入要求与限制](#h2-file-format-constraints)。模型文件夹需直接放在 OSS Bucket 中，系统会自动识别。 - +导入页基础模型字段由平台接口动态返回当前支持导入的基础模型清单，可能随版本更新，请以控制台可选列表为准。当前支持的基础模型如下： -## 支持导入的基础模型 +**当前支持的基础模型清单（点击展开）** -当前支持导入以下基础模型的 LoRA 微调版本： +以下清单由后端接口动态返回，可能随版本更新而调整，请以控制台实际可选项为准。 -模型系列 +**模型系列** -模型名称 +**模型名称** 千问3 千问3-32B +千问3 + 千问3-14B +千问3 + 千问3-8B +千问3 + 千问3-4B-Instruct-2507 千问3-VL @@ -46,249 +40,320 @@ 千问2.5-72B-Instruct +千问2.5 + 千问2.5-32B-Instruct +千问2.5 + 千问2.5-14B-Instruct +千问2.5 + 千问2.5-7B-Instruct 千问2.5-VL 千问2.5-VL-72B-Instruct +千问2.5-VL + 千问2.5-VL-7B-Instruct -## 操作步骤 +训练方式可选项取决于所选基础模型的声明，选择基础模型后训练方式字段自动渲染可选项并默认选中第一项。当前版本仅支持导入 LoRA 模型，全参微调模型不可导入。 -按照以下步骤将 LoRA 模型从 OSS 导入到百炼平台： +导入来源仅支持「从 OSS 导入」一项，表单中默认选中且无其他选项，暂不支持从其他渠道导入模型。 -1. 在**我的模型**页面，点击**导入模型**按钮。 - -2. 在导入模型页面中，填写以下信息： - - - **模型名称**：输入模型的显示名称，最多50个字符。 - - - **基础模型**：选择该 LoRA 模型对应的基础模型。 - - - **训练方式**：选择模型的训练方式。可选项取决于所选基础模型，选择基础模型后自动显示。 - - - **导入来源**：当前仅支持"从OSS导入"，系统已默认选中。 - - - **Bucket**：选择存储模型文件的 OSS Bucket。 - - - **模型目录**：在选定的 Bucket 中浏览并选择模型 Checkpoint 所在的目录。 - - - **模型加密**：为保障您的数据安全，平台会为导出的模型文件开启 OSS 服务端加密，使用 OSS 完全托管密钥进行加解密（SSE-OSS），加密算法为 AES256。 - -3. 确认信息无误后，点击**确定**提交导入请求。如需放弃导入，点击**取消**返回列表页。系统将自动验证模型文件格式和完整性，验证通过后开始导入。导入完成后，您可以在**[我的模型](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/model_center)**页面查看导入的模型，并进行部署、增量训练或删除等操作。 - - 导入后的模型状态包括：创建中（正在导入）、创建成功（导入成功，可部署）、创建失败（导入失败）、已失效（模型文件已变更）。列表页还展示模型名称/ID、基础模型、来源、支持部署方式和创建时间等信息。您可以通过页面顶部的搜索框按名称筛选模型。 - +导入成功后的模型状态为创建成功，可在[我的模型](https://bailian.console.aliyun.com/#/efm/model_center)列表中部署服务；从 OSS 导入的模型不支持增量训练，如需迭代请重新训练后再次导入。导入后将使用百炼提供的免费存储空间存放模型记录。 + +## 首次导入前完成 OSS 授权 -## 导入要求与限制 +首次从 OSS 导入模型前，须完成 OSS 服务关联角色授权，并为目标 Bucket 添加访问标签。授权通过导入页一键完成，授权后系统自动创建服务关联角色 AliyunServiceRoleForSFMDataHubOSSImport（服务名 datahub.sfm.aliyuncs.com，权限策略 AliyunServiceRolePolicyForSFMDataHubOSSImport），通常秒级生效。服务关联角色说明详见[OSS 服务关联角色](#)，主账号与子账号的概念与区别详见[主账号与子账号](#)。 -**重要** +在[导入模型页](https://bailian.console.aliyun.com/#/efm/model_center/import_model)的「导入来源」选择「从 OSS 导入」后，若未授权，OSS 字段区会提示「您还未授权OSS」并显示「前往授权」链接，确定按钮在未授权时不可用。主账号与子账号的授权路径不同，请按账号类型选择。 -重要：当前版本仅支持导入 LoRA（Low-Rank Adaptation）模型，不支持导入全参微调模型。 +【截图：ss-oss-auth-01.png — 导入页 OSS 字段区「您还未授权OSS」提示与「前往授权」入口，及授权弹窗】 -导入 LoRA 模型前，请确保满足以下要求： +### 使用主账号 -- **必需文件**：OSS Bucket 中需包含以下文件： +1. 在导入模型页「导入来源」选择「从 OSS 导入」后，页面提示「您还未授权OSS」，在提示栏右侧点击「前往授权」。 - - **adapter\_model.safetensors**：LoRA 适配器的权重文件，采用 SafeTensors 格式存储。 - - - **adapter\_config.json**：LoRA 适配器的配置文件，包含 rank、alpha 等关键参数信息。 - -- **rank 参数限制**：rank 值必须为 8、16、32 或 64 中的一个，且同一模型的所有 LoRA 层必须使用相同的 rank 值。 +2. 在弹出的对话框中点击「立即授权」，系统自动创建服务关联角色 AliyunServiceRoleForSFMDataHubOSSImport，通常秒级生效，服务高峰期可能稍有延迟。 + +3. 为目标 OSS Bucket 添加标签：访问 OSS 管理控制台的 Bucket 列表，找到目标 Bucket，悬停标签列图标并点击「前往编辑」→「创建标签」，添加标签名 bailian-datahub-access、标签值 read，保存。 -- **修改词汇表的模型**：如果训练过程中添加了新 token 或修改了原始词汇表（vocab），该模型无法导入。系统要求使用与基础模型完全一致的词汇表。 +4. 返回导入模型页，重新选择目标 Bucket 再尝试导入。百炼不支持访问 Bucket 根目录下的文件，须选择 Bucket 下已有的子目录或新建子目录。 -- **修改对话模板的模型**：如果训练过程中修改了 chat\_template 配置，该模型无法导入。系统仅支持使用与对应开源基础模型默认配置一致的 chat\_template。 + +### 使用子账号 + +1. 在导入模型页「导入来源」选择「从 OSS 导入」后，页面提示「您还未授权OSS」，点击「前往授权」。 + +2. 在弹出的对话框中点击「立即授权」，界面提示「授权失败：当前用户没有创建服务关联角色的权限」。对话框显示 Service Name 为 datahub.sfm.aliyuncs.com，服务关联角色名称为 AliyunServiceRoleForSFMDataHubOSSImport，所需用户权限为 ram:CreateServiceLinkedRole。须先由主账号授予子账号创建服务关联角色的权限，再由子账号完成授权。 - chat\_template 配置通常位于以下位置： +3. 由主账号授予子账号创建服务关联角色的权限： - - 模型的 config.json 文件中的 `chat_template` 字段。 + 1. 主账号登录 RAM 控制台，在左侧导航栏选择「权限管理 → 权限策略」，点击「创建权限策略」。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1185949671/p1049475.png) + 2. 选择「脚本编辑」，在 Effect、Action、Resource、Condition 中分别输入以下脚本内容，点击「确定」： - - tokenizer\_config.json 文件中的 `chat_template` 字段。 + ``` + { + "Action": [ + "ram:CreateServiceLinkedRole" + ], + "Resource": "*", + "Effect": "Allow", + "Condition": { + "StringEquals": { + "ram:ServiceName": "datahub.sfm.aliyuncs.com" + } + } + } + ``` + + 3. 输入权限策略名称（示例：服务关联角色）后点击「确定」。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1185949671/p1049476.png) + 4. 在左侧导航栏选择「身份管理 → 用户」，找到待授权的子账号，点击操作列**新增授权**。 -- **未冻结 VIT 的视觉语言模型**：对于 VL（Vision-Language）模型，必须冻结 Vision Transformer（VIT）部分。如果 LoRA adapter 中包含 visual 相关的权重参数（即未冻结 VIT），该模型无法导入。 + 5. 选择刚才创建的自定义权限策略，点击「确认新增授权」，子账号即拥有创建服务关联角色的权限。 + +4. 返回导入模型页点击「前往授权」，在弹出的对话框中点击「立即授权」，系统自动创建服务关联角色，通常秒级生效。 - 可以运行以下代码判断。 +5. 为目标 OSS Bucket 添加 bailian-datahub-access=read 标签（操作同主账号步骤 3），然后返回导入模型页重新选择目标 Bucket。百炼不支持访问 Bucket 根目录下的文件。 - ``` - from safetensors import safe_open - import argparse + +授权后须为目标 OSS Bucket 添加标签：标签名为 bailian-datahub-access，标签值为 read。该标签用于标记百炼可访问的 Bucket，未添加此标签的 Bucket 在下拉列表中不可选，须到[OSS 管理控制台](https://oss.console.aliyun.com/)添加标签后重新选择。 + +OSS Bucket 存储类型不支持归档、冷归档或深度冷归档，支持内容加密的 Bucket 与私有 Bucket。百炼不支持访问 Bucket 根目录下的文件，须选择 Bucket 下已有的子目录或新建子目录。 + +### 旧授权方式升级 + +若您此前使用的是旧授权方式，导入页 Bucket 字段下方会提示「建议转换为新的 Bucket 授权方式，提升安全性」并提供「直接转换」链接。点击后弹出确认框，确认即可升级为服务关联角色授权方式，升级不影响原有数据。 + +### 未开通 OSS 产品 + +若主账号尚未开通对象存储 OSS，导入页会提示「您还未开通OSS」并提供前往购买的链接。须由主账号前往 OSS 控制台开通 OSS 后返回导入页重新授权。 + +## 准备 LoRA 模型文件 + +导入前须将 LoRA 模型文件按以下要求存放在 OSS Bucket 的子目录中（不支持 Bucket 根目录），并在提交前通过系统的自动校验。模型文件须直接放在所选子目录下，系统会自动识别。 + +当前版本仅支持导入 LoRA 模型，不支持导入全参微调模型。 + +### 必需文件与目录结构 + +子目录中须包含以下文件：adapter\_model.safetensors（LoRA 适配器权重，SafeTensors 格式）、adapter\_config.json（含 rank、alpha 等参数的配置文件）、config.json（基础模型配置）。选中目录后系统会自动校验这些文件的格式与完整性。 + +### 训练参数约束 + +- **rank 取值**：rank 必须为 8、16、32 或 64 之一，且同一模型的所有 LoRA 层须使用相同的 rank 值，否则无法导入。 - def print_safetensor_structure(file_path): - print(f"Loading safetensor file: {file_path}") - print("="*80) - - with safe_open(file_path, framework="pt") as f: - keys = f.keys() - print(f"Found {len(keys)} tensors in the file:\n") - - for key in sorted(keys): - tensor = f.get_tensor(key) - shape = tuple(tensor.shape) - dtype = str(tensor.dtype) - device = tensor.device if hasattr(tensor, 'device') else 'cpu' - - lora_tag = " [LoRA]" if "lora_A" in key or "lora_B" in key else "" - - print(f"[{dtype:>14}] {shape} | {key} {lora_tag}") +- **词汇表不可修改**：训练中添加新 token 或修改原始词汇表的模型无法导入，须与基础模型词汇表完全一致。 - if __name__ == "__main__": - parser = argparse.ArgumentParser(description="Print structure of a .safetensors LoRA adapter.") - parser.add_argument("filepath", type=str, help="Path to the .safetensors file") - args = parser.parse_args() +- **对话模板不可修改**：训练中修改 chat\_template 的模型无法导入，须与基础模型默认配置一致。chat\_template 位于 config.json 或 tokenizer\_config.json 的 chat\_template 字段。 - print_safetensor_structure(args.filepath) - ``` +- **视觉模型须冻结 VIT**：视觉语言模型必须冻结 Vision Transformer 部分。若 LoRA 适配器中包含 visual 相关权重参数（即未冻结 VIT），该模型无法导入。 - 判断方法：检查 adapt\_model.safetensors 文件中是否包含 `visual` 相关的权重参数。如果文件中存在以 `visual` 开头的参数键（例如 `visual.encoder.layer.0...`），说明 VIT 部分未被冻结，该模型无法导入。 + +可在导入前运行以下脚本检查 adapter\_model.safetensors 是否含 visual 开头的参数键，以判断 VIT 是否冻结。 + +``` +from safetensors import safe_open +import argparse + +def print_safetensor_structure(file_path): + print(f"Loading safetensor file: {file_path}") + print("="*80) + + with safe_open(file_path, framework="pt") as f: + keys = f.keys() + print(f"Found {len(keys)} tensors in the file:\n") + + for key in sorted(keys): + tensor = f.get_tensor(key) + shape = tuple(tensor.shape) + dtype = str(tensor.dtype) + device = tensor.device if hasattr(tensor, 'device') else 'cpu' + + lora_tag = " [LoRA]" if "lora_A" in key or "lora_B" in key else "" + + print(f"[{dtype:>14}] {shape} | {key} {lora_tag}") + +if __name__ == "__main__": + parser = argparse.ArgumentParser(description="Print structure of a .safetensors LoRA adapter.") + parser.add_argument("filepath", type=str, help="Path to the .safetensors file") + args = parser.parse_args() + + print_safetensor_structure(args.filepath) +``` + +判断方法：若脚本输出中存在以 visual 开头的参数键（如 visual.encoder.layer.0...），说明 VIT 部分未被冻结，该模型无法导入；若仅含 lora\_A、lora\_B 等 LoRA 相关参数键，则 VIT 已冻结，可正常导入。 + +【截图：ss-safetensors-out-01.png — safetensors 结构检查脚本输出示例，展示冻结 VIT 的 adapter 文件仅含 lora\_A、lora\_B 参数键】 + +### 提交前自动校验 + +在[导入页](https://bailian.console.aliyun.com/#/efm/model_center/import_model)选中模型目录后，系统会自动调用文件校验接口检查目录下模型文件的格式与完整性。校验失败会在目录字段下方显示红标提示并阻断提交，须按提示修正文件后再提交。常见失败原因如缺少必需文件，对应错误码 AvailableModelFileNotFound。 + +【截图：ss-dir-browser-01.png — 模型目录浏览器选中子目录后的文件自动校验结果】 + +## 导入 LoRA 模型 + +完成 OSS 授权并准备好模型文件后，在[百炼控制台·我的模型](https://bailian.console.aliyun.com/#/efm/model_center)页面点击右上角「导入模型」按钮进入创建页，按以下步骤完成导入。 + +1. 在「我的模型」页面，点击右上角的「导入模型」按钮进入导入模型创建页。 - ![冻结VIT的adapter文件示例](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1185949671/p1049184.png) +2. 按下方表格填写模型信息，其中基础模型须与 LoRA 训练时的基座一致，Bucket 须已添加 bailian-datahub-access=read 标签，模型目录选中后系统会自动校验文件。 - ![未冻结VIT的adapter文件示例（包含visual参数）](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1185949671/p1049183.png) +3. 确认信息无误后点击确定提交，系统自动验证文件格式和完整性，通过后开始导入；如需放弃点击取消返回列表页。导入后可在[管理导入的模型](#sec-manage)查看状态。 -## 常见问题 +导入模型表单各字段含义如下： -#### **为什么导入的模型与本地使用 vLLM、SGLang 推理的效果不一致？** +【截图：ss-import-form-01.png — 导入模型创建页 7 字段表单全貌】 -百炼平台的推理引擎参数设置可能与您本地使用的推理框架默认值不同。为确保效果一致，建议在调用 API 时调整以下参数： +**字段** -参数名称 +**说明** -推荐值（对应 vLLM 默认值） +**约束** -temperature +模型名称 -取值范围：\[0, 2)。设置为 1.0 等同于 vLLM 引擎默认值。 +输入模型的显示名称。 -top\_p +必填，最多 50 字符 -取值范围：(0, 1.0\]。设置为 1.0 等同于 vLLM 引擎默认值。 +基础模型 -top\_k +选择 LoRA 训练时的基座模型，须与训练基座一致。 -取值为 None 或大于 100 时，表示不启用 top\_k 策略，此时仅有 top\_p 策略生效。设置为 99 不支持全采样，该值接近 vLLM 默认值 0（全采样）。 +必填，下拉选择 -presence\_penalty +训练方式 -取值范围：\[-2.0, 2.0\]。设置为 0 等同于 vLLM 引擎默认值。 +可选项取决于所选基础模型，选择基础模型后自动渲染并默认选中第一项。 -repetition\_penalty（DashScope 协议） +必填，下拉选择 -提高 repetition\_penalty 可以降低模型生成的重复度，1.0 表示不做惩罚。取值范围：大于 0。设置为 1.0 等同于 vLLM 引擎默认值。 +导入来源 -说明：以上参数值基于 vLLM 引擎的默认配置。如果您的本地环境使用 SGLang 或其他推理框架，请参考对应框架的文档调整参数。 +当前仅支持「从 OSS 导入」，无其他选项。 -**首次从 OSS 向阿里云百炼导入文件，应该如何操作？** +只读，默认选中 -如果您是首次从 OSS 向阿里云百炼导入文件，请先按照界面提示完成授权，并为目标 OSS Bucket 添加`bailian-datahub-access`标签，然后再进行导入。 +Bucket -> 如果您尚不清楚主账号和子账号的概念和区别，请先阅读[权限管理](https://help.aliyun.com/zh/model-studio/permission-management-overview)。 +选择存放模型文件的 OSS Bucket，仅列出已添加 bailian-datahub-access=read 标签的 Bucket。 -## 使用主账号 +必填，下拉选择 -1. 单击**前往授权**。 - - 在**导入来源**中选择**从OSS导入**后，页面会显示"您还未授权OSS"的提示信息，在提示栏右侧找到 **前往授权** 链接。 - -2. 在弹出的对话框中，单击**确认授权**，系统将为您自动开通[OSS服务关联角色](https://help.aliyun.com/zh/model-studio/bailian-service-linked-role#32a41eac73z64)（必要条件）。 - - > 通常秒级生效，服务高峰期可能会稍有延迟。 - -3. 为目标 OSS Bucket 添加`bailian-datahub-access`标签。 - - > 该标签用于标记阿里云百炼可访问的 Bucket，未标记的 Bucket 阿里云百炼无法访问。 - - 1. 访问[OSS管理控制台](https://oss.console.aliyun.com/)，单击左侧导航栏中的****Bucket 列表****，即可查看您已创建的Bucket。 - - 2. 在待添加标签的Bucket**标签**列，悬停鼠标于![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1134056371/p903505.png)图标上，然后单击**前往编辑**。 - - 3. 单击**创建标签**。 - - 4. 单击**标签**，添加标签名为`bailian-datahub-access`，标签值为`read`的标签，然后单击**保存**。 - -4. 返回**导入模型**界面，重新选择目标 Bucket 再尝试导入。 - - > **请注意，阿里云百炼不支持访问保存在 Bucket 根目录下的文件。**请您选择 Bucket 下的现有文件夹或新建一个文件夹供阿里云百炼访问。 - +模型目录 -## 使用**子账号** +在选定 Bucket 中浏览并选择模型 Checkpoint 所在子目录，不支持选 Bucket 根目录。 -1. 单击**前往授权**。 - -2. 在弹出的对话框中，单击**确认授权**。界面会提示**授权失败**、**当前用户没有创建服务关联角色的权限**（因为当前子账号没有创建服务关联角色的权限。接下来需要先授予子账号创建服务关联角色的权限，再授予子账号通过阿里云百炼访问OSS的权限）。 - - 对话框中显示 Service Name 为 `datahub.sfm.aliyuncs.com`，服务关联角色名称为 `AliyunServiceRoleForSFMDataHubOSSImport`，执行该操作所需的用户权限为 `ram:CreateServiceLinkedRole`。 - -3. 授予子账号创建服务关联角色的权限。 - - 1. **需主账号登录**[RAM控制台](https://ram.console.aliyun.com/)，在左侧导航栏，选择**权限管理** > **权限策略**，然后单击界面上的**创建权限策略**。 - - 2. 在**脚本编辑**的`Effect`、`Action`、`Resource`、`Condition`中分别输入以下脚本中的对应内容后，单击**确定**。 - - ``` - { - "Action": [ - "ram:CreateServiceLinkedRole" - ], - "Resource": "*", - "Effect": "Allow", - "Condition": { - "StringEquals": { - "ram:ServiceName": "datahub.sfm.aliyuncs.com" - } - } - } - ``` - - 3. 输入权限策略名称后，单击**确定**。 - - 本示例中，权限策略名称为`服务关联角色`。 - - 4. 在左侧导航栏，选择**身份管理** > **用户**。在页面列表中找到待授权的子账号，然后单击子账号**操作**列的**添加权限**。 - - 5. 在权限策略中选择刚才创建的权限策略（自定义策略），单击**确认新增授权**。至此，子账号拥有了创建服务关联角色的权限。 - -4. 授权子账号通过阿里云百炼访问OSS。 - - 1. 返回**导入模型**界面，单击**前往授权**。 - - 在**导入来源**中选择**从OSS导入**后，界面提示**您还未授权OSS**。 - - 2. 在弹出的对话框中，单击**确认授权**，系统将为您自动开通[OSS服务关联角色](https://help.aliyun.com/zh/model-studio/bailian-service-linked-role#32a41eac73z64)（必要条件）。 - - > 通常秒级生效，服务高峰期可能会稍有延迟。 - -5. 为目标 OSS Bucket 添加`bailian-datahub-access`标签。 - - > 该标签用于标记阿里云百炼可访问的 Bucket，未标记的 Bucket 阿里云百炼无法访问。 +必填，树形选择 + +模型加密 + +平台自动为导出的模型文件开启 OSS 服务端加密（SSE-OSS），使用 OSS 完全托管密钥，加密算法为 AES256。 + +只读，平台强制 + +若尚未完成 OSS 授权，确定按钮不可用，须先完成[首次导入前完成 OSS 授权](#sec-authorize-oss)后再提交。导入成功后即可部署，部署操作详见[部署运维](#)。 + +导入后的模型状态包括创建中（正在导入）、创建成功（可部署）、创建失败（导入失败）和已失效（源文件已变更）。 + +## 管理导入的模型 + +导入提交后返回[我的模型](https://bailian.console.aliyun.com/#/efm/model_center)列表页，可查看与管理所有导入的模型。列表展示模型名称、模型 ID（附复制按钮）、基础模型、来源、支持部署方式、状态、创建时间与操作列，右上角提供导入模型入口与刷新按钮。 + +【截图：ss-my-model-list-01.png — 我的模型列表页，含名称、基础模型、来源、部署方式、状态、创建时间、操作列】 + +### 状态与流转 + +模型状态包括创建中、创建成功、创建失败和已失效。创建中表示正在导入；创建成功表示导入完成可部署；创建失败表示导入未成功；已失效表示创建成功后 OSS 源模型文件发生变更。列表对处于创建中状态的模型每 3 秒自动静默刷新，属正常行为，非接口异常。 + +创建失败状态旁附「详情」链接，悬停可查看失败错误码（如 AvailableModelFileNotFound）与对应的 oss://bucket/path 路径，用于定位失败文件。 + +已失效状态可悬停查看弹出框，展示「如下文件检测到更新」及发生变更的源文件名列表，提示须重新导入。 + +【截图：ss-status-popover-01.png — 已失效状态气泡弹出框文件变更列表与创建失败悬浮提示错误码】 + +### 操作列可用性 + +- **部署**：仅创建成功状态可点击，点击后跳转部署创建页，部署操作详见[部署运维](#)；其余状态或无可选部署方式时按钮不可用。 - 1. 访问[OSS管理控制台](https://oss.console.aliyun.com/)，单击左侧导航栏中的****Bucket 列表****，即可查看您已创建的Bucket。 - - 2. 在待添加标签的Bucket**标签**列，悬停鼠标于![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1134056371/p903505.png)图标上，然后单击**前往编辑**。 - - 3. 单击**创建标签**。 - - 4. 单击**标签**，添加标签名为`bailian-datahub-access`，标签值为`read`的标签，然后单击**保存**。 - -6. 返回**导入模型**界面，重新选择目标 Bucket 再尝试导入。 +- **增量训练**：从 OSS 导入的模型不支持增量训练，按钮不可用。 - > **请注意，阿里云百炼不支持访问保存在 Bucket 根目录下的文件。**请您选择 Bucket 下的现有文件夹或新建一个文件夹供阿里云百炼访问。 +- **删除**：创建中状态不可删除；量化模型须前往模型压缩界面删除。 -**遇到“10041495”报错怎么办？** +在操作列点击删除并确认后，仅删除百炼侧的模型记录，不影响 OSS 源文件。其余状态删除时会调用删除接口清理记录。 -一般是由于主账号尚未开通对象存储服务 OSS，处理步骤： +删除操作不可恢复：仅移除百炼侧的模型记录，须重新导入才能恢复；不会删除 OSS 中的源文件。 -1. 需主账号前往[OSS管理控制台](https://oss.console.aliyun.com/)，按界面指引开通 OSS。 - -2. 返回阿里云百炼**导入模型**界面，再尝试授权。 +列表顶部搜索框可按模型名称筛选模型，支持清空重置。来源列按导入来源渲染：OSS 导入显示 oss://bucket/path，训练任务显示来源任务 ID（已删除则显示「训练任务已删除」），并标注全参、LoRA 或量化标签。通过 API 调用已部署模型详见[使用 API 进行模型部署](#)。 + +所有操作按单个模型进行，不支持批量删除或批量部署。 + +## 常见问题 + +汇总导入过程中常见的问题与处理方式。 + +**导入失败提示 AvailableModelFileNotFound 怎么办？** + +该错误表示模型目录文件校验未通过（格式或完整性问题），不是单纯重传文件可解决。请检查所选目录是否包含齐全合规的 adapter\_model.safetensors、adapter\_config.json、config.json，并确认 rank、词汇表、chat\_template 等约束均满足，修正后重新选择目录提交。 + +**模型状态显示「已失效」是怎么回事？** + +已失效表示该模型创建成功后，OSS 源模型文件发生了变更，属正常检测行为而非故障。将鼠标悬停在已失效状态上可查看发生变更的文件名列表，需重新导入模型方可恢复可用。 + +**遇到「10041495」报错怎么办？** + +一般是由于主账号尚未开通对象存储服务 OSS。须由主账号前往 OSS 管理控制台按界面指引开通 OSS，再返回百炼导入模型界面重新尝试授权。 + +**子账号授权 OSS 失败怎么办？** + +子账号授权失败是因为没有创建服务关联角色的权限，并非真失败。须先由主账号在 RAM 控制台创建自定义权限策略（操作为 ram:CreateServiceLinkedRole，针对服务 datahub.sfm.aliyuncs.com）并授予子账号，再由子账号重新点击授权。OSS 服务关联角色说明详见[OSS 服务关联角色](#)。 + +**Bucket 下拉列表中目标 Bucket 不可选怎么办？** + +这是授权要求而非故障。新授权方式下，未添加 bailian-datahub-access=read 标签的 Bucket 在下拉中不可选。须到 OSS 管理控制台为目标 Bucket 添加该标签后返回导入页重新选择。 + +**我的模型列表每隔几秒自动刷新是故障吗？** + +不是故障。列表检测到有处于创建中状态的模型时，会每 3 秒静默刷新以获取最新状态，无创建中状态时自动停止，属正常行为。 + +**为什么导入的模型与本地使用 vLLM、SGLang 推理的效果不一致？** + +百炼推理引擎的参数默认值可能与本地推理框架不同，并非模型导入有误。为对齐 vLLM 默认值，建议调用时参考下表设置参数；使用 SGLang 等其他框架请参考对应文档调整。该参数对照属部署与调用范畴，详见[使用 API 进行模型部署](#)。 + +**参数名称** + +**推荐值（对应 vLLM 默认值）** + +temperature + +取值范围 \[0, 2)，设置为 1.0 等同于 vLLM 引擎默认值。 + +top\_p + +取值范围 (0, 1.0\]，设置为 1.0 等同于 vLLM 引擎默认值。 + +top\_k + +取值为 None 或大于 100 时不启用 top\_k 策略，仅 top\_p 生效；设置为 99 不支持全采样，接近 vLLM 默认值 0（全采样）。 + +presence\_penalty + +取值范围 \[-2.0, 2.0\]，设置为 0 等同于 vLLM 引擎默认值。 + +repetition\_penalty（DashScope 协议） + +提高可降低生成重复度，1.0 表示不惩罚；取值范围大于 0，设置为 1.0 等同于 vLLM 引擎默认值。 + +**删除导入的模型会影响 OSS 中的源文件吗？** + +不会。删除仅移除百炼侧的模型记录，需重新导入方可恢复；OSS 源文件归您所有，百炼仅通过 bailian-datahub-access=read 标签读取访问，删除模型不会改动 OSS 中的任何文件。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md index 37f2cbeb..230abbfc 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md @@ -1,6 +1,6 @@ # 预置吞吐长输入与缓存 -本文介绍 PTU（预置吞吐）部署的长输入和前缀缓存能力，包括额度消耗规则、容量计算器使用方法和 API 响应字段说明。 +本文介绍 PTU（预置吞吐）部署的长输入和前缀缓存能力，包括额度消耗规则、**预置吞吐额度计算器**使用方法和 API 响应字段说明。 ## 功能概述 @@ -12,12 +12,12 @@ PTU 部署支持长输入请求（部分模型最高 200K token）和前缀缓 - 前缀缓存优惠：部分模型支持前缀缓存，命中缓存的输入 token 按折扣系数消耗额度（具体折扣率因模型而异），可降低多轮对话和重复前缀场景的额度消耗。 -- 自动转按量计费：超出 PTU 额度或输入超过模型上限（千问 128K / DeepSeek 64K）时，请求自动转为按量计费，无需修改调用代码。 +- 溢出策略：创建 PTU 时可选——自动溢出至按量计费（默认，业务不中断）或仅使用 PTU 容量（超出返回 429、不产生额外费用）。输入超过模型上限（千问 128K / DeepSeek 64K）仍自动转为按量计费。 **重要** -自动转按量计费后，费用按对应模型的按量付费单价计算。建议通过容量计算器合理规划 PTU 额度，避免意外费用。 +自动溢出策略下转为按量计费后，费用按对应模型的按量付费单价计算（仅使用 PTU 容量策略下超出返回 429、不产生额外费用）。建议通过**预置吞吐额度计算器**合理规划 PTU 额度，避免意外费用。 常见于长文档分析（合同、研报摘要）和多轮对话（客服、编程助手）等输入超 32K token 的场景。 @@ -44,6 +44,8 @@ glm-5.1 \[0, 32K)：输入 1.0 / 输出 1.0 \[32K, 200K\]：输入 1.33 / 输出 1.17 + + deepseek-v4-pro @@ -87,17 +89,17 @@ qwen3.7-plus-2026-05-26 输入合计 = 31.940 KTPM（比无缓存节省 43%） ``` -## 使用容量计算器估算额度 +## 使用**预置吞吐额度计算器**估算额度 **说明** 建议在创建或扩容前使用计算器评估长输入场景的额度需求，避免额度不足导致请求转为按量计费。购买上限以控制台实际展示为准。 -前提条件：已开通百炼服务并具备 PTU 部署权限。登录[百炼控制台](https://bailian.console.aliyun.com/#/efm/model_deploy/create)，在**模型部署** > **创建部署**页面（或在已有部署详情页单击**扩容**），选择可部署的PTU（预置吞吐）模型后，展开**容量计算器**。 +前提条件：已开通百炼服务并具备 PTU 部署权限。登录[百炼控制台](https://bailian.console.aliyun.com/#/efm/model_deploy/create)，在**模型部署** > **创建部署**页面（或在已有部署详情页单击**扩容**），选择可部署的PTU（预置吞吐）模型后，展开**预置吞吐额度计算器**。 ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4645961871/p1082157.png) -容量计算器根据业务负载自动推荐 TPM 额度。填写以下参数后，计算器输出推荐的输入 TPM 和输出 TPM。 +**预置吞吐额度计算器**根据业务负载自动推荐额度。填写以下参数后，计算器输出建议购买的输入 KTPM 和输出 KTPM。 **参数** @@ -109,25 +111,25 @@ qwen3.7-plus-2026-05-26 业务高峰期每分钟的请求数。 -RPM 越大，建议购买的输入和输出 TPM 同比增大。 +RPM 越大，建议购买输入 KTPM 和输出 KTPM 同比增大。 平均输入长度（token） 每条请求的平均输入 token 数。 -输入越长，所处阶梯越高，系数越大，建议购买的输入 TPM 越高。不同模型的阶梯边界不同，以控制台实际展示为准。 +输入越长，所处阶梯越高，系数越大，建议购买输入 KTPM 越高。不同模型的阶梯边界不同，以控制台实际展示为准。 平均输出长度（token） 每条请求的平均输出 token 数。 -输出越长，系数可能越大，建议购买的输出 TPM 越高。 +输出越长，系数可能越大，建议购买输出 KTPM 越高。 -预估缓存命中率（%） +缓存命中率（%） 请求中重复前缀被缓存命中的比例。实际命中率取决于请求内容的重复程度，以运行结果为准。 -命中率越高，输入容量消耗越慢，建议购买的输入 TPM 越低。仅影响输入 TPM，不影响输出 TPM。 +命中率越高，输入容量消耗越慢，建议购买输入 KTPM 越低。仅影响输入 KTPM，不影响输出 KTPM。 ## API 响应字段说明 @@ -271,7 +273,7 @@ PTU 部署的运行监控通过百炼平台的模型监控功能实现，支持 - Token 用量与缓存命中：包含 `cached_tokens` 数据系列，可查看缓存命中量占总输入的比例。 -- 配额内/外调用次数：了解超出 PTU 额度后转为按量计费的请求占比。 +- 配额内/外调用次数：了解超出 PTU 额度后的请求占比（自动溢出策略下转为按量计费，仅使用 PTU 容量策略下返回 429）。 更多监控指标和操作方式，请参见[模型监控](https://help.aliyun.com/zh/model-studio/model-telemetry)。 @@ -280,7 +282,7 @@ PTU 部署的运行监控通过百炼平台的模型监控功能实现，支持 **Q: 超出 PTU 额度时会怎样？** -请求自动转为按量计费。API 响应中 `service_tier` 字段不返回或返回 `default`，同时响应头包含 `x-dashscope-ptu-overflow:true`。业务不会中断。 +取决于创建时选择的溢出策略：「自动溢出」策略下，请求自动转为按量计费，API 响应中 `service_tier` 字段不返回或返回 `default`，同时响应头包含 `x-dashscope-ptu-overflow:true`，业务不会中断；「仅使用 PTU 容量」策略下，超出请求返回 429 错误，不产生额外费用。 **Q: 单次输入超过模型上限时会怎样？** @@ -296,4 +298,4 @@ PTU 部署的运行监控通过百炼平台的模型监控功能实现，支持 **Q: 利用率为什么超过 100%？** -部分模型（如 glm-5.1）的长输入阶梯系数使实际额度消耗高于原始 token 数。利用率 = 折算后消耗 ÷ 购买额度。超过 100% 表示消耗速度超过购买额度，超出部分自动转为按量计费，不影响服务可用性。 +部分模型（如 glm-5.1）的长输入阶梯系数使实际额度消耗高于原始 token 数。利用率 = 折算后消耗 ÷ 购买额度。超过 100% 表示消耗速度超过购买额度，超出部分按溢出策略处理（自动溢出则转为按量计费、不影响服务可用性；仅使用 PTU 容量则返回 429）。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md index 4cb4529c..8c5b19fd 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md @@ -96,7 +96,7 @@ BLEU/ROUGE/余弦等算法 ## 创建评测维度模板 -登录[百炼控制台](https://bailian.console.aliyun.com/#/efm/model_evaluate/dimension_template)，在左侧导航栏选择**模型评测** > **评测维度**，单击**创建维度模板**进入创建页面。 +登录[百炼控制台](https://bailian.console.aliyun.com/#/efm/model_evaluate/dimension_template)，在左侧导航栏选择**模型评测** > **评测维度**，单击**创建评测维度**进入创建页面。 创建评测维度模板时，需先填写以下公共字段： @@ -189,7 +189,7 @@ BLEU/ROUGE/余弦等算法 4. 配置**Pass 标签**和**Fail 标签**：定义通过和未通过的分类标签，每个标签不超过 20 个字符。Pass 和 Fail 标签不可重复。 -5. 单击**完成**提交。 +5. 单击**保存**。 ### 大模型评估-数值型 @@ -206,7 +206,7 @@ BLEU/ROUGE/余弦等算法 5. 设置**通过阈值**：评分达到该值及以上判定为 Pass，步长 0.1。阈值随评分范围自动联动。 -6. 单击**完成**提交。 +6. 单击**保存**。 ### 规则评估-字符串匹配 @@ -219,7 +219,7 @@ BLEU/ROUGE/余弦等算法 3. 配置**模型输出**（右侧）：输入待匹配的文本，可使用变量。评测输入和模型输出至少一侧包含变量。 -4. 单击**完成**提交。 +4. 单击**保存**。 ### 规则评估-文本相似度 @@ -234,7 +234,7 @@ BLEU/ROUGE/余弦等算法 4. 设置**通过阈值**：范围 0~1，步长 0.01，相似度达到该值及以上判定为 Pass。 -5. 单击**完成**提交。 +5. 单击**保存**。 ### 人工评估-分类型 @@ -245,7 +245,7 @@ BLEU/ROUGE/余弦等算法 2. 配置**Fail 标签**：定义未通过的标签含义。Pass 和 Fail 标签不可重复。 -3. 单击**完成**提交。 +3. 单击**保存**。 ## 配置评分器Prompt diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md index 9f575021..0b09f108 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md @@ -4,13 +4,13 @@ ## 模型评测概述 -模型评测是百炼平台提供的模型能力评估功能，支持自定义评测和基线评测两种方式，通过评测维度对模型推理结果进行打分和对比，生成评测报告和排行榜。您可以使用预置模型或调优后的模型作为评测对象，量化模型表现并辅助选型决策。当前仅支持文本生成类模型评测。更多功能介绍请参见[模型评测产品概览](#)。 +模型评测是百炼平台提供的模型能力评估功能，支持自定义评测和基线评测两种方式，通过评测维度对模型推理结果进行打分和对比，生成评测报告和排行榜。您可以使用预置模型或调优后的模型作为评测对象，量化模型表现并辅助选型决策。当前仅支持文本生成类模型评测。更多功能介绍请参见[模型评测产品概览](https://help.aliyun.com/zh/model-studio/model-evaluation-introduction/)。 ### 使用场景 - **模型选型对比**：使用相同的评测数据集和维度评测多个候选模型，通过排行榜横向对比综合得分和通过率，用数据驱动选型决策。 -- **调优效果验证**：对模型调优前后分别执行评测，对比评分变化，量化调优带来的能力提升。调优操作请参见[模型调优](#)。 +- **调优效果验证**：对模型调优前后分别执行评测，对比评分变化，量化调优带来的能力提升。调优操作请参见[模型调优](https://help.aliyun.com/zh/model-studio/model-training-overview)。 - **能力量化评估**：生成包含综合得分、通过率和分数分布的评测报告，为团队协作和管理层汇报提供客观数据支撑。 @@ -154,7 +154,7 @@ Function Calling、NL2SQL ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9455233871/p1085617.png) -在[百炼控制台的模型评测页面](https://bailian.console.aliyun.com/#/efm/model_evaluate/dimension_template)，选择**评测维度** Tab，单击**创建评测维度**。各维度类型的完整配置详解也可参见[评测维度](#)。 +在[百炼控制台的模型评测页面](https://bailian.console.aliyun.com/#/efm/model_evaluate/dimension_template)，选择**评测维度** Tab，单击**创建评测维度**。各维度类型的完整配置详解也可参见[评测维度](https://help.aliyun.com/zh/model-studio/evaluation-metrics)。 ### 大模型评估 @@ -184,7 +184,7 @@ Function Calling、NL2SQL ### 选择评测模型 -从控制台模型列表中选择要评测的目标模型。支持预置模型和调优后的模型，具体支持的模型请参见[预置模型列表](#)。 +从控制台模型列表中选择要评测的目标模型。支持预置模型和调优后的模型，具体支持的模型请参见[预置模型列表](https://help.aliyun.com/zh/model-studio/model-deployment-introduction)。 ### 配置数据来源 @@ -451,7 +451,7 @@ Function Calling、NL2SQL ## 计费说明 -模型评测的费用由两部分构成：被评测模型的推理费用和裁判模型的评分费用。具体模型单价请参见[预置模型列表](#)中的定价说明。 +模型评测的费用由两部分构成：被评测模型的推理费用和裁判模型的评分费用。具体模型单价请参见[预置模型列表](https://help.aliyun.com/zh/model-studio/model-deployment-introduction)中的定价说明。 不同维度类型的费用构成如下： @@ -520,7 +520,7 @@ Function Calling、NL2SQL 首先检查评分器 Prompt 是否有明确的评分标准和正确的变量引用。模糊的评分标准会导致裁判模型评分集中在某个分数段。建议为每个分数档提供清晰的判定条件描述。 -其次确认裁判模型的推理能力是否足够（推荐千问-Max）。LLM 评分器存在位置偏差和自我偏好偏差，建议定期人工抽查校准。优化评分器 Prompt 可参考[Prompt 最佳实践](#)。 +其次确认裁判模型的推理能力是否足够（推荐千问-Max）。LLM 评分器存在位置偏差和自我偏好偏差，建议定期人工抽查校准。优化评分器 Prompt 可参考[Prompt 最佳实践](https://help.aliyun.com/zh/model-studio/prompt-engineering-guide)。 **问题**：创建评测维度时选错了维度类型，想修改但找不到入口。 @@ -544,4 +544,4 @@ Function Calling、NL2SQL **问题**：评测发现模型在特定场景表现不足。 -可对模型进行调优后重新评测验证效果（请参见[模型调优](#)）。如果模型缺失特定领域知识，可引入知识库增强模型能力（请参见知识库）。 +可对模型进行调优后重新评测验证效果（请参见[模型调优](https://help.aliyun.com/zh/model-studio/model-training-overview)）。如果模型缺失特定领域知识，可引入知识库增强模型能力（请参见知识库）。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/image-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/image-model.md index 5e614b57..eee5b087 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/image-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/image-model.md @@ -43,20 +43,22 @@ FLUX.2 - 写实人像和产品照片 -### 何时使用qwen-image-2.0-pro +### 何时使用Qwen Image - 需要使用负向提示词排除输出中的特定元素 - 需要每次调用生成最多6张图片变体（Wan标准模式最多支持4张） +- 需要复杂版面生成、小字精准渲染或多语言字体支持 → 选择 qwen-image-3.0-pro（邀测中） + ## 图片编辑 推荐使用`wan2.7-image-pro`，它支持多图参考（最多9张输入图片）、边界框交互式编辑以及角色一致性多图生成。详细使用方法请参见[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide)和[图像编辑-万相2.7/2.6/2.5](https://help.aliyun.com/zh/model-studio/wan-image-edit)。 -### 何时使用qwen-image-2.0-pro +### 何时使用Qwen Image -如果编辑时需要使用负向提示词，请使用`qwen-image-2.0-pro`（生成和编辑使用同一个模型ID）。 +如果编辑时需要使用负向提示词，请使用`qwen-image-3.0-pro`（邀测中）或`qwen-image-2.0-pro`（生成和编辑使用同一个模型ID）。 ## 推荐模型 @@ -108,9 +110,9 @@ FLUX.2 2048x2048 -`qwen-image-2.0-pro` +`qwen-image-3.0-pro` -负向提示词、最多6张图片变体 +复杂版面生成、小字渲染、多语言字体 支持 @@ -272,6 +274,16 @@ qwen-image-2.0-pro的快速版本 **最大分辨率** +`qwen-image-3.0-pro` + +支持 + +支持 + +6 + +2048x2048 + `qwen-image-2.0-pro` 支持 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/s2s-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/s2s-model.md index 9a3c0b1a..dc841b66 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/s2s-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/s2s-model.md @@ -14,13 +14,13 @@ OpenAI GPT Realtime、Gemini 3.1 Live -`qwen3.5-omni-plus-realtime` +`qwen-audio-3.0-realtime-plus` 成本敏感对话 OpenAI gpt-4o-mini Realtime -`qwen3.5-omni-flash-realtime` +`qwen-audio-3.0-realtime-flash` 实时翻译 / 同传 @@ -74,7 +74,7 @@ Gemini 3.1 Live ## 实时还是文件模式？ -- **实时（WebSocket）**：适用于语音助手、呼叫中心、同声传译等实时语音交互场景。音频流式输入，语音流式输出。模型名称中包含`-realtime`。 +- **实时（WebSocket）**：适用于语音助手、呼叫中心、同声传译等实时语音交互场景。音频流式输入，语音流式输出。 - **文件模式（HTTP）**：可以用延迟换取更好的效果，适用于视频配音、播客翻译、离线内容处理等场景。文件模式下还支持 Function Calling、联网搜索、思考模式、视频上下文等附带能力（详见下方“S2S 单模型的附带能力”）。 @@ -91,13 +91,13 @@ Gemini 3.1 Live 语音助手 / 客服对话 -`qwen3.5-omni-plus-realtime` +`qwen-audio-3.0-realtime-plus` WebSocket 成本敏感的对话 -`qwen3.5-omni-flash-realtime` +`qwen-audio-3.0-realtime-flash` WebSocket @@ -127,27 +127,27 @@ WebSocket ## S2S 单模型的附带能力 -以下能力由 Qwen3.5-Omni / Qwen3-Omni 模型在 S2S 单模型路线下直接提供。Pipeline 路线中，对应能力需要由其中的 LLM 等组件分别支持。 +以下能力由 Qwen3.5-Omni / Qwen3-Omni 模型在 S2S 单模型路线下直接提供；其中 Function Calling 也可由 Qwen-Audio Realtime 提供。Qwen-Audio Realtime 不支持联网搜索和思考模式。Pipeline 路线中，对应能力需要由其中的 LLM 等组件分别支持。 ### Function Calling -让模型根据听到和看到的内容执行操作 -- 查询知识库、查询日程、触发工作流。使用Qwen3.5 Omni（WebSocket与HTTP模式） 或 Qwen3 Omni（HTTP模式）。 +让模型根据听到和看到的内容执行操作 -- 查询知识库、查询日程、触发工作流。使用 Qwen3.5 Omni（WebSocket 与 HTTP 模式）、Qwen3 Omni（HTTP 模式）或 Qwen-Audio Realtime（WebSocket 模式）。 **说明** -Qwen3.5-Omni/Qwen3-Omni实时（WebSocket）模式和Livetranslate模型不支持此功能。Qwen-Audio Realtime（WebSocket）支持Function Calling。 +Qwen3.5-Omni / Qwen3-Omni 实时（WebSocket）模式和 Livetranslate 模型不支持此功能。 ### 联网搜索 -让模型检索实时信息，回答关于时事、股价、天气等问题。使用Qwen3.5 Omni（HTTP和WebSocket），包括Plus和Flash系列。模型自主决定是否搜索。 +让模型检索实时信息，回答关于时事、股价、天气等问题。使用 Qwen3.5 Omni（HTTP 和 WebSocket），包括 Plus 和 Flash 系列。模型自主决定是否搜索。Qwen-Audio Realtime 不支持此功能。 **说明** -Qwen3-Omni-Flash和Livetranslate模型不支持此功能。 +Qwen3-Omni-Flash 和 Livetranslate 模型不支持此功能。 ### 思考模式 -当回答质量比延迟更重要时，使用Qwen3 Omni（HTTP模式）。模型在回复前会逐步推理，适用于视频分析、批量打标等场景。 +当回答质量比延迟更重要时，使用 Qwen3 Omni（HTTP 模式）。模型在回复前会逐步推理，适用于视频分析、批量打标等场景。Qwen-Audio Realtime 不支持此功能。 **说明** @@ -157,7 +157,7 @@ Qwen3-Omni-Flash和Livetranslate模型不支持此功能。 以下模型系列均支持语音翻译： -- **Qwen3.5-Livetranslate**：支持 60 种语言互译，其中 29 种支持音频+文本输出、31 种仅支持文本输出，覆盖中文、英语、法语、德语、俄语、日语、韩语、西班牙语、葡萄牙语、阿拉伯语等主流语种。。 +- **Qwen3.5-Livetranslate**：支持 60 种语言互译，其中 29 种支持音频+文本输出、31 种仅支持文本输出，覆盖中文、英语、法语、德语、俄语、日语、韩语、西班牙语、葡萄牙语、阿拉伯语等主流语种。 - **Qwen3-Livetranslate**：支持18种语言 + 5种中文方言，约3秒延迟，开箱即用。文件模式支持输入视频以获得上下文感知的翻译精度。其中7种语言仅输出文本（不输出语音）。 @@ -992,6 +992,50 @@ HTTP ## 所有模型 +### Qwen-Audio + +**模型** + +**API** + +**输入** + +**Function Calling** + +**联网搜索** + +**思考模式** + +**翻译** + +`qwen-audio-3.0-realtime-plus` + +WebSocket + +音频、文本 + +支持 + +\-- + +\-- + +\-- + +`qwen-audio-3.0-realtime-flash` + +WebSocket + +音频、文本 + +支持 + +\-- + +\-- + +\-- + ### Qwen3.5-Omni **模型** @@ -1256,50 +1300,6 @@ HTTP 18 -### Qwen-Audio - -**模型** - -**API** - -**输入** - -**Function Calling** - -**联网搜索** - -**思考模式** - -**翻译** - -`qwen-audio-3.0-realtime-plus` - -WebSocket - -音频、文本 - -支持 - -\-- - -\-- - -\-- - -`qwen-audio-3.0-realtime-flash` - -WebSocket - -音频、文本 - -支持 - -\-- - -\-- - -\-- - ### 旧版模型 以下模型不再更新，新项目建议使用Qwen3.5-Omni。 @@ -1356,6 +1356,8 @@ WebSocket 选定模型后，参考对应的调用文档： +- Qwen-Audio Realtime（WebSocket，实时语音对话）→ [实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides) + - Qwen3.5-Omni / Qwen3-Omni（WebSocket，实时）→ [实时（Qwen-Omni-Realtime）](https://help.aliyun.com/zh/model-studio/realtime) - Qwen3.5-Omni / Qwen3-Omni（HTTP，文件）→ [非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni) @@ -1363,5 +1365,3 @@ WebSocket - Qwen3.5-Livetranslate（WebSocket，实时）→ [实时语音/音视频翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-5-livetranslate-flash-realtime) - Qwen3-Livetranslate（HTTP，文件）→ [音视频文件翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-livetranslate-flash) - -- Qwen-Audio Realtime（WebSocket，实时语音对话）→ [实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/text-generation-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/text-generation-model.md index 560fb0a8..35e48604 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/text-generation-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/text-generation-model.md @@ -4,7 +4,7 @@ ## 做 AI 编程或 Agent 开发（OpenClaw、Claude Code、Hermes 等）该选哪个模型？ -推荐 `qwen3.7-plus`——能力与成本均衡，完整工具调用支持，1M 上下文适合大型代码库。如需最强推理能力，可选择 `qwen3.7-max`。 +推荐 `qwen3.7-plus`——能力与成本均衡，完整工具调用支持，1M 上下文适合大型代码库。如需最强推理能力，可选择`qwen3.8-max-preview`（ Token Plan 可用），或 `qwen3.7-max`。 ## 从闭源模型迁移到百炼? @@ -18,7 +18,7 @@ GPT-5.5、Claude Opus 4.7、Gemini 3.1 Pro -`qwen3.7-max` +`qwen3.7-max`、`qwen3.8-max-preview`（仅 Token Plan 可用） 平衡 @@ -30,19 +30,19 @@ GPT-5.4、Claude Sonnet 4.6、Gemini 3 Pro GPT-5.4-mini、Claude Haiku 4.5、Gemini 3.1 Flash -`qwen3.6-flash`、`deepseek-v4-flash`、`MiniMax-M2.5` +`qwen3.7-flash`、`deepseek-v4-flash`、`MiniMax-M2.5` ## 应用场景 -聊天机器人、内容生成、摘要总结、文档处理等场景，推荐使用 `qwen3.7-plus`，能力与成本均衡，拥有100万上下文窗口和完整的内置工具。确认效果满足需求后，可以尝试 `qwen3.6-flash` 来降低成本，效果接近旗舰模型，且拥有相同的上下文长度和功能支持。如需最强推理能力，可选择 `qwen3.7-max`（百万 token 上下文），但成本较高。 +聊天机器人、内容生成、摘要总结、文档处理等场景，推荐使用 `qwen3.7-plus`，能力与成本均衡，拥有100万上下文窗口和完整的内置工具。确认效果满足需求后，可以尝试 `qwen3.7-flash` 来降低成本，效果接近旗舰模型，且拥有相同的上下文长度和功能支持。如需最强推理能力，可选择 `qwen3.7-max`（百万 token 上下文）；也可选择 `qwen3.8-max-preview`（ Token Plan 可用）。 ### 办公场景（非编程） 处理日常办公任务（文档撰写、邮件处理、会议纪要整理、数据分析等）时，推荐 `qwen3.7-plus`——能力与成本均衡，拥有 100 万上下文窗口，支持 Function Calling 和内置工具，适合文档摘要、内容生成等办公类任务。 -确认效果满足需求后，可尝试 `qwen3.6-flash` 降低成本，效果接近旗舰模型，且拥有相同的上下文长度和功能支持。 +确认效果满足需求后，可尝试 `qwen3.7-flash` 降低成本，效果接近旗舰模型，且拥有相同的上下文长度和功能支持。 -如需最强推理能力（如复杂数据分析、多步逻辑推演），可选择 `qwen3.7-max`，但成本较高。 +如需最强推理能力（如复杂数据分析、多步逻辑推演），可选择 `qwen3.7-max`，但成本较高；也可选择 `qwen3.8-max-preview`（仅 Token Plan 可用）。 处理超长文档（如同时审阅多份合同、大规模文献梳理）时，推荐 `qwen-long`——上下文窗口达 1000 万 Token，可完整处理大体量文档。 @@ -52,7 +52,7 @@ GPT-5.4-mini、Claude Haiku 4.5、Gemini 3.1 Flash 100万Token约相当于70万个汉字或10本小说。 -- 长文档或大型代码库：`qwen3.7-plus` / `qwen3.6-flash`（100万）。 +- 长文档或大型代码库：`qwen3.7-plus` / `qwen3.7-flash` （100万）。 - 常规任务：128k-256k已足够。 @@ -138,6 +138,22 @@ GPT-5.4-mini、Claude Haiku 4.5、Gemini 3.1 Flash 支持 +`qwen3.7-flash` + +查看快照版本 + +`qwen3.7-flash-2026-07-15` + +1M + +支持 + +支持 + +支持 + +支持 + `qwen3.6-flash` 查看快照版本 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md index 65c53e25..e784e1ba 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md @@ -54,7 +54,7 @@ ElevenLabs Multilingual v3 `qwen-audio-3.0-tts-plus`、`MiniMax/speech-2.8-hd` -`qwen-audio-3.0-tts-flash`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） +`qwen-audio-3.0-tts-plus`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） - **使用标准语音合成**：当内置音色库能满足需求，希望快速上手、无需额外配置时。 @@ -89,7 +89,7 @@ ElevenLabs Multilingual v3 推荐模型 -`qwen-audio-3.0-tts-flash`、`MiniMax/speech-2.8-hd` +`qwen-audio-3.0-tts-plus`、`qwen-audio-3.0-tts-flash`、`MiniMax/speech-2.8-hd` `cosyvoice-v3.5-plus`、`cosyvoice-v3.5-flash` @@ -145,7 +145,7 @@ Qwen-Audio-TTS WebSocket / HTTP -不支持 +支持 不支持 @@ -205,7 +205,7 @@ HTTP WebSocket / HTTP -不支持 +支持 不支持 @@ -225,7 +225,7 @@ WebSocket / HTTP - 系统音色（因音色而异）：中文（普通话）、英文 -- 声音复刻音色（方言通过指令控制功能进行设置）：中文（普通话、广东话、重庆话、东北话、甘肃话、贵州话、浙江话、河北话、河南话、湖北话、湖南话、江西话、宁波话、宁夏话、青岛话、陕西话、山西话、山东话、上海话、四川话、云南话）、英文、日语、韩语、德语、法语、意大利语、俄语、葡萄牙语、泰语、印尼语、马来语、越南语 +- 声音复刻音色（方言通过指令控制功能进行设置）：中文（普通话、广东话、重庆话、东北话、甘肃话、贵州话、浙江话、河北话、河南话、湖北话、湖南话、江西话、宁波话、宁夏话、青岛话、陕西话、山西话、山东话、上海话、四川话、云南话）、英语、日语、韩语、俄语、法语、德语、葡萄牙语、泰语、印尼语、越南语、西班牙语、意大利语、马来西亚语、菲律宾语、阿拉伯语 ### CosyVoice diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/video-generate-edit-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/video-generate-edit-model.md index 59bdcc23..6cd607e5 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/video-generate-edit-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/video-generate-edit-model.md @@ -156,8 +156,6 @@ Seedance 2.0、Runway Gen-4 ### HappyHorse 1.1 -以下模型适用于中国内地和国际部署范围。 - **模型ID** **类型** @@ -192,8 +190,6 @@ Seedance 2.0、Runway Gen-4 ### HappyHorse 1.0 -以下模型适用于中国内地和国际部署范围。 - **模型ID** **类型** @@ -236,8 +232,6 @@ Seedance 2.0、Runway Gen-4 ### Wan 2.7 -以下模型适用于中国内地和国际部署范围。 - **模型ID** **类型** @@ -304,8 +298,6 @@ Seedance 2.0、Runway Gen-4 ### Wan 2.6 -以下模型适用于中国内地和国际部署范围。 - **模型ID** **类型** @@ -372,8 +364,6 @@ Seedance 2.0、Runway Gen-4 ### Wan 2.5 -以下模型适用于中国内地和国际部署范围。 - **模型ID** **类型** @@ -400,8 +390,6 @@ Seedance 2.0、Runway Gen-4 ### Wan 2.2 -以下模型适用于中国内地和国际部署范围。 - **模型ID** **类型** @@ -460,8 +448,6 @@ wan-std / wan-pro 两种模式 ### Wan 2.1（推荐使用Wan 2.7） -以下模型适用于中国内地和国际部署范围。 - **模型ID** **类型** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/vision-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/vision-model.md index da8325e3..9d006ab7 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/vision-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/vision-model.md @@ -28,11 +28,11 @@ GPT-5.4、Claude Sonnet 4.6、Gemini 3 Pro GPT-5.4-mini、Gemini 3.1 Flash -`qwen3.6-flash` +`qwen3.7-flash` ## 图像与视频理解 -推荐从`qwen3.7-plus`开始，它是千问旗舰模型，支持1M上下文、最长2小时视频、Function Calling和内置工具等完整功能。当您的场景稳定后，可以尝试`qwen3.6-flash`来降低成本，它提供接近旗舰的效果，并支持相同的上下文长度和功能集。 +推荐从`qwen3.7-plus`开始，它是千问旗舰模型，支持1M上下文、最长2小时视频、Function Calling和内置工具等完整功能。当您的场景稳定后，可以尝试`qwen3.7-flash`来降低成本，它提供接近旗舰的效果，并支持相同的上下文长度和功能集。 ### 图像分辨率 @@ -40,7 +40,7 @@ GPT-5.4-mini、Gemini 3.1 Flash ### 视频支持 -- 最长2小时 / 2GB：`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`、`qwen3.5-plus`、`qwen3.5-flash` +- 最长2小时 / 2GB：`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.7-flash`、`qwen3.6-flash`、`qwen3.5-plus`、`qwen3.5-flash` - 最长1小时 / 2GB：`qwen3-vl-plus`、`qwen3-vl-flash` @@ -53,7 +53,7 @@ GPT-5.4-mini、Gemini 3.1 Flash - Function Calling：Qwen3.7、Qwen3.6、Qwen3.5和Qwen3-VL系列模型均支持 -- 内置工具（联网搜索、代码执行，无需额外配置）：仅`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`、`qwen3.5-plus`、`qwen3.5-flash` +- 内置工具（联网搜索、代码执行，无需额外配置）：仅`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.7-flash`、`qwen3.6-flash`、`qwen3.5-plus`、`qwen3.5-flash` ### 结构化输出 @@ -64,7 +64,7 @@ Qwen3.7、Qwen3.6、Qwen3.5和Qwen3-VL系列在非思考模式下支持此功能 ## OCR与文档提取 -`qwen3.5-ocr` 专为文档、表格、试卷和手写内容的文字提取而优化。您也可以使用`qwen3.7-plus`或`qwen3.6-flash`进行通用图片文字提取。 +`qwen3.5-ocr`专为文档、表格、试卷和手写内容的文字提取而优化。您也可以使用`qwen3.7-plus`或`qwen3.7-flash`进行通用图片文字提取。 ## 推荐模型 @@ -108,7 +108,7 @@ Qwen3.7、Qwen3.6、Qwen3.5和Qwen3-VL系列在非思考模式下支持此功能 支持 -`qwen3.6-flash` +`qwen3.7-flash` 1M @@ -148,8 +148,6 @@ Qwen3.7、Qwen3.6、Qwen3.5和Qwen3-VL系列在非思考模式下支持此功能 支持 -## 所有模型 - ### Qwen3.7 **模型 ID（Model ID）** @@ -232,6 +230,46 @@ Qwen3.7、Qwen3.6、Qwen3.5和Qwen3-VL系列在非思考模式下支持此功能 支持 +`qwen3.7-flash` + +文本、图像、视频 + +文本 + +1M + +64k + +256 + +64 + +支持 + +支持 + +支持 + +`qwen3.7-flash-2026-07-15` + +文本、图像、视频 + +文本 + +1M + +64k + +256 + +64 + +支持 + +支持 + +支持 + ### Qwen3.6 **模型 ID（Model ID）** @@ -354,6 +392,8 @@ Qwen3.7、Qwen3.6、Qwen3.5和Qwen3-VL系列在非思考模式下支持此功能 支持 +## 旧版模型 + ### Qwen3.5 **模型 ID（Model ID）** @@ -546,57 +586,34 @@ Qwen3.7、Qwen3.6、Qwen3.5和Qwen3-VL系列在非思考模式下支持此功能 #### Qwen3-VL -- `qwen3-vl-plus` - -- `qwen3-vl-plus-2026-01-25` +- `qwen3-vl-plus`及其快照版本 -- `qwen3-vl-flash` - -- `qwen3-vl-flash-2026-01-25` - - -#### Qwen2.5-VL - -- `qwen2.5-vl-72b-instruct` - -- `qwen2.5-vl-32b-instruct` - -- `qwen2.5-vl-7b-instruct` - -- `qwen2.5-vl-3b-instruct` +- `qwen3-vl-flash`及其快照版本 #### Qwen-Omni -- `qwen3-omni-flash` - -- `qwen3-omni-flash-2025-10-22` +- `qwen3-omni-flash`及其快照版本 - `qwen-omni-turbo`及其快照版本 #### Qwen-OCR -- `qwen-vl-ocr` +- `qwen-vl-ocr`及其快照版本 - `qwen-vl-ocr-latest` -- `qwen-vl-ocr-2025-07-14` - #### QVQ - `qvq-max` -- `qvq-max-2025-08-28` - - `qvq-plus` -- `qvq-plus-2025-08-27` - #### 旧版Qwen-VL -- `qwen-vl-max`及其快照版本 +- `qwen-vl-max` -- `qwen-vl-plus`及其快照版本 +- `qwen-vl-plus` diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md index 8e525acd..ecfee151 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md @@ -10,7 +10,7 @@ TPM 预留为指定模型锁定专属推理容量，确保业务高峰期不受 - 专属模型 code：创建 TPM 预留后，系统自动生成专属模型 code，您需要将 API 请求中的 `model` 参数替换为该 code。 -- 超额不中断：超出预留容量的请求自动降级为按量计费处理，无需修改代码。 +- 溢出策略：创建时可选超额处理方式——自动溢出至按量计费（默认，业务不中断）或仅使用预留容量（超出返回 429，不产生额外费用）。 ## **方案对比与选型** @@ -61,7 +61,7 @@ TPM 预留 **流量可预估、不能接受限流** -超出自动降级公共池按量，不中断 +可选：自动溢出按量（默认）/仅预留容量返回429 替换 model 参数即可 @@ -73,7 +73,7 @@ PTU（[模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-in **高吞吐高性能** -超出转按量 +可选：自动溢出按量（默认）/仅PTU容量返回429 替换 model 参数即可 @@ -199,7 +199,7 @@ DeepSeek-v4-Pro `退款 = 降量部分预付费 - (降量部分预付费 × 已用时长/购买时长 × 1.5)` -- 超出保障额度后自动降级为标准按量计费，服务不中断。可在详情页**超额降级统计**中查看降级次数。 +- 溢出策略为「自动溢出」时：超出保障额度自动降级为标准按量计费，服务不中断，可在详情页**超额降级统计**查看降级次数；为「仅使用预留容量」时：超出返回 429，不产生额外费用。 - 服务到期后 2 小时内：实例仍为运行中，可继续调用，可续费；到期后 2~14 小时：实例已停止，不可调用，仍可续费；到期 14 小时后：实例已删除，不可恢复。 @@ -232,6 +232,7 @@ glm-5.1 \[0, 32K)：输入 1.0 / 输出 1.0 \[32K, 200K\]：输入 1.33 / 输出 1.17 + deepseek-v4-pro @@ -357,6 +358,14 @@ Qwen 系列 在输入框中输入天数，取值范围与购买时长一致。 + 溢出策略 + + 预留容量耗尽时，超出部分请求的处理方式。 + + 是 + + 自动溢出至按 token 付费（默认，超出转按量、业务不中断）/ 仅使用预留容量（超出返回 429、不产生额外费用） + 2. 确认参数后单击**立即购买**，在费用确认弹窗中核对费用，单击**确认支付**。 3. 在 TPM 预留详情页的**概览** Tab，找到**专属模型 code**，单击复制。 @@ -451,7 +460,7 @@ RPM 越大，建议购买的输入和输出 TPM 同比增大。 - 使用率趋势：可切换输入/输出方向，展示预留容量线和实际用量。 -- 超额降级统计：展示超出预留容量后被降级处理的次数。 +- 超额降级统计：展示超出预留容量后被降级处理的次数（仅「自动溢出」策略下产生降级）。 #### 监控 @@ -476,6 +485,10 @@ RPM 越大，建议购买的输入和输出 TPM 同比增大。 单击**扩缩容**，在弹窗中调整输入 TPM 和输出 TPM。 +**说明** + +输入 TPM 和输出 TPM 支持调整为 0：归 0 后不再产生容量费用，且专属模型 code 继续保留，避免因到期或退订导致 code 失效。但归 0 属于减配，已使用部分按 1.5 倍系数结算违约金（详见上方计费与使用说明）。 + #### 续费 单击**续订**，选择续费时长并完成支付。如已开启**到期自动续费**，系统在到期前一天 08:00 自动扣款续费。 @@ -524,7 +537,7 @@ RPM 越大，建议购买的输入和输出 TPM 同比增大。 **Q: 超出预留容量时会怎样？** -超出预留容量的请求自动降级为按量计费处理，服务不中断。可在详情页概览 Tab 的**超额降级统计**中查看降级次数和时间。频繁降级时建议扩容。 +取决于创建时选择的溢出策略：「自动溢出」策略下，超出预留容量的请求自动降级为按量计费，服务不中断，可在详情页概览 Tab 的**超额降级统计**查看降级次数和时间，频繁降级时建议扩容；「仅使用预留容量」策略下，超出请求返回 429 错误，不产生额外费用，频繁 429 时建议扩容。 **Q: 专属模型 code 怎么获取？** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-monitoring/model-telemetry.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-monitoring/model-telemetry.md index 27e89c0c..a229ac38 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-monitoring/model-telemetry.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-monitoring/model-telemetry.md @@ -11,9 +11,9 @@ ## 支持的模型 -- **监控：普通监控**支持[选择模型](https://help.aliyun.com/zh/model-studio/models)中的所有模型，包括基于它们调优后的[自定义模型](https://help.aliyun.com/zh/model-studio/model-deployment-introduction#f17bf700c06k5)；**高级监控**支持北京、新加坡、弗吉尼亚地域下的所有模型**。** +- **监控：普通监控**支持[选择模型](https://help.aliyun.com/zh/model-studio/models)中的所有模型，包括基于它们调优后的[自定义模型](https://help.aliyun.com/zh/model-studio/model-deployment-introduction#f17bf700c06k5)；**高级监控**支持北京、上海、新加坡、弗吉尼亚地域下的所有模型**。** -- **告警功能：**支持北京、新加坡地域下的所有模型。 +- **告警功能：**支持北京、新加坡、弗吉尼亚地域下的所有模型。 ## **监控模型运行** @@ -22,9 +22,9 @@ > 列表记录按“模型 + 业务空间”维度生成。新模型在首次数据同步完成后自动加入列表（普通监控的延迟通常为小时级，请耐心等待；如需分钟级的数据洞察，请使用[高级监控](#2e5f2f0dffijg)）。 -列表顶部「监控数据看板」以卡片形式汇总**模型总量**、**总调用次数**、**总失败次数**、**平均调用时长**、**平均首包时长**。 +列表顶部「**监控数据**」以卡片形式汇总**模型总量**、**总调用次数**、**总失败次数**、**平均调用时长**、**平均首包时长**。 -「模型监控」表格列出各模型的**模型 Code**、**业务空间**、**调用总量**、**调用失败量**、**失败率**、**平均调用时长**、**平均首包时长**（除模型 Code、业务空间外均可排序），操作列提供**监控**、**日志**入口。 +「模型监控」表格列出各模型的**模型Code**、**业务空间**、**调用总量**、**调用失败量**、**失败率**、**平均调用时长**、**平均首Token延时**（除模型Code、业务空间外均可排序），操作列提供**监控**、**日志**入口。列表工具栏还提供**日志回流**入口，可将推理日志回流为训练数据集。 > 默认业务空间成员可查看所有业务空间的模型调用情况；子业务空间成员仅能查看当前空间的数据，无法切换查看其他业务空间数据。 @@ -80,7 +80,11 @@ ### **查看某次调用的 Token 消耗** -> 该功能目前仅适用于**华北2（北京）**地域的部分模型。 +> 该功能目前适用于**华北2（北京）**、新加坡地域的部分模型，弗吉尼亚地域同样支持。 + +**说明** + +**数据说明：**推理日志（高级监控）从调用发生到可查询存在分钟级延迟，请耐心等待；普通监控的用量汇总（如调用次数、Token 总量）延迟为小时级，高峰期可能达 1-2 小时。如遇无数据或查不到记录的情况，请先确认已等待足够的数据同步时间。仅记录**开启推理日志后**的调用数据，开通前的历史调用无法追溯。 1. 使用主账号（[或拥有足够权限的子账号](#f9d06146c0xe0)）登录，在目标业务空间的[模型监控（北京）](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)页面，点击右上角的**模型监控配置**，按照指引依次开通审计日志和推理日志。 @@ -88,7 +92,7 @@ 2. 在模型监控列表中找到目标模型，点击其右侧**操作**列的**日志**。 -3. **日志**页签展示该模型的[实时推理](#f131611173sdx)调用记录，**用量**字段即为本次调用的Token消耗。 +3. **日志**页签以表格形式展示该模型的[实时推理](#f131611173sdx)调用记录，表格包含**Request ID/调用时间**、**调用时长**、**状态码**（支持筛选）、**错误码**、**用量**、**请求和响应**、**操作**等列。其中**用量**字段即为本次调用的Token消耗。 ### **创建异常消耗告警** @@ -100,10 +104,14 @@ **重要** -该功能目前仅适用于**华北2（北京）**地域的部分模型。 +该功能目前适用于**华北2（北京）**、新加坡地域的部分模型，弗吉尼亚地域同样支持。 模型监控支持查看模型的每一次对话，包括输入、输出及耗时，是故障排查和内容审计的关键工具。 +**说明** + +**数据说明：**日志从调用发生到可查询存在分钟级延迟，请耐心等待。如遇实时更新延迟或查不到记录的情况，请先确认已等待足够时间（普通监控用量汇总为小时级延迟，高级监控/推理日志为分钟级）。仅记录**开启推理日志后**的调用数据，开通前的历史调用无法追溯。 + ### **步骤一：开通日志** 使用主账号（[或拥有足够权限的子账号](#f9d06146c0xe0)）登录，在目标业务空间的[模型监控（北京）](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)页面，点击右上角的**模型监控配置**，按照指引依次开通审计日志和推理日志。 @@ -116,7 +124,7 @@ 1. 在模型监控列表中找到目标模型，点击其右侧**操作**列的**日志**。 -2. **日志**页签展示该模型的[实时推理](#f131611173sdx)调用记录，**请求和响应**字段分别对应本次调用的输入与输出。 +2. **日志**页签以表格形式展示该模型的[实时推理](#f131611173sdx)调用记录，表格包含**Request ID/调用时间**、**调用时长**、**状态码**（支持筛选）、**错误码**、**用量**、**请求和响应**、**操作**等列。其中**请求和响应**字段分别对应本次调用的输入与输出。 **支持请求和响应的模型** @@ -152,14 +160,20 @@ - 三方模型：deepseek-v3.1、deepseek-v3.2、deepseek-v3.2-exp +并非所有模型都支持推理日志（请求/响应内容记录）。是否支持由模型本身决定，与模型是否为多模态无关。当所选模型不支持时，界面会显示**当前模型暂不支持日志**。 + +请注意：请求和响应内容**仅在开启推理日志后**才会被采集，开通前的历史调用不会补录。若某次调用的输出内容缺失或存在日志缺失，请先确认该模型是否支持推理日志，以及推理日志是否已在调用发生前完成开通。 + ## **建立主动告警** **重要** -该功能目前仅适用于新加坡和华北2（北京）地域。 +该功能目前适用于新加坡、北京和弗吉尼亚地域。 模型的静默失败（如超时、Token消耗突增），传统应用日志难以发现。模型监控支持对监控指标（如成本、失败率、响应延迟）设置告警。一旦指标出现异常，系统立即告警。 +模型告警页面包含**告警规则**和**告警历史**两个页签。**告警历史**页签可按告警时间、告警规则、告警等级、状态筛选查看历史告警记录，点击详情可查看告警详情。 + ### **步骤一：开启高级监控** 1. 使用主账号（[或拥有足够权限的子账号](#54ea9ba526ovz)）登录，在目标业务空间的模型监控（[北京](https://bailian.console.aliyun.com/?tab=model#/model-telemetry) 或 [新加坡](https://modelstudio.console.aliyun.com/?tab=dashboard#/model-telemetry)）页面，点击右上角的**模型监控配置**。 @@ -171,7 +185,7 @@ 1. 在模型告警（[北京](https://bailian.console.aliyun.com/?tab=model#/model-alert) 或[新加坡](https://modelstudio.console.aliyun.com/?tab=dashboard#/model-alert)）页面，点击右上角的**创建告警规则**。 -2. 在对话框中，选择要监控的模型和监控模板，确认无误后点击**创建**。当指定的监控指标（如调用统计或性能指标）出现异常时，系统将通知您的团队。 +2. 在对话框中，选择要监控的模型和监控模板，确认无误后点击**确定**。当指定的监控指标（如调用统计或性能指标）出现异常时，系统将通知您的团队。 - **通知方式：**支持短信、电子邮件、电话、钉钉群机器人、企业微信机器人及Webhook。 @@ -289,12 +303,20 @@ 模型非首包时长p99 + **TPS** + + model\_tps\_per\_request + + 单次请求输出 Token 速度（TPS），每秒生成 Token 数，衡量模型生成速度（仅高级监控支持） + **用量** model\_usage 模型用量总和 + **关于 TPS 指标：**`model_tps_per_request` 仅在**高级监控**中展示，高级监控为收费功能。TPS（每秒生成 Token 数）与非首包时长（每 Token 的平均生成耗时）呈倒数关系（TPS ≈ 1 ÷ 非首包时长均值）。排查响应慢的问题时，建议结合首 Token 延时（TTFT）、非首 Token 延时及输入 Token 量综合分析，单次调用总耗时还受输入长度、网络等因素影响，不能仅凭 TPS 判断。TPS 触发的是按请求维度的限流，区别于 TPM（每分钟 Token 数）的按账号维度限流。 + - **HTTP API：**`{HTTP API}`需替换为前面[步骤一](#title-tkb-ds1-4p5)获取的HTTP API地址。 - **Authorization：**需将阿里云账号的 `AccessKey:AccessKeySecret` 拼接后进行Base64编码，并以 `Basic 编码后字符串` 的形式提供。 @@ -530,6 +552,12 @@ 模型非首包时长p99 + **TPS** + + model\_tps\_per\_request + + 单次请求输出 Token 速度（TPS），每秒生成 Token 数，衡量模型生成速度（仅高级监控支持） + **用量** model\_usage @@ -715,6 +743,12 @@ } ``` +- **API Key 用量限额：**当前不支持为单个 API Key 设置月度或每日 Token 消耗上限并自动停服（即不支持硬性阻断，无法实现额度用尽后自动禁用 API 调用）。如需防止意外欠费，可采用以下替代方案： + + - **告警通知 + 手动禁用 Key：**开启**高级监控**（收费功能）后，可对 Token 消耗配置告警阈值，当消耗超出阈值时接收消费预警通知，再人工介入处理（如手动禁用对应 API Key）。普通监控仅支持基础用量查看，不支持配置告警。 + + - **免费额度用完即停：**适用于仅使用免费配额的场景，开启后免费额度耗尽时自动停止调用。 + ## **计费说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md index a108688f..7c8723b1 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md @@ -33,6 +33,30 @@ **功能说明** +7月21日 + +平台功能 + +记忆库商业化通知 + +记忆库商业化通知[了解详情](https://www.aliyun.com/notice/118464) + +7月16日 + +平台功能 + +企业知识库（旧）下线通知 + +企业知识库（旧）下线通知[了解详情](https://www.aliyun.com/notice/118448) + +7月16日 + +平台功能 + +Managed Agent商业化通知 + +Managed Agent商业化通知[了解详情](https://www.aliyun.com/notice/118456) + 7月14日 平台功能 @@ -177,7 +201,7 @@ Coding Plan Coding Plan 联网搜索 MCP 升级 -Coding Plan 联网搜索 MCP 升级 Streamable HTTP 协议，前 2000 次免费，[了解详情](https://help.aliyun.com/zh/model-studio/web-search-for-coding-plan) +Coding Plan 联网搜索 MCP 升级 Streamable HTTP 协议，前 2000 次免费，[了解详情](https://help.aliyun.com/zh/model-studio/web-search-mcp) 6月10日 @@ -339,7 +363,7 @@ Token Plan Token Plan 团队版团队管理上线 -Token Plan 团队版新增团队管理：支持 SSO/钉钉登录、席位分配、Credits 用量监控，[了解详情](https://help.aliyun.com/zh/model-studio/token-plan-team#tp05-h-enter) +Token Plan 团队版新增团队管理：支持 SSO/钉钉登录、席位分配、Credits 用量监控，[了解详情](https://help.aliyun.com/zh/model-studio/token-plan-team-management#tp05-h-enter) 5月4日 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/newly-released-models.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/newly-released-models.md index 1fffce5d..c2cc4d82 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/newly-released-models.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/newly-released-models.md @@ -8,5036 +8,7260 @@ **时间** -**服务部署范围** - -**模型规格** +**模型ID** **功能说明** -视频对口型 +文本生成、深度思考、视觉理解 -2026-07-15 +2026-07-21 -中国内地 +`qwen3.7-flash` -pixverse/pixverse-lipsync +`qwen3.7-flash-2026-07-15` -爱诗视频对口型模型，输入视频和音频，生成口型与音频同步的视频。 +Qwen3.7原生视觉语言系列Flash模型，相较3.6-Flash全面提升多模态理解与Agent执行能力。重点强化多模态基础能力、万物识别能力更强，真实世界感知与空间智能进一步提升，Search Agent、CI Agent等多模态Agent场景能力显著升级、端到端任务执行更稳定，多模态Coding能力优化、vibe coding 体验更加流畅。 -视频动作模仿 +图片生成 -2026-07-15 +2026-07-20 -中国内地 +`qwen-image-3.0-pro` -pixverse/pixverse-motioncontrol +内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。 细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。 知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。 Qwen-Image-3.0-Pro 不只是在追求"好看"，更在追求“好用”——让图像生成真正成为可落地的生产力工具。 -爱诗视频动作模仿模型，输入视频和参考动作视频，生成模仿参考动作的视频。 +文本生成、深度思考、视觉理解 -视频超清 +2026-07-17 + +`kimi/kimi-k3` + +Kimi K3 是 Kimi 迄今能力最强的旗舰模型，拥有 2.8 万亿参数，基于 KDA 混合线性注意力机制（Kimi Delta Attention）和注意力残差（Attention Residuals）技术构建，原生支持视觉理解，并拥有 100 万 token 上下文窗口。它是全球首个开源的 3 万亿级别模型，面向长程编程、知识工作和推理等前沿智能场景而设计。 + +文本向量 2026-07-15 -中国内地 +`qwen3.7-text-embedding` -pixverse/pixverse-upscale +是通义实验室基于Qwen3.7训练的多语言文本统一向量模型，相较text-embedding-v4版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升20%；支持256~2560维用户自定义向量维度。 -爱诗视频超清模型，将低分辨率视频提升至更高分辨率。 +实时语音合成 + +2026-07-14 + +`qwen-audio-3.0-tts-plus` -实时多模态 +qwen-audio-3.0-tts-plus是面向高质量语音生成场景打造的高性能语音合成大模型。相比前一版本，模型支持更多小语种和中文方言，显著提升方言发音的正宗程度，并增强了 free-style 指令遵循能力和细粒度标签控制能力，可更准确地控制情绪、语气、角色、语速、音量和合成风格。同时，模型在噪声、混响等复杂声学条件下具备更强鲁棒性，进一步提升了音质、清晰度、分辨率和整体表现力。Plus 版本更强调合成效果和细节表现，适用于有更高音质、自然度和表现力要求的专业场景，如内容创作、有声书、影视配音、品牌声音设计和高品质语音服务。 + +实时语音合成 2026-07-14 -中国内地 +`qwen-audio-3.0-tts-flash` -qwen-audio-3.0-realtime-plus、qwen-audio-3.0-realtime-flash +qwen-audio-3.0-tts-flash是面向实时交互场景优化的高性能语音合成大模型。相比前一版本，模型支持更多小语种和中文方言，提升了方言发音的正宗程度，并增强了 free-style 指令遵循能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。同时，模型在噪声、混响等复杂声学条件下具备更强鲁棒性，提升了音质、清晰度和整体表现力。Flash 版本重点优化实时合成体验，首包延时控制在 200ms 以内，适用于语音助手、实时对话、智能客服等低延迟交互场景。 -Qwen-Audio端到端实时语音大模型兼顾语音推理能力与双工对话节奏，在保持流畅、自然的实时交互体验的同时，通过并行推理、全向流式等工程优化，有效控制端到端响应时延。[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides) +实时语音对话 -语音合成 +2026-07-14 + +`qwen-audio-3.0-realtime-plus` + +千问实时语音大模型 是一款登顶全球权威评测的下一代实时双工语音大模型，它在全球权威第三方评测平台 Artificial Analysis Speech-to-Speech子项中取得综合排名第一。千问实时语音大模型 兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现"又快又聪明"的对话体验。标准版更注重高质量的回复结果。 + +实时语音对话 2026-07-14 -中国内地 +`qwen-audio-3.0-realtime-flash` + +千问实时语音对话大模型3.0 是一款登顶全球权威评测的下一代实时双工语音大模型，它在全球权威第三方评测平台 Artificial Analysis Speech-to-Speech子项中取得综合排名第一。千问实时语音对话大模型3.0兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现"又快又聪明"的对话体验。极速版更注重极致的响应速度 -qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash +视频生成 -Qwen-Audio-TTS语音合成模型上线，新增更多小语种和中文方言支持，增强了指令遵循与细粒度标签控制能力，音质和表现力全面提升。其中 Plus 版本面向高品质专业场景，Flash 版本面向低延迟实时交互场景，首包延时控制在 200ms 以内。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +2026-07-14 -文生图/参考生图 +`pixverse/pixverse-upscale` -2026-07-13 +Upscale 支持将不同分辨率的视频超分至 4K，提升画面清晰度与细节表现，适合高质量展示与二次分发。 -中国内地 +视频生成 -vidu/vidu-image\_reference2image、vidu/viduq3-fast\_reference2image、vidu/viduq2-pro\_reference2image、vidu/viduq2-fast\_reference2image +2026-07-14 -由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。 [Vidu-图像生成](https://help.aliyun.com/zh/model-studio/vidu-image-generation-api-reference) +`pixverse/pixverse-motioncontrol` -参考生视频 +Motion Control 支持从参考视频中提取动作，并迁移到目标人物图片上，生成角色复现相同动作的新视频。 -2026-07-13 +视频生成 -中国内地 +2026-07-14 -vidu/viduq3-ad\_reference2video、vidu/viduq3-drama\_reference2video +`pixverse/pixverse-lipsync` -Vidu-参考生视频系列模型（广告、短剧方向），支持传入参考图片和视频，生成对应场景的视频内容。[Vidu-参考生视频](https://help.aliyun.com/zh/model-studio/vidu-reference-to-video-api-reference) +对口型能力可将视频中人物的嘴部动作与输入音频或 TTS 精准同步，让角色说话更自然，提升视频表现力与叙事感染力。 -图生视频 +视频生成 -2026-07-13 +2026-07-09 -中国内地 +`vidu/viduq3-pro-fast_img2video` -vidu/viduq3-pro-fast\_img2video +输入图片与文本描述，生成视频。ViduQ3-Pro-fast生成速度更快，性价比更高；较 ViduQ2-Pro-fast 生成时长 10 秒扩展至 16 秒 ，可实现更复杂的镜头切换与叙事逻辑。 -Vidu-图生视频模型，根据输入图像和文本提示词快速生成视频。[Vidu-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/vidu-image-to-video-api-reference) +图片生成 -参考生视频 +2026-07-09 -2026-07-01 +`vidu/viduq3-fast_reference2image` -中国内地 +输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，主打高速高质与低成本，成本比Pro降低约50%。 -wan2.7-r2v-2026-06-12 +视频生成 -万相2.7参考生视频模型快照版本，支持主体参考和音色定制，并可输入单张多宫格故事板直接生成剧本化视频。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +2026-07-09 -图像生成 +`vidu/viduq3-drama_reference2video` -2026-06-25 +ViduQ3-Drama 是面向精品剧/AI漫剧生产的专用模型，主打"一致性+动效+细节美学+性价比"四大维度的全面升级，效果更稳、情绪更真、动态更强。 -中国内地 +视频生成 -qwen-image-2.0-pro-2026-06-22 +2026-07-09 -Qwen-Image-2.0系列模型最新快照，融合图片生成与编辑能力。相较于4月22日快照，文字渲染能力进一步增强，支持最长1k token的指令输入；真实质感与写实场景细节刻画更加细腻；语义遵循能力更强。[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api)、[千问-图像编辑](https://help.aliyun.com/zh/model-studio/qwen-image-edit-api) +`vidu/viduq3-ad_reference2video` -文生视频 +ViduQ3-Ad是面向广告行业的专用模型，主打"营销级切镜+智能运镜+直出音效"三大能力，上传商品图即可生成16秒广告视频，降低广告视频的创作门槛与制作成本。 -2026-06-22 +图片生成 -中国内地 +2026-07-09 -happyhorse-1.1-t2v +`vidu/viduq2-pro_reference2image` -HappyHorse 1.1系列文生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-文生视频](https://help.aliyun.com/zh/model-studio/happyhorse-text-to-video-api-reference) +输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，擅长处理复杂逻辑，具备超强上下文一致性和工业级稳定性。适合专业设计、漫剧制作等。 -图生视频 +图片生成 -2026-06-22 +2026-07-09 -中国内地 +`vidu/viduq2-fast_reference2image` -happyhorse-1.1-i2v +输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，语义理解能力大幅提升，支持更多风格。 -HappyHorse 1.1系列图生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/happyhorse-image-to-video-api-reference) +图片生成 -参考生视频 +2026-07-09 -2026-06-22 +`vidu/vidu-image_reference2image` -中国内地 +输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，对中英文字的精准渲染、UI/图表等设计细节的像素级还原，适合制作海报、信息图等。 -happyhorse-1.1-r2v +文本生成 -HappyHorse 1.1系列参考生视频模型，支持多参考图输入生成有声视频，可生成3~15秒、720P/1080P视频。[HappyHorse-参考生视频](https://help.aliyun.com/zh/model-studio/happyhorse-reference-to-video-api-reference) +2026-07-09 -推理模型 +`glm-5.2-fast-preview` -2026-06-18 +GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。 -中国内地 +视频生成 -kimi/kimi-k2.7-code-highspeed +2026-07-01 -月之暗面直供的高速编程模型，与 kimi/kimi-k2.7-code 功能完全一致，速度提升5~6倍。[Kimi-月之暗面](https://help.aliyun.com/zh/model-studio/kimi-api-by-moonshot-ai) +`wan2.7-t2v-2026-06-12` -文字提取 +万相2.7-文生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。该版本为2026年6月12日快照。 -2026-06-16 +视频生成 -中国内地 +2026-07-01 -qwen3.5-ocr +`wan2.7-r2v-2026-06-12` -千问文字提取模型，基于Qwen3.5架构，速度更快，效果更强。上下文长度扩展至128K，支持多轮对话。信息抽取能力大幅提升，覆盖多种国内外卡证。[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr) +万相2.7-参考生视频，更加稳定的角色、道具与场景参考，支持最大5个图/视频混合参考，支持音频音色参考，搭配基础能力升级实现更强表演能力。该版本为2026年6月12日快照。 -推理模型 +图片生成 -2026-06-15 +2026-06-25 -中国内地 +`qwen-image-2.0-pro-2026-06-22` -kimi/kimi-k2.7-code +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。 -月之暗面直供的 Kimi K2.7 Code 模型，以编码为中心的智能体模型，专为长程软件工程任务优化，仅支持思考模式。擅长跨多文件重构、功能实现、长会话调试等复杂工作流。[Kimi-月之暗面](https://help.aliyun.com/zh/model-studio/kimi-api-by-moonshot-ai) +文本生成、深度思考、视觉理解 -推理模型 +2026-06-17 -2026-06-10 +`kimi/kimi-k2.7-code-highspeed` -中国内地 +K2.7 Code高速版与普通版是同一个模型，但输出速度约为普通版的 5-6 倍，常规编程场景下（取输入长度中位数）输出速度约 180 Token/s，短上下文场景可达 260 Token/s ，带来更极致的编程体验。 -qwen3.7-max-2026-06-08 +语音识别 -Qwen3.7系列中规模最大、综合能力最强的Max模型，相较于5月20日快照增加了视觉模态理解能力，能够感知真实世界场景，具备多模态交互混合智能体能力。 +2026-06-17 -参考生视频 +`fun-asr-flash-2026-06-15` -2026-06-08 +百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。 -中国内地 +视觉理解 -pixverse/pixverse-v6-r2v +2026-06-16 -爱诗V6-参考生视频模型，参考多张图像生成视频。相较于v5.6全面升级，通用场景推荐使用。支持将一张多宫格分镜拼接图一键转化为视频。[爱诗-参考生视频](https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference) +`qwen3.5-ocr` -推理模型 +Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景中）抽取上效果显著提升。 -2026-06-01 +视频生成 -中国内地 +2026-06-16 -qwen3.7-plus、qwen3.7-plus-2026-05-26 +`happyhorse-1.1-t2v` -千问3.7Plus系列，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 +HappyHorse-1.1-T2V支持文生视频，进一步提升文本语义理解、镜头调度与动态生成表现。模型能够更精准地还原创作意图，在人物动作、场景氛围、视觉美感和物理运动上生成更流畅自然、细节丰富且一致性更高的高质量视频。 -推理模型 +视频生成 -2026-05-29 +2026-06-16 -中国内地 +`happyhorse-1.1-r2v` -vanchin/deepseek-v4-pro +HappyHorse-1.1-R2V支持参考生视频，进一步提升主体、场景风格与画面一致性的稳定保持。模型最多支持9张参考图片输入，能够更精准理解并延续创作意图，在人物、场景、风格和镜头表现上带来更强的可控性与表现力。 -由快手万擎直供的 DeepSeek 模型推理服务。[DeepSeek-快手万擎](https://help.aliyun.com/zh/model-studio/deepseek-api-by-vanchin) +视频生成 -推理模型 +2026-06-16 -2026-05-29 +`happyhorse-1.1-i2v` -中国内地 +HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。 -stepfun/step-3.7-flash +文本生成、深度思考 -阶跃星辰直供的 Step 3.7 Flash 模型，Flash 档位新一代旗舰。在搜索、Agent、编码、多模态四大方向全面升级，深度检索与多模态图搜能力大幅增强，Agent 与工具调用能力显著提升，编码能力在 Flash 档位多项基准中表现优异，多模态理解能力对标头部旗舰。 +2026-06-16 -推理模型 +`glm-5.2` -2026-05-25 +`ZHIPU/GLM-5.2` -中国内地 +GLM-5.2是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 -qwen3.7-max-preview、qwen3.7-max-2026-05-17 +文本生成、深度思考、视觉理解 -Qwen Max 系列模型。仅支持纯文本输入，仅支持思考模式。 +2026-06-15 -推理模型 +`kimi-k2.7-code` -2026-05-21 +`kimi/kimi-k2.7-code` -中国内地 +kimi-k2.7-code是 kimi 迄今最智能的coding模型，在长上下文中更可靠地遵循指令，能以更高的成功率完成编程任务，同时支持文本、图片与视频输入，思考模式，对话与 Agent 任务。 -qwen3.7-max、qwen3.7-max-2026-05-20 +文本生成、深度思考、视觉理解 -Qwen Max 系列新一代旗舰模型。仅支持纯文本输入，默认开启思考模式，支持显式缓存，在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 +2026-06-09 -音视频翻译 +`qwen3.7-max-2026-06-08` -2026-05-19 +Qwen3.7系列中规模最大、综合能力最强的Max模型，相较于5月20日快照增加了视觉模态理解能力，能够感知真实世界场景，具备多模态交互混合智能体能力。该版本为2026年6月8日快照。 -中国内地 +视频生成 -qwen3.5-livetranslate-flash-realtime、qwen3.5-livetranslate-flash-realtime-2026-05-19 +2026-06-08 -一款多语言音视频实时翻译模型，可识别 60 种语言，并实时翻译为 29 种语言的音频。[实时语音/音视频翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-5-livetranslate-flash-realtime) +`pixverse/pixverse-v6-r2v` -推理模型 +V6是PixVerse在26年3月底推出的新模型，r2v（多主体参考生成视频）模型全球排名第二，输入2-7张图像，智能融合不同主体，适合复杂的中景、远景视频镜头。同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力、更强的情绪表现力、和更流畅的高速运动画面。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。 -2026-05-19 +文本生成、深度思考、视觉理解 -中国内地 +2026-06-01 -xiaomi/mimo-v2.5-pro +`qwen3.7-plus` -小米直供的 MiMo-V2.5-Pro 模型，在通用智能体能力、复杂软件工程以及长程任务等方面提升显著。[MiMo-小米](https://help.aliyun.com/zh/model-studio/mimo) +`qwen3.7-plus-2026-05-26` -推理模型 +Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 -2026-05-19 +语音合成 -中国内地 +2026-06-01 -ZHIPU/GLM-5.1、ZHIPU/GLM-5 +`fun-music-preview` -智谱直供的 glm-5、glm-5.1模型，适用于智能交互、企业应用及开发辅助等场景。[GLM-智谱](https://help.aliyun.com/zh/model-studio/glm-zhipu) +百聆音乐生成preview版大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。本次版本为预览快照版 -音乐生成 +文本生成、深度思考、视觉理解 -2026-05-06 +2026-06-01 -中国内地 +`MiniMax/MiniMax-M3` -fun-music-v1 +MiniMax M3 凭借业界领先的 Coding 与 Agentic 能力、1M 超长上下文窗口以及原生多模态特性，可出色胜任企业级长文档理解、高质量内容生成、代码编写、Bug 修复及原生应用构建等任务；强大的 Agentic 能力端到端贯通工作流，原生多模态更带来流畅自然的图文混合交互体验。 -百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。[音乐生成](https://help.aliyun.com/zh/model-studio/fun-music/) +文本生成 -推理模型 +2026-05-29 -2026-04-29 +`vanchin/deepseek-v4-pro` -中国内地 +DeepSeek-V4系列是强大的混合专家（MoE）语言模型，包含DeepSeek-V4-Pro（1.6T总参数，49B激活参数）。支持高达100万（1M）token的上下文长度，是在超过32T高质量多样化token上预训练的开源模型。 -deepseek-v4-pro +文本生成、视觉理解 -deepseek-v4-pro `cached_token` 单价调整为 **1 元/百万 token**，标准 `input_token` 单价不变。详见[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)。 +2026-05-25 -参考生视频 +`stepfun/step-3.7-flash` -2026-04-29 +Step 3.7 Flash 是阶跃星辰最新推出的生产级 Agent 高效率 Flash 模型，专为 Agent、Coding、Search 与多模态工作流打造，在速度、成本、执行可靠性与复杂任务完成能力之间实现了更优平衡。具备多模态感知与执行、视觉搜索与工具增强、高可靠工具调用与编排，以及 Agent 生态兼容优化等核心能力。 -中国内地 +文本生成、深度思考 -vidu/viduq3-mix\_reference2video、vidu/viduq3\_reference2video、vidu/viduq3-turbo\_reference2video +2026-05-21 -Vidu-参考生视频系列模型，支持传入参考图片和文本提示词，将图片中的主体角色融合到提示词描述的场景中，生成流畅的视频内容。[Vidu-参考生视频](https://help.aliyun.com/zh/model-studio/vidu-reference-to-video-api-reference) +`qwen3.7-max` -图生视频 +`qwen3.7-max-2026-05-20` -2026-04-29 +Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 -中国内地 +文本生成、深度思考 -vidu/viduq2-pro-fast\_img2video +2026-05-20 -Vidu-首帧生视频系列模型，根据输入图像和文本提示词，生成一段流畅的视频。[Vidu-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/vidu-image-to-video-api-reference) +`qwen3.7-max-preview` -3D生成 +`qwen3.7-max-2026-05-17` -2026-04-29 +Qwen3.7系列中规模最大、综合能力最强的Max模型预览版，仅支持思考模式，开放纯文本模型能力供体验。主要优化面向用户的通用对话场景，例如知识问答、指令跟随、创意写作等。 -中国内地 +实时语音翻译 -Tripo/Tripo-H3.1 +2026-05-19 -Tripo 高精度3D生成模型，支持最高200万面，可实现文生3D、单图生3D和多图生3D任务。[Tripo-3D模型生成](https://help.aliyun.com/zh/model-studio/tripo-3d-generation-api-reference) +`qwen3.5-livetranslate-flash-realtime` -3D生成 +`qwen3.5-livetranslate-flash-realtime-2026-05-19` -2026-04-29 +Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。 -中国内地 +文本生成 -Tripo/Tripo-P1.0 +2026-05-18 -Tripo 专业3D生成模型，最高2万面，生成速度较快，可实现文生3D、单图生3D和多图生3D任务。[Tripo-3D模型生成](https://help.aliyun.com/zh/model-studio/tripo-3d-generation-api-reference) +`xiaomi/mimo-v2.5-pro` -文生视频 +MiMo-V2.5-Pro 是小米发布的最新旗舰模型。与前代模型相比，它在通用智能体能力、复杂软件工程以及长程任务等方面都有显著提升，在 ClawEval、GDPVal 和 SWE-bench Pro 等基准测试中均位列前茅。它能够独立且完全自主地完成需要人类专家耗时数天甚至数周的专业任务，涉及上千次工具调用。其高达 100 万 token 的上下文长度，非常适合集成到各种智能体框架中使用。 -2026-04-27 +文本生成 -中国内地 +2026-05-18 -happyhorse-1.0-t2v +`ZHIPU/GLM-5.1` -HappyHorse系列文生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-文生视频](https://help.aliyun.com/zh/model-studio/happyhorse-text-to-video-api-reference) +GLM-5.1 是智谱最新旗舰模型，代码能力大大增强，长程任务显著提升，能够在单次任务中持续、自主地工作长达 8 小时，完成从规划、执行到迭代优化的完整闭环，交付工程级成果。 在综合能力与 Coding 能力上，GLM-5.1 整体表现对齐 Claude Opus 4.6，并在长程自主执行、复杂工程优化与真实开发场景中展现出更强的持续工作能力，是构建 Autonomous Agent 与长程 Coding Agent 的理想基座。 -图生视频 +语音合成 -2026-04-27 +2026-05-06 + +`fun-music-v1` -中国内地 +百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。 -happyhorse-1.0-i2v +视频生成 -HappyHorse系列图生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/happyhorse-image-to-video-api-reference) +2026-04-28 -参考生视频 +`vidu/viduq3-mix_reference2video` -2026-04-27 +输入1-7张参考图片与文本描述，生成视频。ViduQ3-mix画面质感强，均衡性好。ViduQ3参考生视频为剧而生，万物可参，声画同出：6大特效（粒子 / 流体 / 动力学 / 运镜 / 转场 / 光影）、5大音效（环境 / 动态 / 氛围 / 拟音 / 情绪）、4大场景（短剧 / 漫剧 / 影视剧 / 广告）全面融合，轻松驾驭短剧、漫剧、广告等多元创作。 + +视频生成 -中国内地 +2026-04-27 -happyhorse-1.0-r2v +`vidu/viduq3_reference2video` -HappyHorse系列参考生视频模型，支持多参考图输入生成有声视频，可生成3~15秒、720P/1080P视频。[HappyHorse-参考生视频](https://help.aliyun.com/zh/model-studio/happyhorse-reference-to-video-api-reference) +输入1-7张参考图片与文本描述，生成视频。ViduQ3支持智能切镜，多机位一致性更出色。ViduQ3参考生视频为剧而生，万物可参，声画同出：6大特效（粒子 / 流体 / 动力学 / 运镜 / 转场 / 光影）、5大音效（环境 / 动态 / 氛围 / 拟音 / 情绪）、4大场景（短剧 / 漫剧 / 影视剧 / 广告）全面融合，轻松驾驭短剧、漫剧、广告等多元创作。 -视频编辑 +视频生成 2026-04-27 -中国内地 +`vidu/viduq3-turbo_reference2video` -happyhorse-1.0-video-edit +输入1-7张参考图片与文本描述，生成视频。ViduQ3-Turbo生成速度快，性价比高。ViduQ3参考生视频为剧而生，万物可参，声画同出：6大特效（粒子 / 流体 / 动力学 / 运镜 / 转场 / 光影）、5大音效（环境 / 动态 / 氛围 / 拟音 / 情绪）、4大场景（短剧 / 漫剧 / 影视剧 / 广告）全面融合，轻松驾驭短剧、漫剧、广告等多元创作。 -HappyHorse系列视频编辑模型，支持对视频进行编辑处理。[HappyHorse-视频编辑](https://help.aliyun.com/zh/model-studio/happyhorse-video-edit-api-reference) +视频生成 -文生视频 +2026-04-27 -2026-07-01 +`vidu/viduq2-pro-fast_img2video` -中国内地 +输入图片与文本描述，生成视频。ViduQ2-Pro-fast价格触底、效果稳定，生成速度较turbo提高2-3倍。ViduQ2图生视频是全球首创「万物可参考」视频模型。支持特效、表情、纹理、动作、人物、场景等六大维度参考，实现编辑全面进化。通过可控式增、删、改，达成精细化视频编辑，专为漫剧、短剧、影视制作打造的生产级创作引擎。 -wan2.7-t2v-2026-06-12 +3D生成 -万相2.7-文生视频模型快照版本，模型能力与wan2.7-t2v一致。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +2026-04-27 -文生视频 +`Tripo/Tripo-P1.0` -2026-04-26 +Tripo P1.0 是面向实时应用与生产管线的 3D 生成模型，专为需要干净拓扑和引擎可用网格的开发者与创作者设计。模型可在约 2 秒内生成具备专业级拓扑结构的 3D 资产，适用于游戏、Web3D 与各类实时交互场景。针对 UGC 内容生产中对“速度”和“开箱即用”的需求，Tripo P1.0 在保证质量的同时大幅提升生成效率，使资产能够快速接入实时引擎与开发流程。 -中国内地 +3D生成 + +2026-04-27 -wan2.7-t2v-2026-04-25 +`Tripo/Tripo-H3.1` -万相2.7-文生视频模型快照版本，模型能力与wan2.7-t2v一致。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +Tripo H3.1 是 Tripo 推出的高精度 3D 生成模型，专为需要极致视觉质量与细节表现的创作者设计。模型通过核心算法升级与模块优化，参数规模达 200 亿级，支持十亿体素级三维分辨率与最高 200 万面多边形生成。在保持高精度几何与真实纹理的同时，Tripo H3.1 对输入参考图的还原度与对齐度进一步提升，在角色形体、面部细节与几何文字等复杂结构上实现更稳定、细致的表达，适用于高质量视觉制作与 3D 打印等高精度资产生产场景。 -图生视频 +视频生成 2026-04-26 -中国内地 +`wan2.7-t2v-2026-04-25` + +万相2.7-文生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。该版本为2026年4月25日快照。 + +视频生成 + +2026-04-26 -wan2.7-i2v-2026-04-25 +`wan2.7-i2v-2026-04-25` -万相2.7-图生视频模型快照版本，模型能力与wan2.7-i2v一致。[万相2.7-图生视频](https://help.aliyun.com/zh/model-studio/image-to-video-general-api-reference) +万相2.7-图生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。该版本为2026年4月25日快照。 -文生文与视觉理解 +文本生成、深度思考、视觉理解 2026-04-26 -中国内地 +`kimi/kimi-k2.6` -kimi/kimi-k2.6 +Kimi K2.6 是 Kimi 最新最智能的模型，Kimi K2.6 的通用 Agent、代码、视觉理解等综合能力得到全面提升，其中在博士级难度的完整版人类最后的考试（Humanity’s Last Exam）、在考察模型真实软件工程能力的 SWE-Bench Pro、评估 Agent 深度检索能力的 DeepSearchQA 等基准测试中均取得行业领先的成绩，同时支持文本、图片与视频输入，思考与非思考模式，对话与 Agent 任务。 -kimi/kimi-k2.6 是 Kimi 最新最智能的模型，月之暗面直供，在通用 Agent、代码、视觉理解等综合能力得到全面提升，同时支持文本、图片与视频输入、思考与非思考模式、对话与 Agent 任务。[Kimi-月之暗面](https://help.aliyun.com/zh/model-studio/kimi-api-by-moonshot-ai) +视频生成 -推理模型 +2026-04-26 -2026-04-24 +`happyhorse-1.0-video-edit` -中国内地 +HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。 -deepseek-v4-pro、deepseek-v4-flash +视频生成 -DeepSeek-V4系列模型，阿里直供。deepseek-v4-pro为旗舰模型，deepseek-v4-flash为轻量级高速模型。[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api) +2026-04-26 -图像生成 +`happyhorse-1.0-r2v` -2026-04-23 +HappyHorse-1.0-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。 -中国内地 +文本生成、深度思考 -qwen-image-2.0-pro-2026-04-22 +2026-04-24 -Qwen-Image-2.0系列模型，实现了图片生成和图片编辑的融合。相较于3月3日快照，本模型在画面质感，尤其是纹理细节、光影、材质上有明显跃升；支持多语言的图内文字生成；艺术风格表现更加均衡。[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api)、[千问-图像编辑](https://help.aliyun.com/zh/model-studio/qwen-image-edit-api) +`deepseek-v4-pro` -推理模型 +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 -2026-04-23 +文本生成、深度思考 -中国内地 +2026-04-24 -qwen3.5-plus-2026-04-20 +`deepseek-v4-flash` -Qwen3.5原生视觉语言系列Plus模型新快照，相较2月15日快照Agentic coding能力大幅提升，推理速度显著提升；知识、推理与长上下文能力保持较高水准，适合编码智能体、生产工作流和高吞吐场景。 +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 -推理模型 +文本生成、深度思考、视觉理解 2026-04-23 -中国内地 +`qwen3.5-plus-2026-04-20` -qwen3.6-27b +Qwen3.5原生视觉语言系列Plus模型，相较于2月15日快照，本模型在Agentic coding能力上大幅提升；推理速度显著提升；知识、推理与长上下文能力保持较高水准，满足复杂Agent任务的需求，适合应用于编码智能体、生产工作流和高吞吐场景。该版本为2026年4月20日快照。 -Qwen3.6系列27B原生视觉语言Dense模型，相较3.5-27B重点提升Agentic coding能力，STEM与推理能力进一步增强；视觉模态方面空间智能、物体定位与检测能力显著增强，视频理解、文档OCR及视觉Agent能力稳步提升。 +图片生成 -文生文与视觉理解 +2026-04-23 -2026-04-21 +`qwen-image-2.0-pro` -中国内地 +`qwen-image-2.0-pro-2026-04-22` -kimi-k2.6 +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。 -Kimi最新最智能的模型，具备更强更稳的长程代码编写能力，指令遵循和自我纠错能力显著提升。同时支持文本、图片与视频输入、思考与非思考模式、对话与 Agent 任务。[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api) +文本生成、深度思考、视觉理解 -推理模型 +2026-04-22 -2026-04-20 +`qwen3.6-27b` -中国内地 +Qwen3.6系列27B原生视觉语言Dense模型，模型效果相较3.5-27B重点提升了Agentic coding能力、模型STEM与推理能力进一步增强；视觉模态方面在空间智能、物体定位与检测能力上显著增强，视频理解、文档OCR及视觉Agent能力稳步提升。 -qwen3.6-max-preview +视频生成 -Qwen3.6系列规模最大的闭源模型，Coding能力进一步提升、Agent执行更加高效。仅支持纯文本输入，支持思考模式（默认开启），支持显式缓存和Function Calling。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +2026-04-22 -> 不支持图像与视频输入。 +`happyhorse-1.0-i2v` -推理模型 +HappyHorse-1.0-I2V支持图生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 -2026-04-16 +文本生成、深度思考、视觉理解 -中国内地 +2026-04-21 -qwen3.6-flash、qwen3.6-flash-2026-04-16、qwen3.6-35b-a3b +`kimi-k2.6` -Qwen3.6 原生视觉语言 Flash 系列模型，在整体性能上较 Qwen3.5-Flash 显著提升。重点增强了智能体编程能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；在视觉能力方面，空间智能显著增强，其中物体定位和目标检测表现尤为突出。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +kimi-k2.6是Kimi最新最智能的模型，具备更强更稳的长程代码编写能力，指令遵循和自我纠错能力显著提升，同时支持文本、图片与视频输入，思考与非思考模式，对话与Agent任务。 -语音识别 +视频生成 -2026-04-16 +2026-04-21 -中国内地 +`happyhorse-1.0-t2v` -fun-asr、fun-asr-2025-11-07 +HappyHorse-1.0-T2V支持文生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 -Fun-ASR 实时语音识别能力升级： +文本生成、深度思考 -- 方言支持：覆盖汉语七大方言（官话/吴/湘/赣/客/闽/粤）及 20+ 地区口音 - -- 古诗词优化：提升古诗词识别准确率，适用于文化教育、有声读物等场景 - -- 文本优化：标点预测与文本归一化增强，数字/日期/金额自动转标准格式 - -- 多语种扩展：支持中、英、日、韩、法、德、西班牙等共 30 个语种 - +2026-04-20 -详情请参见[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition) +`qwen3.6-max-preview` -推理模型 +Qwen3.6系列中规模最大、综合能力最强的Max模型Preview版本，当前开放纯文本模型能力供体验。相较于此前发布的Qwen3-Max和Qwen3.6-Plus，本模型在vibe coding能力上进一步提升、coding agent执行更加高效、前端编程开发能力显著提升；长尾知识能力进一步升级。 -2026-04-14 +文本生成、深度思考、视觉理解 -中国内地 +2026-04-16 -glm-5.1 +`qwen3.6-flash` -智谱GLM-5.1模型，专为长程任务设计，支持 200K 上下文，最大输出可达 128K Token。通过强大的逻辑推理、长文本理解及代码生成能力，在多项基准测试中表现优异，适用于智能交互、企业应用及开发辅助等场景。[GLM-阿里云](https://help.aliyun.com/zh/model-studio/glm) +`qwen3.6-flash-2026-04-16` -文生文 +Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。 -2026-04-14 +文本生成、深度思考、视觉理解 -中国内地 +2026-04-16 -vanchin/deepseek-v3.2-think、vanchin/deepseek-v3.1-terminus、vanchin/deepseek-r1、vanchin/deepseek-v3 +`qwen3.6-35b-a3b` -由快手万擎直供的 DeepSeek 模型推理服务。[DeepSeek-快手万擎](https://help.aliyun.com/zh/model-studio/deepseek-api-by-vanchin) +Qwen3.6系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果相较3.5-35B-A3B显著提升了agentic coding能力、数学推理和代码推理能力、空间智能能力、物体定位与目标检测能力。 -文字识别 +文本生成、深度思考 2026-04-14 -中国内地 +`glm-5.1` -vanchin/deepseek-ocr +GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 -由快手万擎直供的 DeepSeek OCR 模型推理服务。[DeepSeek-快手万擎](https://help.aliyun.com/zh/model-studio/deepseek-api-by-vanchin) +视频生成 -文生视频 +2026-04-09 -2026-04-13 +`pixverse/pixverse-c1-t2v` -中国内地 +C1是PixVerse在26年3月底推出的影视行业大模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。 -pixverse/pixverse-c1-t2v +视频生成 -爱诗C1-文生视频模型，基于文本提示词生成视频。支持提示词智能分镜，打斗与特效强化升级，支持真人或动漫风格。[爱诗-文生视频](https://help.aliyun.com/zh/model-studio/pixverse-text-to-video-api-reference) +2026-04-09 -图生视频 +`pixverse/pixverse-c1-r2v` -2026-04-13 +C1是PixVerse在26年3月底推出的影视行业大模型，r2v（多主体参考生成视频）输入2-7张图像，智能融合不同主体，同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力和想象力、更接近影视专业水准的打斗动作和术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合多主体群像、多人对话、多人交互等复杂剧情，适合中景、全景镜头。 如果输入了1张多宫格分镜图片（最高支持九宫格），则可以一键生成连续分镜长视频。 -中国内地 +视频生成 -pixverse/pixverse-c1-it2v +2026-04-09 -爱诗C1-首帧生视频模型，根据输入图像和文本提示词生成视频。支持提示词智能分镜，打斗与特效强化升级，支持真人或动漫风格。[爱诗-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/pixverse-image-to-video-api-reference) +`pixverse/pixverse-c1-kf2v` -图生视频 +C1是PixVerse在26年3月底推出的影视行业大模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。 -2026-04-13 +视频生成 -中国内地 +2026-04-09 -pixverse/pixverse-c1-kf2v +`pixverse/pixverse-c1-it2v` -爱诗C1-首尾帧生视频模型，基于首帧图像、尾帧图像和文本提示词生成平滑过渡的视频。支持提示词智能分镜，打斗与特效强化升级，支持真人或动漫风格。[爱诗-图生视频-基于首尾帧](https://help.aliyun.com/zh/model-studio/pixverse-keyframe-to-video-api-reference) +C1是PixVerse在26年3月底推出的影视行业大模型，it2v（图片生成视频）模型除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征。相比V6可增强提示词，拥有更强的想象力、更接近影视专业水准的打斗动作、术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合单人特写、单人独白、定格/慢动作、空镜转场等短时长镜头。 -参考生视频 +文本生成、深度思考 -2026-04-13 +2026-04-07 -中国内地 +`vanchin/deepseek-v3.2-think` -pixverse/pixverse-c1-r2v +DeepSeek-V3.2 是一款实现了高计算效率与卓越推理及代理（Agent）性能完美协调的模型。该模型建立在 DeepSeek-V3 的基础之上，通过引入 DeepSeek 稀疏注意力（DSA）、可扩展的强化学习框架以及大规模代理任务合成流水线等关键技术突破，推动了开源大语言模型的前沿发展。 -爱诗C1-参考生视频模型，参考多张图像生成视频。支持将一张多宫格分镜拼接图一键转化为视频。[爱诗-参考生视频](https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference) +文本生成、深度思考 -视频编辑 +2026-04-07 -2026-04-03 +`vanchin/deepseek-v3.1-terminus` -中国内地 +DeepSeek-V3.1-Terminus 是 DeepSeek-V3.1 的更新版本，旨在保持模型原有核心能力的同时，针对用户反馈的问题进行了修复和优化。该版本的模型结构与 DeepSeek-V3 保持一致，并在特定领域进行了显著增强。 -wan2.7-videoedit +文本生成 -万相2.7-视频编辑模型，支持指令编辑与视频迁移任务，可修改视频局部或整体画面，同时支持多图参考替换及动作、特效、运镜的复刻。[万相2.7-视频编辑](https://help.aliyun.com/zh/model-studio/wan-video-editing-api-reference) +2026-04-07 -图生视频 +`vanchin/deepseek-v3` -2026-04-03 +DeepSeek-V3 由深度求索（DeepSeek）于 2024 年 12 月发布，是目前开源社区领先的混合专家（MoE）语言模型：总参数 671B，每个 token 仅激活 37B 参数。模型在 14.8 万亿高质量 tokens 上完成预训练，原生支持 128k 上下文。通过创新的无辅助损失负载均衡策略、多头潜在注意力（MLA）架构和 FP8 混合精度训练。 -中国内地 +文本生成、深度思考 -wan2.7-i2v +2026-04-07 -万相2.7-图生视频模型，支持多模态输入（文本/图像/音频/视频），可完成首帧生视频、首尾帧生视频、视频续写三大任务。[万相2.7-图生视频](https://help.aliyun.com/zh/model-studio/image-to-video-general-api-reference) +`vanchin/deepseek-r1` -文生视频 +DeepSeek-R1 是深度求索于 2025 年 1 月开源的 6710 亿参数混合专家（MoE）推理模型，推理时仅激活 370 亿参数。作为首个通过纯强化学习（无监督微调）训练的千亿级模型，实现了链式思维（CoT）的自然涌现。模型在 RL 前加入冷启动数据解决了 R1-Zero 的重复和混语问题，在数学、代码、推理任务上达到 OpenAI o1 水平。 -2026-04-03 +文本生成、视觉理解 -中国内地 +2026-04-07 -wan2.7-t2v +`vanchin/deepseek-ocr` -万相2.7-文生视频模型，新增分辨率档位与宽高比自定义设置，支持灵活适配不同创作场景与平台发布需求。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +DeepSeek-OCR以 “探索视觉 - 文本压缩边界” 为核心目标，从大语言模型（LLM）视角重新定义视觉编码器的功能定位，为文档识别、图像转文本等高频场景提供了兼顾精度与效率的全新解决方案。 -参考生视频 +视频生成 2026-04-03 -中国内地 +`wan2.7-videoedit` -wan2.7-r2v +Wan2.7-VideoEdit，自然语言指令编辑视频，支持局部或全局编辑，可参考图像替换视频元素，支持复刻视频动作、特效、运镜等动态过程。 -万相2.7参考生视频模型，支持主体参考和音色定制，并可输入单张多宫格故事板直接生成剧本化视频。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +视频生成 -推理模型 +2026-04-03 -2026-04-02 +`wan2.7-t2v` -中国内地 +Wan2.7-T2V，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。 -qwen3.6-plus、qwen3.6-plus-2026-04-02 +视频生成 -千问3.6-Plus，代码开发能力重点升级（Agentic Coding、前端编程等），Vibe Coding体验显著提升；泛化场景推理能力进一步增强；多模态方面，万物识别、OCR、物体定位等能力显著提升；同时修复了Qwen3.5-Plus上线后的已知问题。使用方法与qwen3.5-plus一致。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +2026-04-03 -图像生成与编辑 +`wan2.7-r2v` -2026-04-01 +Wan2.7-R2V，更加稳定的角色、道具与场景参考，支持最大5个图/视频混合参考，支持音频音色参考，搭配基础能力升级实现更强表演能力。 -中国内地 +视频生成 -wan2.7-image-pro、wan2.7-image +2026-04-03 -万相2.7-图像生成与编辑模型，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循表现更优。Pro系列支持4K输出；加速版兼顾效果与响应速度。[万相-图像生成与编辑2.7](https://help.aliyun.com/zh/model-studio/wan-image-generation-and-editing-api-reference) +`wan2.7-i2v` -全模态 +万相2.7-图生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。 -2026-03-30 +图片生成 -中国内地 +2026-04-01 -qwen3.5-omni-plus、qwen3.5-omni-plus-2026-03-15、qwen3.5-omni-flash、qwen3.5-omni-flash-2026-03-15 +`wan2.7-image-pro` -最新一代全模态大模型，支持长视频分析、会议纪要、字幕输出、安全审核、音视频交互；支持音视频内容的深度理解与生成描述，支持 113 种语言识别和 36 种语言的音频生成，可处理 3 小时音频及1 小时视频输入，支持联网搜索及指令来控制输出音频的音量、语速、情绪。[非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni) +万相2.7-图像生成与编辑旗舰版模型，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现。 -全模态 +图片生成 -2026-03-30 +2026-04-01 -中国内地 +`wan2.7-image` -qwen3.5-omni-plus-realtime、qwen3.5-omni-plus-realtime-2026-03-15、qwen3.5-omni-flash-realtime、qwen3.5-omni-flash-realtime-2026-03-15 +万相2.7-图像生成与编辑，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现 -千问最新推出的实时多模态模型，相比于上一代的 Qwen3-Omni-Flash-Realtime：模型智力大幅提升，与 Qwen3.5-Plus 智能水平相当。原生支持联网搜索（WebSearch），支持语音打断和控制；支持 113 种语种和方言的语音识别，以及 36 种语种和方言的语音生成。[实时（Qwen-Omni-Realtime）](https://help.aliyun.com/zh/model-studio/realtime) +文本生成、深度思考、视觉理解 -文生视频 +2026-04-01 -2026-03-29 +`qwen3.6-plus` -中国内地 +`qwen3.6-plus-2026-04-02` -pixverse/pixverse-v6-t2v +Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。 -爱诗v6-文生视频模型，基于文本提示词，生成一段流畅的视频。[爱诗-文生视频](https://help.aliyun.com/zh/model-studio/pixverse-text-to-video-api-reference) +视频生成 -图生视频 +2026-04-01 -2026-03-29 +`pixverse/pixverse-v6-t2v` -中国内地 +V6是PixVerse在26年3月底推出的新模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。 -pixverse/pixverse-v6-it2v +视频生成 -爱诗v6-首帧生视频模型，根据输入图像和文本提示词，生成一段流畅的视频。[爱诗-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/pixverse-image-to-video-api-reference) +2026-04-01 -图生视频 +`pixverse/pixverse-v6-kf2v` -2026-03-29 +V6是PixVerse在26年3月底推出的新模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。 -中国内地 +视频生成 -pixverse/pixverse-v6-kf2v +2026-04-01 -爱诗v6-首尾帧生视频模型，基于首帧图像、尾帧图像和文本提示词，生成一段平滑过渡的视频。[爱诗-图生视频-基于首尾帧](https://help.aliyun.com/zh/model-studio/pixverse-keyframe-to-video-api-reference) +`pixverse/pixverse-v6-it2v` -文生视频 +V6是PixVerse在26年3月底推出的新模型，it2v（图片生成视频）模型全球排名第二，it2v除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征，拥有更强的人物情绪、高速运动表现力。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。 -2026-03-27 +实时全模态 -中国内地 +2026-03-30 -vidu/viduq3-pro\_text2video、vidu/viduq3-turbo\_text2video、vidu/viduq2\_text2video +`qwen3.5-omni-plus-realtime` -Vidu-文生视频系列模型。基于文本提示词，生成一段流畅的视频。[Vidu-文生视频](https://help.aliyun.com/zh/model-studio/vidu-text-to-video-api-reference) +`qwen3.5-omni-plus-realtime-2026-03-15` -图生视频 +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。 -2026-03-27 +全模态 -中国内地 +2026-03-30 -vidu/viduq3-pro\_img2video、vidu/viduq3-turbo\_img2video、vidu/viduq2-pro\_img2video、vidu/viduq2-turbo\_img2video +`qwen3.5-omni-plus` -Vidu-首帧生视频系列模型，根据输入图像和文本提示词，生成一段流畅的视频。[Vidu-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/vidu-image-to-video-api-reference) +`qwen3.5-omni-plus-2026-03-15` -首尾帧生视频 +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。 -2026-03-27 +实时全模态 -中国内地 +2026-03-30 -vidu/viduq3-pro\_start-end2video、vidu/viduq3-turbo\_start-end2video、vidu/viduq2-pro\_start-end2video、vidu/viduq2-turbo\_start-end2video +`qwen3.5-omni-flash-realtime` -Vidu-首尾帧生视频系列模型，基于首帧图像、尾帧图像和文本提示词，生成一段平滑过渡的视频。[Vidu-图生视频-基于首尾帧](https://help.aliyun.com/zh/model-studio/vidu-keyframe-to-video-api-reference) +`qwen3.5-omni-flash-realtime-2026-03-15` -参考生视频 +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。 -2026-03-27 +全模态 -中国内地 +2026-03-30 -vidu/viduq2-pro\_reference2video、vidu/viduq2\_reference2video +`qwen3.5-omni-flash` -Vidu-参考生视频系列模型，支持传入参考图片和文本提示词，将图片中的主体角色融合到提示词描述的场景中，生成流畅的视频内容。[Vidu-参考生视频](https://help.aliyun.com/zh/model-studio/vidu-reference-to-video-api-reference) +`qwen3.5-omni-flash-2026-03-15` -图像生成 +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。 -2026-03-24 +视频生成 -中国内地 +2026-03-26 -kling/kling-v3-image-generation、kling/kling-v3-omni-image-generation +`vidu/viduq3-turbo_text2video` -可灵V3图像生成系列模型，支持文生图、参考图生图两种任务。[可灵-图像生成](https://help.aliyun.com/zh/model-studio/kling-image-generation-api-reference) +输入一段文本，生成视频。ViduQ3-Turbo文生视频是高性能加速版模型。生成效率极高，兼具优质画质与动态表现，尤其在打斗场面、情绪渲染及语义理解上表现出色。性价比突出，适合图片社交、AI陪伴及特效素材等泛娱乐场景。 视频生成 -2026-03-24 +2026-03-26 -中国内地 +`vidu/viduq3-turbo_start-end2video` -kling/kling-v3-video-generation、kling/kling-v3-omni-video-generation +输入首帧图、尾帧图与文本描述，生成视频。ViduQ3-Turbo首尾帧生视频是高性能加速版模型。生成效率极高，兼具优质画质与动态表现，尤其在打斗场面、情绪渲染及语义理解上表现出色。性价比突出，适合图片社交、AI陪伴及特效素材等泛娱乐场景。 -可灵V3视频生成系列模型，支持文生视频、图生视频-基于首帧、图生视频-基于首尾帧、参考生视频以及视频编辑。[可灵-视频生成](https://help.aliyun.com/zh/model-studio/kling-video-generation-api-reference/) +视频生成 -参考生视频 +2026-03-26 -2026-03-20 +`vidu/viduq3-turbo_img2video` -中国内地 +输入图片与文本描述，生成视频。ViduQ3-Turbo图生视频是高性能加速版模型。生成效率极高，兼具优质画质与动态表现，尤其在打斗场面、情绪渲染及语义理解上表现出色。性价比突出，适合图片社交、AI陪伴及特效素材等泛娱乐场景。 -pixverse/pixverse-v5.6-r2v +视频生成 -参考多张图像生视频。[爱诗-参考生视频](https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference) +2026-03-26 -图生视频 +`vidu/viduq3-pro_text2video` -2026-03-20 +输入一段文本，生成视频。ViduQ3-Pro文生视频是旗舰级音视频原生模型。支持长达16秒的音画同步生成，实现多镜头自由切换，精准把控节奏、情绪与叙事连贯性。参数量领先，画质、人物一致性及情绪表现卓越，达电影级标准。适用于广告（电商、TVC、效果投放）、漫剧、真人剧及游戏等专业生产场景。 -中国内地 +视频生成 -pixverse/pixverse-v5.6-kf2v +2026-03-26 -基于首帧和尾帧生视频。[爱诗-图生视频-基于首尾帧](https://help.aliyun.com/zh/model-studio/pixverse-keyframe-to-video-api-reference) +`vidu/viduq3-pro_start-end2video` -图生视频 +输入首帧图、尾帧图与文本描述，生成视频。ViduQ3-Pro首尾帧生视频是旗舰级音视频原生模型。支持长达16秒的音画同步生成，实现多镜头自由切换，精准把控节奏、情绪与叙事连贯性。参数量领先，画质、人物一致性及情绪表现卓越，达电影级标准。适用于广告（电商、TVC、效果投放）、漫剧、真人剧及游戏等专业生产场景。 -2026-03-20 +视频生成 -中国内地 +2026-03-26 -pixverse/pixverse-v5.6-it2v +`vidu/viduq3-pro_img2video` -基于首帧生视频。[爱诗-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/pixverse-image-to-video-api-reference) +输入图片与文本描述，生成视频。ViduQ3-Pro图生视频是旗舰级音视频原生模型。支持长达16秒的音画同步生成，实现多镜头自由切换，精准把控节奏、情绪与叙事连贯性。参数量领先，画质、人物一致性及情绪表现卓越，达电影级标准。适用于广告（电商、TVC、效果投放）、漫剧、真人剧及游戏等专业生产场景。 -文生视频 +视频生成 -2026-03-20 +2026-03-26 -中国内地 +`vidu/viduq2_text2video` -pixverse/pixverse-v5.6-t2v +输入一段文本，生成视频。ViduQ2文生视频是精准指令遵循与细腻情感捕捉模型。具备卓越的剧情控制力，能深刻理解并表现微表情变化；镜头语言丰富，运镜流畅，画面张力十足。广泛适用于影视动漫、广告电商、短剧及文旅等行业。 -文生视频系列模型。[爱诗-文生视频](https://help.aliyun.com/zh/model-studio/pixverse-text-to-video-api-reference) +视频生成 -多模态向量 +2026-03-26 -2026-03-19 +`vidu/viduq2_reference2video` -中国内地 +输入参考图片与文本描述，生成视频。ViduQ2参考生视频是精准指令遵循与细腻情感捕捉模型。具备卓越的剧情控制力，能深刻理解并表现微表情变化；镜头语言丰富，运镜流畅，画面张力十足。广泛适用于影视动漫、广告电商、短剧及文旅等行业。 -tongyi-embedding-vision-plus-2026-03-06、tongyi-embedding-vision-flash-2026-03-06 +视频生成 -相较于 tongyi-embedding-vision 主线模型，tongyi-embedding-vision-plus-2026-03-06 / flash-2026-03-06 视觉向量化能力全面升级，基于 Qwen3 底座，支持多分辨率（0-3 档可调）、多维度（Plus 64-1152 维 / Flash 64-768 维）、30+ 语言，以及文本/图片/视频融合向量化，IPC/自驾/电商/视频营销等场景图像搜索效果大幅提升。 +2026-03-26 -界面交互 +`vidu/viduq2-turbo_start-end2video` -2026-03-18 +输入首帧图、尾帧图与文本描述，生成视频。ViduQ2-Turbo首尾帧生视频是极速生成引擎。720P 5s视频最快仅需19秒，1080P 5s视频约27秒。人物动作与表情自然逼真，真实感强，在打斗等高动态场景中效果出色，运动幅度大。 -中国内地 +视频生成 -gui-plus-2026-02-26 +2026-03-26 -相较于`gui-plus`，`gui-plus-2026-02-26`模型能力全面升级，支持思考与非思考模式，模型在处理跨平台、多 APP 任务的效果上得到大幅提升。[界面交互](https://help.aliyun.com/zh/model-studio/gui-automation) +`vidu/viduq2-turbo_img2video` -语音识别 +输入图片与文本描述，生成视频。ViduQ2-Turbo图生视频是极速生成引擎。720P 5s视频最快仅需19秒，1080P 5s视频约27秒。人物动作与表情自然逼真，真实感强，在打斗等高动态场景中效果出色，运动幅度大。 -2026-03-05 +视频生成 -中国内地 +2026-03-26 -fun-asr-realtime-2026-02-28 +`vidu/viduq2-pro_start-end2video` -Fun-ASR实时语音识别新增快照模型，较 fun-asr-realtime-2025-11-07 效果更优。[实时语音识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition) +输入首帧图、尾帧图与文本描述，生成视频。ViduQ2-Pro首尾帧生视频是全球首创「万物可参考」视频模型。支持特效、表情、纹理、动作、人物、场景等六大维度参考，实现编辑全面进化。通过可控式增、删、改，达成精细化视频编辑，专为漫剧、短剧、影视制作打造的生产级创作引擎。 -图像生成与编辑 +视频生成 -2026-03-03 +2026-03-26 -中国内地 +`vidu/viduq2-pro_reference2video` -qwen-image-2.0、qwen-image-2.0-2026-03-03、qwen-image-2.0-pro、qwen-image-2.0-pro-2026-03-03 +输入参考视频、图片与文本描述，生成视频。ViduQ2-Pro参考生视频是全球首创「万物可参考」视频模型。支持特效、表情、纹理、动作、人物、场景等六大维度参考，实现编辑全面进化。通过可控式增、删、改，达成精细化视频编辑，专为漫剧、短剧、影视制作打造的生产级创作引擎。 -千问-Image2.0系列，同时支持图像生成和编辑。Pro系列文字渲染、真实质感、语义遵循能力更强；加速版兼顾效果与响应速度。[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api)、[千问-图像编辑](https://help.aliyun.com/zh/model-studio/qwen-image-edit-api) +视频生成 -语音识别 +2026-03-26 -2026-03-03 +`vidu/viduq2-pro_img2video` -中国内地 +输入图片与文本描述，生成视频。ViduQ2-Pro图生视频是全球首创「万物可参考」视频模型。支持特效、表情、纹理、动作、人物、场景等六大维度参考，实现编辑全面进化。通过可控式增、删、改，达成精细化视频编辑，专为漫剧、短剧、影视制作打造的生产级创作引擎。 -qwen3-asr-flash-2026-02-10 +视频生成 -千问录音文件识别新增快照模型，较 qwen3-asr-flash-2025-09-08 效果更优。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +2026-03-26 -语音合成 +`kling/kling-v3-video-generation` -2026-03-02 +智能分镜可读懂剧本场景流转，自动调度机位和景别。原生多模态框架支持音画一致性。打破时长限制，多镜头故事创作更自由。 -中国内地 +视频生成 -cosyvoice-v3.5-plus、cosyvoice-v3.5-flash +2026-03-26 -CosyVoice3.5 模型上线，专注声音复刻与设计，支持指令控制语音合成效果。[实时语音合成-CosyVoice /Sambert](https://help.aliyun.com/zh/model-studio/text-to-speech) +`kling/kling-v3-omni-video-generation` -推理模型 +新增“全能参考”，支持3-8秒视频或多图锚定角色元素。可匹配原声及口型驱动，实现角色本色呈现。视频一致性更强，表现更灵动。支持音画同步、智能分镜。 -2026-02-24 +图片生成 -中国内地 +2026-03-26 -MiniMax-M2.5 +`kling/kling-v3-omni-image-generation` -稀宇科技（MiniMax）推出的新模型，响应速度快，擅长编程、办公等任务。[使用方法](https://help.aliyun.com/zh/model-studio/minimax-api) +解锁影视级叙事画面，新增系列组图及2K/4K直出。深度解析提示词视听元素，精确响应创作指令。支持自由多参考图及全面效果升级，适合分镜、剧情概念图及场景设定。 -推理模型 +图片生成 -2026-02-24 +2026-03-26 -中国内地 +`kling/kling-v3-image-generation` -qwen3.5-flash、qwen3.5-flash-2026-02-23、qwen3.5-122b-a10b、qwen3.5-27b、qwen3.5-35b-a3b +支持最多10张参考图，可锁定主体、元素和色调，保证风格一致。融合风格转绘、人像/角色参考、多图融合及局部重绘，操作灵活。人像细节真实，整体画面细腻丰富，色彩氛围兼具影视感。 -阿里巴巴推出的最新模型千问3.5-Flash和开源模型，支持文本、图像和视频输入，响应速度快，综合表现接近qwen3.5-plus，支持内置[工具调用](https://help.aliyun.com/zh/model-studio/tool-calls/)。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +文本生成、深度思考 -代码模型 +2026-03-26 -2026-02-20 +`ZHIPU/GLM-5` -中国内地 +智谱新一代旗舰基座,面向AgenticEngineering,实现从代码到工程的范式跃迁,擅长复杂系统工程与长程Agent任务。 -qwen3-coder-next +文本生成 -Qwen3系列新一代开源代码生成模型，支持多轮工具交互，提升了对仓库级别代码的理解能力和对AI编程工具的适配性。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +2026-03-23 -推理模型 +`qwen-deep-research-2025-12-15` -2026-02-18 +千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。 -中国内地 +多模态向量 -glm-5 +2026-03-20 -智谱推出的最新模型，专为编程与智能体场景打造，擅长复杂的系统工程与长程Agent任务。[GLM-阿里云](https://help.aliyun.com/zh/model-studio/glm) +`tongyi-embedding-vision-plus-2026-03-06` -推理模型 +Tongyi-Embedding-Vision是基于LLM底座的视觉多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文、以文搜图文等下游多样化任务场景。 2026-03-06版本在保留极致性价比优势的同时，基于Qwen3底座实现了效果与功能全面升级，包括全场景性能提升、多分辨率模式/多向量维度/多语言能力/融合向量等能力的支持。 -2026-02-16 +多模态向量 -中国内地 +2026-03-20 -qwen3.5-plus、qwen3.5-plus-2026-02-15、qwen3.5-397b-a17b +`tongyi-embedding-vision-flash-2026-03-06` -阿里巴巴推出的最新模型千问3.5-Plus和开源模型，支持文本、图像和视频输入，在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中表现卓越，支持内置[工具调用](https://help.aliyun.com/zh/model-studio/tool-calls/)。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +Tongyi-Embedding-Vision是基于LLM底座的视觉多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文、以文搜图文等下游多样化任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，具备极高性价比。 2026-03-06版本在保留极致性价比优势的同时，基于Qwen3底座实现了效果与功能全面升级，包括全场景性能提升、多分辨率模式/多向量维度/多语言能力/融合向量等能力的支持。 -语音识别 +文本生成、深度思考 -2026-02-13 +2026-03-20 -中国内地 +`MiniMax/MiniMax-M2.7` -qwen3-asr-flash-realtime-2026-02-10 +M2.7 能够自行构建复杂 Agent Harness，并基于 Agent Teams、复杂 Skills、Tool Search tool 等能力，完成高度复杂的生产力任务。 -千问实时语音识别新增最新快照模型，较 qwen3-asr-flash-realtime-2025-10-27 效果更优。[实时语音识别](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition-user-guide) +视频生成 -语音识别 +2026-03-19 -2026-02-12 +`pixverse/pixverse-v5.6-t2v` -中国内地 +输入文字描述，秒级生成与语义精准匹配的高质量视频，支持多种风格。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。 -fun-asr-flash-8k-realtime、fun-asr-flash-8k-realtime-2026-01-28 +视频生成 -新增基于Fun-ASR大模型架构的小尺寸ASR模型，专为8kHz场景优化，适合对成本敏感的客户。[实时语音识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition) +2026-03-19 -语音合成 +`pixverse/pixverse-v5.6-r2v` -2026-02-10 +输入2–7张图像，智能融合不同主体，保持风格统一与动作协调，轻松构建丰富叙事场景，提升内容可控性与创意自由度。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。 -中国内地 +视频生成 -qwen3-tts-instruct-flash、qwen3-tts-instruct-flash-2026-01-26 +2026-03-19 -千问语音合成上线Instruct（指令控制）模型，支持通过自然语言指令精准控制合成效果。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +`pixverse/pixverse-v5.6-kf2v` -语音合成 +在任意两张图片之间实现无缝转换，实现更流畅自然的场景过渡，打造视觉冲击力强的画面效果。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。 -2026-02-10 +视频生成 -中国内地 +2026-03-19 -qwen3-tts-vd-2026-01-26 +`pixverse/pixverse-v5.6-it2v` -千问语音合成上线声音设计模型，可通过文本描述创建定制化音色。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +上传任意图片，自由定制剧情、节奏与风格，生成生动连贯的视频。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。 语音合成 -2026-02-10 - -中国内地 +2026-03-19 -qwen3-tts-vc-2026-01-22 +`MiniMax/speech-2.8-turbo` -千问语音合成上线声音复刻模型，可基于真实音频样本快速复刻音色。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。 语音合成 -2026-02-04 +2026-03-19 -中国内地 +`MiniMax/speech-2.8-hd` -qwen3-tts-instruct-flash-realtime、qwen3-tts-instruct-flash-realtime-2026-01-22 +MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。 -千问实时语音合成新增Instruct（指令控制）模型，支持通过自然语言指令精准控制合成效果。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +语音合成 -参考生视频 +2026-03-19 -2026-02-02 +`MiniMax/speech-02-turbo` -中国内地 +MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。 -wan2.6-r2v-flash +语音合成 -基于参考视频和图像的角色形象，生成多镜头视频，支持自动配音。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +2026-03-19 -文生文与视觉理解 +`MiniMax/speech-02-hd` -2026-01-30 +MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。 -中国内地 +深度思考、视觉理解 -kimi-k2.5 +2026-03-18 -由月之暗面（Moonshot AI）公司推出的视觉理解模型，在代码生成、视觉理解等通用智能任务中表现突出。同时支持图像、视频与文本输入、对话与 Agent 任务。[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api) +`gui-plus-2026-02-26` -文生文 +GUI系列模型，相较于GUI-Plus的模型，支持思考模式与非思考模式，模型在处理跨平台、多app任务的理解和执行等效果得到了全方位的大幅度提升。支持工具调用。此版本为2026年2月26日快照。 -2026-01-29 +语音识别 -中国内地 +2026-03-05 -tongyi-xiaomi-analysis-flash、tongyi-xiaomi-analysis-pro +`fun-asr-realtime-2026-02-28` -通义晓蜜对话分析专注于对话信息抽取、场景分类、满意度判定等分析需求，擅长处理复杂业务逻辑的质检规则，支持自定义分析标准，具备强大的多轮对话理解和语义推理能力。[对话分析（Tongyi-Xiaomi-Analysis）](https://help.aliyun.com/zh/model-studio/dialogue-analysis) +通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。此版本为2026年2月28日的快照版本，在近场VAD方面进行了优化。 语音识别 -2026-01-28 +2026-03-03 -中国内地 +`qwen3-asr-flash-2026-02-10` -qwen3-asr-flash-filetrans、qwen3-asr-flash-filetrans-2025-11-17 +通义千问3-ASR-Flash，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2026年2月10日的快照版本。 -千问3-ASR-Flash-Filetrans系列模型现已支持词级别时间戳，通过设置新参数 `enable_words`，获取毫秒级的词/字对齐信息，并体验更符合语义的精细化断句。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +图片生成 -推理模型 +2026-03-03 -2026-01-27 +`qwen-image-2.0-pro-2026-03-03` -中国内地 +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。该版本为2026年3月3日快照。 -qwen3-max-2026-01-23 +图片生成 -相较于 2025 年 9 月 23 日的快照版本，有效融合了思考模式与非思考模式，显著提升了模型的整体性能。在思考模式下，模型集成了 Web 搜索、网页信息提取和代码解释器三项工具，通过在思考过程中引入外部工具，在复杂问题上实现更高的准确率。[OpenAI兼容-Responses](https://help.aliyun.com/zh/model-studio/compatibility-with-openai-responses-api) +2026-03-03 -推理模型 +`qwen-image-2.0` -2026-01-27 +`qwen-image-2.0-2026-03-03` -中国内地 +Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。 -MiniMax-M2.1 +文本生成 -MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，其核心优势在于多语言编程能力以及作为Agent解决复杂任务的能力。[使用方法](https://help.aliyun.com/zh/model-studio/minimax-api) +2026-02-28 -视觉理解 +`qwen-flash-character-2026-02-26` -2026-01-23 +千问系列多语言角色扮演模型，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。本版本为2026年2月26日快照。 -中国内地 +语音合成 -qwen3-vl-flash-2026-01-22 +2026-02-27 -千问VL的全新快照版模型，有效融合了思考模式与非思考模式，相较于 2025 年 10 月 15 日的快照版本，显著提升了模型的整体性能，在通用视觉识别、安防、巡店、巡检、拍照解题等业务场景中实现了更高准确率的推理。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +`cosyvoice-v3.5-plus` -图生视频 +CosyVoice-v3.5-Plus是通义实验室CosyVoice系列的超高表现力语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。 -2026-01-17 +语音合成 -中国内地 +2026-02-27 -wan2.6-i2v-flash +`cosyvoice-v3.5-flash` -支持生成有声与无声视频，两类视频按各自计费规则独立计费；同时具备多镜头叙事能力与音频处理能力。[万相-图生视频-基于首帧（2.1-2.6）](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +CosyVoice-v3.5-Flash是通义实验室CosyVoice系列的高性能语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。 -图像编辑 +文本生成、深度思考 -2026-01-17 +2026-02-24 -中国内地 +`MiniMax-M2.5` -qwen-image-edit-max、qwen-image-edit-max-2026-01-16 +MiniMax-M2.5是MiniMax推出的旗舰级开源大模型，经过数十万个真实复杂环境中的大规模强化学习训练，M2.5 在编程、工具调用和搜索、办公等生产力场景都达到或者刷新了行业的 SOTA。 -千问图像编辑模型Max系列，具备更稳定、丰富的编辑能力，增强了工业设计与几何推理能力，并提升了角色一致性与编辑的精准度。[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide) +文本生成、深度思考、视觉理解 -语音合成 +2026-02-23 -2026-01-16 +`qwen3.5-flash` -中国内地 +`qwen3.5-flash-2026-02-23` -qwen3-tts-vc-realtime-2026-01-15 +Qwen3.5原生视觉语言系列Flash模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步；响应速度快，兼具推理速度和性能。 -千问实时语音合成新增最新快照模型，[声音复刻（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning)效果进一步优化，较 qwen3-tts-vc-realtime-2025-11-27 更自然、更贴近原声。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +文本生成、深度思考、视觉理解 -文生图 +2026-02-23 -2026-01-12 +`qwen3.5-35b-a3b` -中国内地 +Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。 -qwen-image-plus-2026-01-09 +文本生成、深度思考、视觉理解 -千问图像生成的全新快照版模型，为qwen-image-max的蒸馏加速版，支持快速生成高质量图像。[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api) +2026-02-23 -推理模型 +`qwen3.5-27b` -2026-01-12 +Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。 -中国内地 +文本生成、深度思考、视觉理解 -deepseek-v3.2 +2026-02-23 -deepseek-v3.2 模型支持隐式缓存与显式缓存，可提升响应速度，并在不影响回复效果的前提下降低使用成本。[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache) +`qwen3.5-122b-a10b` -语音识别 +Qwen3.5系列122B-A10B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现仅次于Qwen3.5-397B-A17B，文本能力显著优于Qwen3-235B-2507，视觉能力优于Qwen3-VL-235B。 -2026-01-06 +文本生成 -中国内地 +2026-02-19 -qwen3-asr-flash、qwen3-asr-flash-2025-09-08 +`qwen3-coder-next` -千问3-ASR-Flash支持OpenAI兼容模式。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +Qwen3系列新一代代码生成模型，效果接近Qwen3-Coder-Plus兼具更优性能。模型重点优化仓库级别理解、支持多轮工具交互、提升对于agentic coding类工具的适配能力。 -语音合成 +文本生成、深度思考 -2026-01-05 +2026-02-18 -中国内地 +`glm-5` -cosyvoice-v3-flash +GLM-5是面向Coding与Agent场景的新一代大模型，在复杂系统工程与长程任务中达到开源 SOTA，真实编程体验逼近 Claude Opus 级别；基于 744B 新基座、异步强化学习与稀疏注意力，实现从“写代码”到“写工程”的全面升级。 -语音合成CosyVoice新增24个音色（详情请参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)）： +文本生成、深度思考、视觉理解 -- 方言：龙嘉怡、龙老铁 - -- 出海营销：loongkyong、loongtomoka - -- 诗词朗诵：龙飞 - -- 语音助手：龙小淳、龙小夏、YUMI - -- 社交陪伴：龙橙、龙泽、龙哲、龙颜、龙星、龙天、龙婉、龙嫱、龙菲菲、龙浩 - -- 有声书：龙三叔、龙媛、龙悦、龙修、龙楠 - -- 新闻播报：龙书 - +2026-02-15 -文生图 +`qwen3.5-plus` -2025-12-31 +`qwen3.5-plus-2026-02-15` -中国内地 +Qwen3.5原生视觉语言系列Plus模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在多项任务评测中，3.5系列均展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 -qwen-image-max、qwen-image-max-2025-12-30 +文本生成、深度思考、视觉理解 -千问图像生成模型Max系列，相较于Plus系列提升了图像的真实感与自然度，有效降低了AI合成痕迹，在人物质感、纹理细节和文字渲染等方面表现突出。[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api) +2026-02-15 -推理模型 +`qwen3.5-397b-a17b` -2025-12-29 +Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。 -中国内地 +语音识别 -glm-4.7 +2026-02-13 -智谱最新旗舰模型，编程能力更强，多步骤推理与执行更稳定。模型总参数达355B，支持复杂任务的分步规划、代码生成与工具协同；问答流畅，写作沉浸感强，创意角色扮演表现突出。[GLM-阿里云](https://help.aliyun.com/zh/model-studio/glm) +`qwen3-asr-flash-realtime-2026-02-10` -图像编辑 +Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2026年2月10日的快照版本。 -2025-12-23 +文本生成、深度思考、视觉理解 -中国内地 +2026-02-13 -qwen-image-edit-plus-2025-12-15 +`kimi/kimi-k2.5` -千问图像编辑发布的最新快照模型，相较于上一版本提升了角色一致性、工业设计能力和几何推理能力，并优化了编辑后的图片与原图在空间布局、纹理和风格上的匹配度，编辑效果更精准。[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide) +Kimi K2.5 是 Kimi 在2026年最新推出的智能模型，在 Agent、代码、视觉理解及一系列通用智能任务上取得开源 SoTA 表现。同时 Kimi K2.5 也是 Kimi 迄今最全能的模型，原生的多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与 Agent 任务。 -文生图 +文本生成、深度思考 -2025-12-19 +2026-02-13 -中国内地 +`MiniMax/MiniMax-M2.5` -z-image-turbo +智能体世界的SOTA，专为智能体2.0设计，将编码扩展到现实世界包括工作空间、娱乐和个人助理。模型亮点：全球SOTA开源编码与智能体模型；SWE-bench Pro和SWE-bench Verified得分高于Opus 4.6；在Excel、搜索与研究以及文档摘要方面的全球SOTA；未来工作空间的完美主力模型；闪电般快速：优化思维效率，100+ TPS，实现比 Opus 快 3 倍的速度；极致性价比，以支持始终在线的智能体。 -轻量级文生图模型，可快速生成高质量图像，支持中英双语渲染、复杂语义理解和多风格题材，并可灵活适配多种分辨率与宽高比。[文生图Z-Image](https://help.aliyun.com/zh/model-studio/z-image-api-reference) +文本生成、深度思考 -视觉理解 +2026-02-13 -2025-12-19 +`MiniMax/MiniMax-M2.1` -中国内地 +M2.1 的设计初衷在于打破“最顶级的 Agent 能力仅存在于闭源模型”的壁垒。我们在模型层面进行了针对性优化，显著提升了模型在代码生成、工具调用、复杂指令遵循及长程规划任务中的性能。从自动化进行多语言的软件开发，到执行多步骤的复杂办公工作流，MiniMax-M2.1 均表现出卓越的稳定性。我们致力于为开发者提供一个完全透明、可控且高可用的基础模型，以构建下一代自主智能体应用。 -qwen3-vl-plus-2025-12-19 +实时语音识别 -千问VL的全新快照版模型，指令遵循能力更强，具有更低的延迟。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +2026-02-12 -语音识别 +`fun-asr-flash-8k-realtime` -2025-12-19 +`fun-asr-flash-8k-realtime-2026-01-28` -中国内地 +通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。 -qwen3-asr-flash-filetrans、qwen3-asr-flash-filetrans-2025-11-17、qwen3-asr-flash、qwen3-asr-flash-2025-09-08 +语音合成 -新增捷克语、丹麦语等共 9 种语言的语音识别支持。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +2026-02-10 -语音识别 +`qwen3-tts-vd-2026-01-26` -2025-12-17 +Qwen3-TTS-VD模型是通义最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月26日快照版本模型。 -中国内地 +语音合成 -qwen3-asr-flash-realtime、qwen3-asr-flash-realtime-2025-10-27 +2026-02-10 -新增捷克语、丹麦语等共 9 种语言的语音识别支持。[实时语音识别](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition-user-guide) +`qwen3-tts-vc-2026-01-22` -语音识别 +Qwen3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月22日快照版本模型。 -2025-12-17 +语音合成 -中国内地 +2026-02-10 -qwen3-asr-flash、qwen3-asr-flash-2025-09-08 +`qwen3-tts-instruct-flash` -支持任意采样率和声道的音频。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +`qwen3-tts-instruct-flash-2026-01-26` -语音识别 +Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。 -2025-12-17 +文本生成、深度思考、视觉理解 -中国内地 +2026-01-30 -fun-asr-mtl、fun-asr-mtl-2025-08-25 +`kimi-k2.5` -支持对中、英、日、韩等共 31 种语言的语音识别，尤其适合东南亚出海场景。[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition) +kimi-k2.5是月之暗面迄今发布最全能的模型，原生多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与Agent任务。 -文生图 +视频生成 -2025-12-16 +2026-01-29 -中国内地 +`wan2.6-r2v-flash` -wan2.6-t2i +万相2.6-参考生视频-Flash，生成更快性价比更高。支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍 -新增同步接口。支持在总像素面积与宽高比约束内，自由选尺寸。[万相-文生图V2](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference) +文本向量 -图像生成与编辑 +2026-01-29 -2025-12-16 +`qwen3-vl-rerank` -中国内地 +Qwen3-VL-Rerank重排模型，它能够深入理解文本、图片、视频的丰富多模态信息。在初步检索获得结果后，Qwen3-VL-Rerank 能够运用其先进的跨模态关联能力，对候选项目进行智能化的二次排序，将最相关的结果置于显要位置。通用用于提升跨模态搜索的准确率、优化图搜和视频检索的精准度、辅助图像聚类的分组质量、以及实现复杂多模态信息的高效检索和精确打标。 -wan2.6-image +文本生成、深度思考 -支持图像编辑和图文混合输出。[万相-图像生成与编辑2.6](https://help.aliyun.com/zh/model-studio/wan-image-generation-api-reference) +2026-01-28 -图生视频-基于首帧 +`siliconflow/deepseek-v3.2` -2025-12-16 +DeepSeek-V3.2 是一款兼具高计算效率与卓越推理和 Agent 性能的模型。其方法建立在三大关键技术突破之上：DeepSeek 稀疏注意力（DSA），一种高效的注意力机制，在保持模型性能的同时显著降低了计算复杂性，并特别针对长上下文场景进行了优化；可扩展的强化学习框架，通过该框架，模型性能可与 GPT-5 相媲美，其高算力版本在推理能力上可与 Gemini-3.0-Pro 匹敌；以及大规模 Agent 任务合成管线，旨在将推理能力整合到工具使用场景中，从而提高在复杂交互环境中的指令遵循和泛化能力。该模型在 2025 年国际数学奥林匹克（IMO）和国际信息学奥林匹克（IOI）中取得了金牌表现 -中国内地 +文本生成、深度思考 -wan2.6-i2v +2026-01-28 -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[万相-图生视频-基于首帧（2.1-2.6）](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +`siliconflow/deepseek-v3.1-terminus` -参考生视频 +DeepSeek-V3.1-Terminus 是由深度求索（DeepSeek）发布的 V3.1 模型的更新版本，定位为混合智能体大语言模型。此次更新在保持模型原有能力的基础上，专注于修复用户反馈的问题并提升稳定性。它显著改善了语言一致性，减少了中英文混用和异常字符的出现。模型集成了“思考模式”（Thinking Mode）和“非思考模式”（Non-thinking Mode），用户可通过聊天模板灵活切换以适应不同任务。作为一个重要的优化，V3.1-Terminus 增强了代码智能体（Code Agent）和搜索智能体（Search Agent）的性能，使其在工具调用和执行多步复杂任务方面更加可靠 -2025-12-16 +文本生成 -中国内地 +2026-01-28 -wan2.6-r2v +`siliconflow/deepseek-v3-0324` -基于参考视频的角色形象和音色，生成多镜头视频，支持自动配音。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +新版 DeepSeek-V3 （DeepSeek-V3-0324）与之前的 DeepSeek-V3-1226 使用同样的 base 模型，仅改进了后训练方法。新版 V3 模型借鉴 DeepSeek-R1 模型训练过程中所使用的强化学习技术，大幅提高了在推理类任务上的表现水平，在数学、代码类相关评测集上取得了超过 GPT-4.5 的得分成绩。此外该模型在工具调用、角色扮演、问答闲聊等方面也得到了一定幅度的能力提升。 -文生视频 +文本生成、深度思考 -2025-12-16 +2026-01-28 -中国内地 +`siliconflow/deepseek-r1-0528` -wan2.6-t2v +DeepSeek-R1-0528 是一款强化学习（RL）驱动的推理模型，解决了模型中的重复性和可读性问题。在 RL 之前，DeepSeek-R1 引入了冷启动数据，进一步优化了推理性能。它在数学、代码和推理任务中与 OpenAI-o1 表现相当，并且通过精心设计的训练方法，提升了整体效果。 -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +深度思考、视觉理解 -声音设计 +2026-01-26 -2025-12-16 +`qwen3-vl-plus` -中国内地 +Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。 -qwen-voice-design +文本生成、深度思考 -千问发布声音设计模型，通过文本描述生成定制化音色。结合qwen3-tts-vd-realtime-2025-12-16模型使用生成语音，覆盖 10 种语言。[声音设计（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design) +2026-01-23 -语音合成 +`qwen3-max-2026-01-23` -2025-12-16 +千问3系列Max模型，相较2025年9月23日快照，此版本实现思考模式和非思考模式的有效融合，模型整体效果得到全方位的大幅度提升。在思考模式下，同时发布Web搜索、Web信息提取和代码解释器工具能力，使得模型在慢思考的同时，能够通过引入外部工具，以更高的准确性解决更有难度的问题。此版本为2026年1月23日快照。 -中国内地 +文本生成、深度思考 -qwen3-tts-vd-realtime-2025-12-16（快照版） +2026-01-23 -千问实时语音合成发布全新快照版模型，可使用[声音设计（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design)生成的音色进行低延迟、高稳定性的实时合成；支持多语言输出；能根据文本自动调节语气，并优化复杂文本的合成表现。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +`MiniMax-M2.1` -语音识别 +MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，以多语言编程与长链 Agent 能力为核心优势。 -2025-12-12 +视觉理解 -中国内地 +2026-01-22 -fun-asr、fun-asr-2025-11-07 +`qwen3-vl-flash-2026-01-22` -录音文件识别-Fun-ASR功能更新： +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于2025年10月15日快照，模型整体效果有大幅提升：在通用视觉识别与推理方面有增强；在安防、巡店、巡检、拍照解题等业务场景识别效果上提升显著。此版本为2026年1月22日快照版本。 -- 支持歌唱识别，能实现整首歌曲的转写，详情请参见[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition)。 - +多模态向量 -语音合成 +2026-01-21 -2025-12-11 +`qwen3-vl-embedding` -中国内地 +基于Qwen3-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景 -cosyvoice-v3-flash、cosyvoice-v3-plus +语音合成 -CosyVoice模型音色更新： +2026-01-21 -- cosyvoice-v3-flash模型新增5个系统音色：longanrou\_v3、longyingjing\_v3、longyingling\_v3、longanling\_v3和longhan\_v3，均支持时间戳与SSML功能，详情请参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)。 - -- cosyvoice-v3-flash和cosyvoice-v3-plus模型的复刻音色功能增强：支持时间戳与SSML功能，韵律效果提升。请参见[CosyVoice声音复刻/设计API](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api)创建新音色体验。 - +`qwen3-tts-instruct-flash-realtime` -全模态 +`qwen3-tts-instruct-flash-realtime-2026-01-22` -2025-12-04 +千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。 -中国内地 +视频生成 -qwen3-omni-flash-2025-12-01 +2026-01-15 -千问Omni发布的最新快照模型，支持的音色增加至49种，模型的指令跟随能力大幅升级，能高效理解文本、图像、音频、视频。[非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni) +`wan2.6-i2v-flash` -实时多模态 +万相2.6-图生视频-Flash，生成更快更高性价比。智能分镜调度支持多镜头叙事，多人稳定对话，更自然真实音色，最高支持15秒时长生成 -2025-12-04 +图片生成 -中国内地 +2026-01-15 -qwen3-omni-flash-realtime**\-**2025-12-01 +`qwen-image-edit-max` -千问Omni 实时版发布的最新快照模型，提供了低延迟的多模态交互能力，支持的音色增加至49种，模型的指令跟随能力和交互体验大幅升级。[实时（Qwen-Omni-Realtime）](https://help.aliyun.com/zh/model-studio/realtime) +`qwen-image-edit-max-2026-01-16` -语音翻译 +千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。 -2025-12-04 +语音合成 -中国内地 +2026-01-14 -qwen3-livetranslate-flash、qwen3-livetranslate-flash-2025-12-01 +`qwen3-tts-vd-realtime-2026-01-15` -千问3-LiveTranslate-Flash 是音视频翻译模型，支持 18 种语言（包括中文、英文、俄文、法文等）互译，可结合视觉上下文提升翻译准确性，并输出文本与语音。[音视频文件翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-livetranslate-flash) +千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。 -推理模型 +语音合成 -2025-12-04 +2026-01-14 -中国内地 +`qwen3-tts-vc-realtime-2026-01-15` -deepseek-v3.2 +千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。 -DeepSeek-V3.2是引入DeepSeek Sparse Attention（一种稀疏注意力机制）的正式版模型，也是DeepSeek推出的首个将思考融入工具使用的模型，同时支持思考模式与非思考模式的工具调用。 +文本生成 -[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api) +2026-01-13 -多语言翻译 +`qwen-flash-character` -2025-11-28 +千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 -中国内地 +文本生成 -qwen-mt-lite +2026-01-09 -千问基础级文本翻译大模型，支持31个语种互译，相较于qwen-mt-flash响应更快，成本更低，适用于等对延迟敏感的场景。[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation) +`tongyi-xiaomi-analysis-pro` -声音复刻 +通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。 -2025-11-27 +文本生成 -中国内地 +2026-01-09 -qwen-voice-enrollment +`tongyi-xiaomi-analysis-flash` -千问发布声音复刻模型，仅需 5 秒以上音频即可快速生成高相似度声音。结合qwen3-tts-vc-realtime-2025-11-27模型使用，可高保真复刻并实时输出某人的声音，覆盖 11 种语言。[声音复刻（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning) +通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。 -语音合成 +图片生成 -2025-11-27 +2026-01-09 -中国内地 +`qwen-image-plus-2026-01-09` -qwen3-tts-vc-realtime-2025-11-27（快照版） +千问系列图像生成模型，具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色。此版本为2026年1月9日快照，为Qwen-Image-Max的蒸馏加速版，可以更快速地生成高质量图片。 -千问实时语音合成发布全新快照版模型，可使用[声音复刻（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning)生成的音色进行低延迟、高稳定性的实时合成；支持多语言输出；能根据文本自动调节语气，并优化复杂文本的合成表现。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +图片生成 -语音合成 +2025-12-30 -2025-11-27 +`qwen-image-max` -中国内地 +`qwen-image-max-2025-12-30` -qwen3-tts-flash-realtime-2025-11-27（快照版） +千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。 -千问实时语音合成发布全新快照版模型，低延迟且稳定性高；音色更丰富，同一音色支持多语言输出；能根据文本自动调节语气，并提升复杂文本的合成表现。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +文本生成、深度思考 -语音合成 +2025-12-25 -2025-11-27 +`glm-4.7` -中国内地 +智谱最新旗舰，具备更强的编程能力与更稳定的多步骤推理/执行能力。总参数355B，支持长程任务规划、编码、工具协同，问答自然、写作沉浸、创意角色扮演能力强。 -qwen3-tts-flash-2025-11-27（快照版） +图片生成 -千问语音合成发布全新快照版模型，音色更丰富；同一音色支持多语言输出；可自适应文本调节语气，并优化复杂文本的合成能力。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +2025-12-18 -文字提取 +`z-image-turbo` -2025-11-21 +Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。 -中国内地 +深度思考、视觉理解 -qwen-vl-ocr-2025-11-20（快照版） +2025-12-18 -千问文字提取模型，该快照版基于Qwen3-VL架构，大幅提升文档解析、文字定位能力。[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr) +`qwen3-vl-plus-2025-12-19` -语音识别 +Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合。相较于9月23日快照，在推理及分析任务、风格控制上表现更优；同时拥有更低的延时和更快的响应速度。此版本为2025年12月19日快照版本。 -2025-11-20 +视频生成 -中国内地 +2025-12-16 -qwen3-asr-flash-filetrans、qwen3-asr-flash-filetrans-2025-11-17（快照版） +`wan2.6-r2v` -千问录音文件识别发布了新模型，专为音频文件的异步转写设计，支持最长12小时录音。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +万相2.6-参考生视频，支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍。提醒：当使用视频进行参考时，输入视频也会计入费用，详见模型计费文档。 -界面交互 +图片生成 -2025-11-20 +2025-12-15 -中国内地 +`wan2.6-t2i` -gui-plus +万相2.6-文生图，画面质感、美学表现、指令遵循升级，在艺术风格精准控制、真实感人像、长文本生图及广泛历史文化IP覆盖上均表现出卓越能力，可生成高质量且富有表现力的视觉内容。 -GUI-Plus 可基于屏幕截图和自然语言指令来解析用户意图，并转换为标准化的图像用户界面（GUI）操作（如点击、输入、滚动等），供外部系统决策或执行。相较于千问VL系列模型，提升了GUI操作的准确性。[界面交互](https://help.aliyun.com/zh/model-studio/gui-automation) +图片生成 -语音识别 +2025-12-15 -2025-11-19 +`wan2.6-image` -中国内地 +万相2.6-图像生成，全能图像生成模型，支持图文一体化推理生成，具备多图创意融合、商用级一致性、美学要素迁移与镜头光影精确控制，全面提升图像生成的一致性、可控性和表现力。 -fun-asr-realtime-2025-11-07（快照版） +图片生成 -Fun-ASR实时语音识别发布了全新快照版模型，优化远场语音活动检测（VAD）以提升识别准确率与稳定性，并在原有中英文识别基础上新增支持中文多地方言及日语。[实时语音识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition) +2025-12-15 -语音识别 +`qwen-image-edit-plus-2025-12-15` -2025-11-19 +千问系列图像编辑Plus模型，相较10月30日快照提升角色一致性、工业设计能力、几何推理能力；同时集成例如打光等Lora能力、减轻偏移问题。此版本为2025年12月15日快照。 + +语音合成 -中国内地 +2025-12-12 -fun-asr-2025-11-07（快照版） +`qwen3-tts-vd-realtime-2025-12-16` -Fun-ASR录音文件识别发布了全新快照版模型，优化远场语音活动检测（VAD）以提升识别准确率与稳定性，并在原有中英文识别基础上新增支持中文多地方言及日语。[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition) +千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年12月16日快照版本模型。 语音合成 -2025-11-19 +2025-12-12 -中国内地 +`qwen-voice-design` -cosyvoice-v3-flash +千问voice-design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 -较之前版本提升了发音准确性、音色相似度，并且增加了更多小语种支持（德、西、法、意、俄）。[实时语音合成-CosyVoice /Sambert](https://help.aliyun.com/zh/model-studio/text-to-speech) +实时全模态 -推理模型 +2025-12-04 -2025-11-11 +`qwen3-omni-flash-realtime` -中国内地 +`qwen3-omni-flash-realtime-2025-12-01` -kimi-k2-thinking +千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。 -由月之暗面（Moonshot AI）公司推出的思考模型，具有通用智能体能力和推理能力，擅长深度推理，并能通过多步工具调用解决各类难题。[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api) +全模态 -多语言翻译 +2025-12-04 -2025-11-10 +`qwen3-omni-flash` -中国内地 +`qwen3-omni-flash-2025-12-01` -qwen-mt-flash +千问3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。 -相较于qwen-mt-turbo支持流式增量输出，整体性能表现有所提升。[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation) +实时语音识别 -推理模型 +2025-12-04 -2025-11-03 +`qwen3-livetranslate-flash` -中国内地 +`qwen3-livetranslate-flash-2025-12-01` -qwen3-max-preview +Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。 -qwen3-max-preview 模型的思考模式：在整体推理能力上显著提升，尤其在智能体编程、常识推理，以及数学、科学和通用任务方面表现更优。[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking) +视频生成 -图像编辑 +2025-12-03 -2025-10-31 +`wan2.6-t2v` -中国内地 +万相2.6-文生视频，智能分镜调度支持多镜头叙事，能够生成主体、场景和氛围一致的多镜头叙事视频，最高支持15秒时长，更高品质的声音生成，更好的指令遵循和视觉质量 -qwen-image-edit-plus、qwen-image-edit-plus-2025-10-30 +视频生成 -在qwen-image-edit的基础上优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间，且支持单次请求返回多张图片。[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide) +2025-12-03 -实时语音识别 +`wan2.6-i2v` -2025-10-27 +万相2.6-图生视频，智能分镜调度支持多镜头叙事，更高品质的声音生成，多人稳定对话，更自然真实音色，最高支持15秒时长生成 -中国内地 +文本生成、深度思考 -qwen3-asr-flash-realtime、qwen3-asr-flash-realtime-2025-10-27 +2025-12-02 -千问实时语音识别大模型具备自动语种识别功能，可识别 11 种语音类型，并能在复杂音频环境下较为准确地转录。[实时语音识别](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition-user-guide) +`deepseek-v3.2` -视觉理解 +DeepSeek-V3.2是引入DeepSeek Sparse Attention（一种稀疏注意力机制）的正式版模型，也是DeepSeek推出的首个将思考融入工具使用的模型，同时支持思考模式与非思考模式的工具调用。 -2025-10-21 +文本生成、深度思考 -中国内地 +2025-12-01 -qwen3-vl-32b-thinking、qwen3-vl-32b-instruct +`qwen-plus-2025-12-01` -Qwen3-VL系列 32B 的Dense模型，文档识别与理解、空间感知与万物识别能力、视觉2D检测与空间推理能力均表现出色，适合通用场景下的复杂感知任务。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +本版本为2025年12月1日快照，相较7月28日快照在推理能力上有提升；智能体能力、多轮工具调用能力进一步增强；主观创作类任务表现更优。支持1M上下文长度，按照上下文长度进行阶梯计费。 -推理模型 +语音合成 -2025-10-21 +2025-11-27 -中国内地 +`qwen3-tts-vc-realtime-2025-11-27` -glm-4.6 +千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年11月27日快照版本模型。 -GLM新一代旗舰模型，核心能力较glm-4.5全面提升。总参数量为3550 亿，激活参数320亿，上下文窗口扩展至200K。[GLM-阿里云](https://help.aliyun.com/zh/model-studio/glm) +语音合成 -视觉理解 +2025-11-27 -2025-10-16 +`qwen3-tts-flash-realtime` -中国内地 +`qwen3-tts-flash-realtime-2025-11-27` -qwen3-vl-flash、qwen3-vl-flash-2025-10-15 +千问3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 -Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于开源版Qwen3-VL-30B-A3B，效果更优，响应速度更快。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +语音合成 -视觉理解 +2025-11-27 -2025-10-14 +`qwen3-tts-flash` -中国内地 +`qwen3-tts-flash-2025-11-27` -qwen3-vl-8b-thinking、qwen3-vl-8b-instruct +Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 -Qwen3-VL系列 8B 的Dense模型，占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面空间感知与万物识别能力。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +语音合成 -视觉理解 +2025-11-27 -2025-10-03 +`qwen-voice-enrollment` -中国内地 +千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 -qwen3-vl-30b-a3b-thinking、qwen3-vl-30b-a3b-instruct +语音识别 -基于Qwen3-VL新一代开源模型，响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级空间感知与万物识别能力，胜任复杂现实任务。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +2025-11-21 -推理模型 +`fun-asr` -2025-09-30 +百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本等同于2025年11月7日的快照版本。 -中国内地 +视觉理解 -deepseek-v3.2-exp +2025-11-20 -混合推理架构模型，同时支持思考模式与非思考模式，引入稀疏注意力机制，旨在提升处理长文本时的训练与推理效率，价格低于 deepseek-v3.1。详情参见[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api)。 +`qwen-vl-ocr` -图生视频 +`qwen-vl-ocr-2025-11-20` -2025-09-23 +千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。 -中国内地 +文本生成 -wan2.5-i2v-preview +2025-11-19 -新增音频能力，支持自动配音，或传入自定义音频文件，实现音画同步。[图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +`qwen-mt-lite` -文生视频 +基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 -2025-09-23 +语音识别 -中国内地 +2025-11-17 -wan2.5-t2v-preview +`qwen3-asr-flash-filetrans` -新增音频能力，支持自动配音，或传入自定义音频文件，实现音画同步。[文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +`qwen3-asr-flash-filetrans-2025-11-17` -图像编辑 +千问3-ASR-Flash的大文件转录版本，千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。 -2025-09-23 +实时语音识别 -中国内地 +2025-11-17 -wan2.5-i2i-preview +`fun-asr-realtime-2025-11-07` -支持输入文本、单图或多图实现基于主体一致性的图像编辑、多图融合和组图生成。[通用图像编辑2.5](https://help.aliyun.com/zh/model-studio/wan2-5-image-edit-api-reference) +通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。此版本为2025年11月7日的快照版本。 -文生图 +语音识别 -2025-09-23 +2025-11-17 -中国内地 +`fun-asr-2025-11-07` -wan2.5-t2i-preview +百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本为2025年11月7日的快照版本。 -取消单边限制，在总像素面积与宽高比约束内，自由选尺寸。[文生图V2](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference) +语音合成 -视觉理解 +2025-11-17 -2025-09-23 +`cosyvoice-v3-flash` -中国内地 +合成能力：CosyVoice-v3-Flash是通义实验室CosyVoice系列最新版高性能的语音合成大模型，较之前版本在自然度、音质、韵律、情感表现力上有更好的表现。该模型支持文本至语音的实时流式合成。克隆能力：CosyVoice-v3-Flash也是通义实验室CosyVoice系列最新版的语音克隆大模型，较之前版本提升了发音准确性、音色相似度，并且增加了更多小语种支持（德、西、法、意、俄）。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。 -qwen3-vl-plus、qwen3-vl-plus-2025-09-23、qwen3-vl-235b-a22b-thinking、qwen3-vl-235b-a22b-instruct +视觉理解 -Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力达到世界顶尖水平。此版本在视觉编码、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +2025-11-12 -文生文 +`gui-plus` -2025-09-23 +GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。 -中国内地 +文本生成、深度思考 -qwen3-max、qwen3-max-2025-09-23 +2025-11-10 -相较qwen3-max-preview版本，在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。模型列表 +`kimi-k2-thinking` -音视频翻译 +kimi-k2-thinking模型是月之暗面提供的具有通用 Agentic能力和推理能力的思考模型，它擅长深度推理，并可通过多步工具调用，帮助解决各类难题。 -2025-09-23 +文本生成 -中国内地 +2025-11-06 -qwen3-livetranslate-flash-realtime-2025-09-22 +`qwen-mt-flash` -qwen3-livetranslate-flash-realtime 是一款多语言音视频实时翻译模型，可识别 18 种语言，并实时翻译为 10 种语言的音频。[实时语音/音视频翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-5-livetranslate-flash-realtime) +基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 -文生图 +图片生成 -2025-09-23 +2025-10-30 -中国内地 +`qwen-image-edit-plus` -qwen-image-plus +`qwen-image-edit-plus-2025-10-30` -在复杂文本渲染方面表现突出，特别是中英文文本渲染，可实现复杂的图文混合布局，相较于qwen-image更具价格优势。[文生图（Qwen-Image)](https://help.aliyun.com/zh/model-studio/qwen-image-api) +千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。 -代码模型 +实时语音识别 -2025-09-23 +2025-10-27 -中国内地 +`qwen3-asr-flash-realtime` -qwen3-coder-plus-2025-09-23 +`qwen3-asr-flash-realtime-2025-10-27` -基于 Qwen3 的代码生成模型，相较上一版本（7月22日快照）在下游任务效果和工具调用方面鲁棒性有所提升，代码安全性增强。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +千问3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。 -多模态向量 +深度思考、视觉理解 -2025-09-23 +2025-10-21 -中国内地 +`qwen3-vl-32b-thinking` -tongyi-embedding-vision-plus、tongyi-embedding-vision-flash +Qwen3-VL系列最大尺寸Dense模型的推理版本，多模态推理能力仅次于Qwen3-VL-235B-Thinking，STEM&数学类解题能力、通用图像和视频理解能力出众，多模态Agent能力达到SOTA，适合做复杂多模态推理任务。 -以Qwen系列大语言模型为基座构建，围绕视觉向量化能力进行增强训练，支持文字、图像、视频三种模态。[文本与多模态向量化](https://help.aliyun.com/zh/model-studio/embedding) +视觉理解 -语音识别 +2025-10-21 -2025-09-23 +`qwen3-vl-32b-instruct` -中国内地 +Qwen3-VL系列最大尺寸Dense模型的非推理版本，综合表现仅次于Qwen3-VL-235B-Instruct，文档识别和理解能力出色，空间感知与万物识别能力强，视觉2D检测/空间推理能力达到SOTA，适合通用场景下的复杂感知任务。 -fun-asr-realtime、fun-asr-realtime-2025-09-15 +文本向量 -集成了创新的 RAG 技术，支持大规模热词自定义、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度，同时支持中英文自由切换，具备更强的噪声鲁棒性，适应多样复杂环境。[实时语音识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition) +2025-10-21 -实时语音合成 +`qwen3-rerank` -2025-09-22 +基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐开源Qwen3-Rerank系列模型 -中国内地 +多模态向量 -qwen3-tts-flash-realtime、qwen3-tts-flash-realtime-2025-09-18 +2025-10-21 -千问最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言、方言。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +`qwen2.5-vl-embedding` -语音合成 +基于Qwen2.5-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景 -2025-09-22 +文本生成、深度思考 -中国内地 +2025-10-21 -qwen3-tts-flash、qwen3-tts-flash-2025-09-18 +`glm-4.6` -千问最新的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言、方言。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +GLM新一代旗舰模型，核心能力较4.5全面提升。总参数量为3550 亿，激活参数320亿，上下文窗口扩展至200K。 -实时全模态 +深度思考、视觉理解 -2025-09-22 +2025-10-14 -中国内地 +`qwen3-vl-flash` -qwen3-omni-flash-realtime、qwen3-omni-flash-realtime-2025-09-15 +`qwen3-vl-flash-2025-10-15` -Qwen3系列多模态大模型的实时版，提供了低延迟的多模态交互能力，支持音频的流式输入，支持输出文本和音频，同时支持多种音色和语言。[实时（Qwen-Omni-Realtime）](https://help.aliyun.com/zh/model-studio/realtime) +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。 -全模态 +深度思考、视觉理解 -2025-09-22 +2025-09-30 -中国内地 +`qwen3-vl-8b-thinking` -qwen3-omni-flash、qwen3-omni-flash-2025-09-15 +Qwen3-VL系列8B Dense模型的Thinking版本，占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面升级图像/视频理解、空间感知与万物识别能力。 -Qwen3系列多模态模型，能高效理解文本、图像、音频和视频，支持119种语言的文本交互，具备卓越的指令跟随与系统提示定制能力，可轻松设定角色和对话风格，可用于构建语音助手、多媒体分析和内容创作应用。[非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni) +视觉理解 -音频理解 +2025-09-30 -2025-09-22 +`qwen3-vl-8b-instruct` -中国内地 +Qwen3-VL系列8B Dense模型的Instruct版本，占用显存更低，全面升级图像/视频理解、长视频长文档等超长上下文支持、空间感知与万物识别能力，胜任复杂现实任务。 -qwen3-omni-30b-a3b-captioner +深度思考、视觉理解 -以千问3-Omni为基座的开源模型，无需任何提示，自动为复杂语音、环境声、音乐、影视声效等生成精准、全面的描述，能识别说话人情绪、音乐元素（如风格、乐器）、敏感信息等，适用于音频内容分析、安全审核、意图识别、音频剪辑等多个领域。[音频理解-Qwen3-Omni-Captioner](https://help.aliyun.com/zh/model-studio/qwen3-omni-captioner) +2025-09-30 -图生视频 +`qwen3-vl-30b-a3b-thinking` -2025-09-19 +Qwen3-VL系列第二大MoE模型的Thinking版本，响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级图像/视频理解、空间感知与万物识别能力，胜任复杂现实任务。 -中国内地 +视觉理解 -wan2.2-animate-move +2025-09-30 -支持将模板视频中角色的动作和表情，迁移至单张静态人物图片上，生成人物动作视频。[万相-图生动作](https://help.aliyun.com/zh/model-studio/wan-animate-move-api) +`qwen3-vl-30b-a3b-instruct` -图生视频 +Qwen3-VL系列第二大MoE模型的Instruct版本，响应速度快，支持长视频长文档等超长上下文；全面升级图像/视频理解、空间感知与万物识别能力；具备视觉2D/3D定位能力，胜任复杂现实任务。 -2025-09-19 +文本生成、深度思考 -中国内地 +2025-09-30 -wan2.2-animate-mix +`deepseek-v3.2-exp` -能够依据人物图片和参考视频，将视频中的主角替换为图片中的角色，同时保留原视频的场景、光照和色调，实现无缝人物替换。[万相-视频换人](https://help.aliyun.com/zh/model-studio/wan-animate-mix-api) +引入了DeepSeek Sparse Attention（一种稀疏注意力机制）的实验性质版本，针对长文本的训练和推理效率进行了探索性的优化和验证。 -视觉理解 +语音识别 -2025-09-12 +2025-09-25 -中国内地 +`fun-asr-mtl` -qwen-vl-ocr-2025-08-18 +`fun-asr-mtl-2025-08-25` -文字定位能力全面升级，通用文字识别、信息抽取能力均有提升。[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#e91318ce09pxi)。 +百聆多语言语音识别大模型，支持超过31种语言，支持语种自由切换，出海用户首推，尤其东南亚出海。fun-asr为该模型的升级版本，建议切换使用fun-asr。 -首尾帧生视频 +图片生成 -2025-09-12 +2025-09-23 -中国内地 +`wan2.5-i2i-preview` -wan2.2-kf2v-flash +万相2.5-图像编辑-Preview，全新升级模型架构。支持指令控制实现丰富的图像编辑能力，指令遵循能力进一步提升，支持高一致性保持的多图参考生成，文字生成表现优异。 -指令理解与运镜控制更准，稳定性全面提升。[图生视频-基于首尾帧](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-by-first-and-last-frame-api-reference) +多模态向量 -推理模型 +2025-09-23 -2025-09-11 +`tongyi-embedding-vision-plus` -中国内地 +Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。 -qwen-plus-2025-09-11 +多模态向量 -属于 Qwen3 系列模型，相较于qwen-plus-2025-07-28，在思考模式下提升了指令遵循能力、总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。在非思考模式下中文理解与逻辑推理能力得到增强，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +2025-09-23 -推理模型 +`tongyi-embedding-vision-flash` -2025-09-11 +Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，在视觉向量化上具备极高性价比。 -中国内地 +深度思考、视觉理解 -qwen3-next-80b-a3b-thinking、qwen3-next-80b-a3b-instruct +2025-09-23 -基于Qwen3的新一代开源模型，thinking模型相较于qwen3-235b-a22b-thinking-2507提升了指令遵循能力，总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型相较于qwen3-235b-a22b-instruct-2507增强了中文理解、逻辑推理及文本生成能力，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +`qwen3-vl-plus-2025-09-23` -语音合成 +Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。此版本为2025年9月23日快照版本。 -2025-09-08 +深度思考、视觉理解 -中国内地 +2025-09-23 -cosyvoice-v3-plus、cosyvoice-v3 +`qwen3-vl-235b-a22b-thinking` -语音合成CosyVoice新增cosyvoice-v3-plus和cosyvoice-v3模型，较之前版本在自然度、音质、韵律、情感表现力上有更好的表现。使用方法请参见[实时语音合成-CosyVoice /Sambert](https://help.aliyun.com/zh/model-studio/text-to-speech)。 +Qwen3系列视觉理解模型，多模态思考能力显著增强，模型在STEM与数学推理方面进行了重点优化；视觉感知与识别能力全面提升、OCR能力迎来重大升级。 -语音识别 +视觉理解 -2025-09-08 +2025-09-23 -中国内地 +`qwen3-vl-235b-a22b-instruct` -qwen3-asr-flash（稳定版，等同qwen3-asr-flash-2025-09-08） +Qwen3系列视觉理解模型，在视觉coding、空间感知等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解，OCR能力迎来重大升级。 -qwen3-asr-flash-2025-09-08（快照版） +文本生成、深度思考 -Qwen3-ASR模型基于Qwen3-Omni模型基座训练而成，支持多语言识别、歌唱识别、噪声拒识等功能。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +2025-09-23 -文生文 +`qwen3-max` -2025-09-05 +`qwen3-max-2025-09-23` -中国内地 +千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。 -qwen3-max-preview +实时语音翻译 -基于Qwen3的Qwen-Max模型（预览版），相较Qwen 2.5系列整体通用能力有大幅度提升，中英文通用文本理解能力、复杂指令遵循能力、主观开放任务能力、多语言能力、工具调用能力均显著增强；模型知识幻觉更少。千问 Max +2025-09-23 -视频生成 +`qwen3-livetranslate-flash-realtime` -2025-08-26 +`qwen3-livetranslate-flash-realtime-2025-09-22` -中国内地 +Qwen3-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。 -wan2.2-s2v-detect、wan2.2-s2v +文本生成 -万相数字人模型支持基于单张图片和音频，生成动作自然的说话、唱歌或表演视频，不限制形象画幅，支持肖像、全身或半身的人物图像。[万相-数字人](https://help.aliyun.com/zh/model-studio/wan-s2v-overview/) +2025-09-23 -图像翻译 +`qwen3-coder-plus-2025-09-23` -2025-08-22 +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。本版本为2025年9月23日快照，相较上一版本（7月22日快照）在下游任务效果和工具调用方面鲁棒性有所提升；代码安全性增强。 -中国内地 +视觉理解 -qwen-mt-image +2025-09-23 -千问图像翻译模型支持将11种语言图片的文字翻译成中文或英文，能精准保留原始排版与内容信息，并提供术语定义、敏感词过滤、图像主体检测等功能。[千问-图像翻译](https://help.aliyun.com/zh/model-studio/qwen-mt-image-api) +`qwen-vl-ocr-latest` -推理模型 +千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。 -2025-08-22 +图片生成 -中国内地 +2025-09-23 -deepseek-v3.1 +`qwen-image-plus` -混合推理架构模型，同时支持思考模式与非思考模式。详情参见[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api)。 +千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。 -深入研究 +实时语音识别 -2025-08-22 +2025-09-23 -中国内地 +`fun-asr-realtime` -qwen-deep-research +`fun-asr-realtime-2025-09-15` -千问深入研究模型，它可以拆解复杂问题，结合互联网搜索进行推理分析并生成研究报告。详情请参见[深入研究（Qwen-Deep-Research）](https://help.aliyun.com/zh/model-studio/qwen-deep-research) +通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。 -语音识别 +图片生成 -2025-08-22 +2025-09-22 -中国内地 +`qwen-image-edit` -fun-asr（稳定版，等同fun-asr-2025-08-25） +千问系列首个图像编辑模型，成功将Qwen-Image的文本渲染能力拓展到编辑任务上。支持精准的中英双语文字编辑、视觉外观与语义双重编辑、具备强大的跨基准性能表现。 -fun-asr-2025-08-25（快照版） +视频生成 -Fun-ASR 是通义实验室推出的端到端语音识别大模型。它基于先进的自研语音技术，具备卓越的上下文感知和高精度转写能力，支持中英文录音文件识别。[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition) +2025-09-19 -图像编辑 +`wan2.5-t2v-preview` -2025-08-19 +万相2.5-文生视频-Preview，全新升级模型架构，支持与画面同步的声音生成，支持10秒长视频生成，更强的指令遵循能力，运动能力、画面质感进一步提升。 -中国内地 +图片生成 -qwen-image-edit +2025-09-19 -千问图像编辑模型支持精准的中英双语文字编辑、调色、细节增强、风格迁移、增删物体、改变位置和动作等操作，可实现复杂的图文编辑。[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide) +`wan2.5-t2i-preview` -视觉理解 +万相2.5-文生图-Preview，全新升级模型架构。画面美学、设计感、真实质感显著提升，精准指令遵循，擅长中英文和小语种文字生成，支持复杂结构化长文本和图表、架构图等内容生成。 -2025-08-15 +视频生成 -中国内地 +2025-09-19 -qwen-vl-plus-2025-08-15 +`wan2.5-i2v-preview` -视觉理解模型。在物体识别与定位、多语言处理的能力上有显著提升。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +万相2.5-图生视频-Preview，全新升级技术架构，支持与画面同步的声音生成，支持10秒长视频生成，更强的指令遵循能力，运动能力、画面质感进一步提升。 -视觉理解 +视频生成 -2025-08-13 +2025-09-19 -中国内地 +`wan2.2-animate-move` -qwen-vl-max-2025-08-13 +wan2.2-animate-move图生动作是一款角色动画生成模型，用户只需上传一张角色照片和一段参考表演视频，即可将视频中的表情和动作迁移到图片角色上，生成高保真的动画视频。 -视觉理解模型。视觉理解指标全面提升，数学、推理、物体识别、多语言处理能力显著增强。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +视频生成 -文生图 +2025-09-19 -2025-08-13 +`wan2.2-animate-mix` -中国内地 +wan2.2-animate-mix视频换人是一款角色替换的模型产品，上传一张角色照片与一段表演视频，即可将原视频中的角色精准替换为照片中的角色，完整保留原始视频的场景、光照和色调等环境细节。 -qwen-image +实时全模态 -千问文生图模型在复杂文本渲染方面表现突出，特别是中英文文本渲染，可实现复杂的图文混合布局**。**[文生图（Qwen-Image)](https://help.aliyun.com/zh/model-studio/qwen-image-api) +2025-09-17 -图生视频 +`qwen3-omni-flash-realtime-2025-09-15` -2025-08-11 +千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。此版本为2025年9月15日的快照版本。 -中国内地 +全模态 -wan2.2-i2v-flash +2025-09-17 -万相2.2极速版模型。相较 2.1 模型，新版本在画面细节表现和运动稳定性方面均有显著提升，生成速度提升达 50%。[首帧生视频](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +`qwen3-omni-flash-2025-09-15` -视觉理解 +Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。此版本为2025年9月15日的快照版本。 -2025-08-06 +语音识别 -中国内地 +2025-09-17 -qwen-vl-plus-2025-07-10 +`qwen3-omni-30b-a3b-captioner` -视觉理解模型。相较于上一版模型，进一步提升监控视频内容的理解能力。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。 -代码模型 +语音合成 -2025-08-05 +2025-09-16 -中国内地 +`qwen3-tts-flash-realtime-2025-09-18` -qwen3-coder-flash、qwen3-coder-flash-2025-07-28 +Qwen3-TTS-Flash-Realtime-2025-09-18模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年9月18日快照版本模型。 -千问Coder系列速度最快、成本最低的模型。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +语音合成 -推理模型 +2025-09-16 -2025-08-05 +`qwen3-tts-flash-2025-09-18` -中国内地 +千问3-TTS-Flash-2025-09-18模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年9月18日快照版本模型。 -qwen-flash、qwen-flash-2025-07-28 +视觉理解 -千问系列速度最快、成本极低的模型，适合简单任务。模型列表 +2025-09-13 -推理模型 +`qwen-vl-ocr-2025-08-28` -2025-07-30 +本模型为2025年8月28日的快照版本，在文字定位能力上实现全面升级，通用文字识别及卡证票据信息抽取能力均有提升。 -中国内地 +视频生成 -qwen-plus-2025-07-28 +2025-09-12 -属于 Qwen3 系列模型，相较于上一版模型，将上下文长度提高到了1,000,000。思考模式请参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)，非思考模式请参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +`wan2.2-kf2v-flash` -推理模型 +全新升级的万相2.2-首尾帧生视频，生成速度更快。优化视频动态稳定性与成功率，更强大的指令遵循能力，两张图片生成丝滑过度视频。 -2025-07-30 +文本生成、深度思考 -中国内地 +2025-09-11 -qwen3-30b-a3b-thinking-2507 +`qwen3-next-80b-a3b-thinking` -qwen3-30b-a3b-instruct-2507 +基于Qwen3的新一代思考模式开源模型，相较上一版本（千问3-235B-A22B-Thinking-2507指令遵循能力有提升、模型总结回复更加精简。 -是qwen3-30b-a3b的升级版。thinking模型逻辑能力、通用能力、知识增强及创作能力提升，参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型创作能力与模型安全性提升，参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +文本生成 -图生视频 +2025-09-11 -2025-07-28 +`qwen3-next-80b-a3b-instruct` -中国内地 +基于Qwen3的新一代非思考模式开源模型，相较上一版本（千问3-235B-A22B-Instruct-2507）中文文本理解能力更佳、逻辑推理能力有增强、文本生成类任务表现更好。 -wan2.2-i2v-plus +文本生成、深度思考 -相较 2.1 模型，新版本在画面细节表现和运动稳定性方面均有显著提升，生成速度提升达 50%。[首帧生视频](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +2025-09-11 -文生视频 +`qwen-plus-2025-09-11` -2025-07-28 +本版本为2025年9月11日快照，相较7月28日快照在思考模式下指令遵循能力有提升、模型总结回复更加精简；在非思考模式下中文文本理解能力更佳、逻辑推理能力有增强。支持1M上下文长度，按照上下文长度进行阶梯计费。 -中国内地 +语音识别 -wan2.2-t2v-plus +2025-09-08 -相较 2.1 模型，新版本在画面细节表现和运动稳定性方面均有显著提升，生成速度提升达 50%。[文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +`qwen3-asr-flash` -文生图 +`qwen3-asr-flash-2025-09-08` -2025-07-28 +千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。 -中国内地 +文本生成、深度思考 -wan2.2-t2i-flash、wan2.2-t2i-plus +2025-09-05 -相较 2.1 模型，新版本在创意性、稳定性、写实质感上全面升级，生成速度提升达 50%。[文生图V2版](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference) +`qwen3-max-preview` -数据挖掘 +Qwen3系列Max模型Preview版本，实现思考模式和非思考模式的有效融合。思考模式下在智能体编程能力、常识知识推理能力、数学/科学/通用类推理等能力上均有显著增强。 -2025-07-26 +语音合成 -中国内地 +2025-09-03 -qwen-doc-turbo +`cosyvoice-v3-plus` -千问数据挖掘模型，可以提取文档中的结构化信息并用于数据标注和内容审核等领域。详情请参见[数据挖掘（Qwen-Doc）](https://help.aliyun.com/zh/model-studio/data-mining-qwen-doc)。 +克隆能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音克隆大模型，具有更好的音质和复刻相似度，适用于更专业的场景。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。合成能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音合成大模型，具有更好的音质和表现力，适用于更专业的场景。该模型支持文本至语音的实时流式合成。 -推理模型 +视频生成 -2025-07-24 +2025-08-25 -中国内地 +`wan2.2-s2v-detect` -qwen3-235b-a22b-thinking-2507、qwen3-235b-a22b-instruct-2507 +wan2.2-s2v-detect 是 wan2.2-s2v 的辅助模型，用于确认输入的人物肖像图片是否符合 wan2.2-s2v 模型所需的人物肖像图片规范。wan2.2-s2v 模型基于 wan2.2-s2v-detect 检测通过的图片和人声音频文件进行视频生成。 -是qwen3-235b-a22b的升级版。thinking模型逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景，参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型创作能力与模型安全性均有提升，参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +视频生成 -代码模型 +2025-08-25 -2025-07-23 +`wan2.2-s2v` -中国内地 +wan2.2-s2v 是一款视频生成模型，可基于人物图片和人声音频文件，生成高质量的人物说话/唱歌/表演动态视频。 -qwen3-coder、qwen3-coder-plus-2025-07-22 +文本生成、深度思考 -基于 Qwen3 的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，代码能力卓越的同时兼具通用能力。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +2025-08-23 -语音合成 +`deepseek-v3.1` -2025-06-26 +DeepSeek-V3.1为混合推理架构模型，同时支持思考模式与非思考模式，具备更高的推理效率和更强的Agent能力。 -中国内地 +图片生成 -qwen-tts-2025-05-22 +2025-08-22 -qwen-tts 模型的2025年5月22日快照版本。新增北京话、吴语和四川话三种音色。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +`qwen-mt-image` -推理模型 +专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。 -2025-06-24 +文本生成 -中国内地 +2025-08-22 -qwen-plus、qwen-turbo +`qwen-deep-research` -qwen-plus 已更新，支持[思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)，与 qwen-plus-2025-04-28 能力相同；qwen-turbo 已更新，支持[思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)，与 qwen-turbo-2025-04-28 能力相同。 +千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。 -语音合成 +语音识别 -2025-06-23 +2025-08-22 -中国内地 +`fun-asr-2025-08-25` -cosyvoice-v2 +百聆新一代语音识别大模型，主打中文、英文、日文语音识别，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境，国内用户首推。此版本为2025年8月25日的快照版本。 -cosyvoice-v2 全新升级，新增多种音色，不仅丰富了语言支持范围（如粤语、韩语、日语），还引入了更多风格化音色（如英式英语、美式英语）。详情请参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)。 +图片生成 -文本向量 +2025-08-13 -2025-06-04 +`qwen-image` -中国内地 +千问系列首个图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA。 -text-embedding-v4 +视频生成 -text-embedding-v4为text-embedding-v3的升级版模型，属于[Qwen3-Embedding](https://qwenlm.github.io/zh/blog/qwen3-embedding/)系列。相较于上一版模型，新版模型涵盖了更多自然语言以及多种编程语言，并新增2048及1536向量维度的选择。[文本与多模态向量化](https://help.aliyun.com/zh/model-studio/embedding) +2025-08-11 -推理模型 +`wan2.2-i2v-flash` -2025-06-04 +全新升级的万相2.2图生视频，生成速度更快。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。 -中国内地 +文本生成 -deepseek-r1-0528 +2025-08-06 -deepseek-r1-0528为deepseek-r1的升级版模型，相较于上一版模型，新版模型在复杂推理任务中的表现有了显著提升。在数学、编程与通用逻辑等多个基准测评中取得了优异成绩。[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api) +`glm-4.5-air` -视觉推理 +GLM-4.5-Air采用混合专家（MoE）架构，总参数量为1060亿，激活参数120亿，相较GLM-4.5更紧凑、轻量，适用于对模型规模和资源消耗有一定限制的场景。 -2025-06-03 +文本生成 -中国内地 +2025-08-06 -qvq-max-2025-05-15 +`glm-4.5` -视觉推理模型。较上一个版本，增强了在数学、编程、视觉分析、创作以及通用任务方面的能力。[视觉推理](https://help.aliyun.com/zh/model-studio/visual-reasoning) +GLM-4.5采用混合专家（MoE）架构，总参数量为3550 亿，激活参数320亿，在复杂推理、代码生成及智能体交互等通用能力上实现了能力融合与技术突破。 -视觉推理 +文本生成 -2025-06-03 +2025-08-05 -中国内地 +`qwen3-coder-flash` -qvq-plus、qvq-plus-latest、qvq-plus-2025-05-15 +`qwen3-coder-flash-2025-07-28` -视觉推理模型。支持视觉输入及思维链输出，继qvq-max模型后推出的plus版本，相较于qvq-max模型，qvq-plus系列模型推理速度更快，效果和成本更均衡。[视觉推理](https://help.aliyun.com/zh/model-studio/visual-reasoning) +基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。 -语音合成 +文本生成、深度思考 -2025-06-03 +2025-08-05 -中国内地 +`qwen-flash` -cosyvoice-v2 +`qwen-flash-2025-07-28` -SSML（Speech Synthesis Markup Language，语音合成标记语言）不仅能指定语音合成读什么内容，还能精细控制其朗读方式，包括断句分词、发音、语速、停顿、语调、音量等语音特征，甚至支持添加背景音乐，从而实现更加自然、富有表现力的语音输出。详细使用方法请参见[SSML标记语言介绍](https://help.aliyun.com/zh/model-studio/introduction-to-cosyvoice-ssml-markup-language)。 +Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。 -视频编辑 +文本生成 -2025-05-19 +2025-07-31 -中国内地 +`qwen3-coder-30b-a3b-instruct` -wanx2.1-vace-plus +基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。 -通用视频编辑模型。模型具备多模态输入能力，融合图片、视频与文本提示词，可执行图生视频（参考图像主体或背景生成视频）、视频重绘（从输入视频中提取运动特征生成视频）等多种任务。[通用视频编辑](https://help.aliyun.com/zh/model-studio/legacy-wanx-vace-api-reference) +文本生成、深度思考 -图像生成 +2025-07-31 -2025-05-15 +`qwen-plus-latest` -中国内地 +千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。本模型是动态更新版本，模型更新不会提前通知。 -aitryon-plus +文本生成、深度思考 -AI试衣模型。相较于基础版aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。[AI试衣-Plus版](https://help.aliyun.com/zh/model-studio/aitryon-plus-api) +2025-07-30 -视觉理解 +`qwen3-30b-a3b-thinking-2507` -2025-05-15 +基于Qwen3的思考模式开源模型，相较上一版本（千问3-30B-A3B）复杂推理类任务性能优秀，包括逻辑推理、数学、科学、代码类等具有一定难度的任务场景，指令遵循、文本理解、多语言翻译等能力显著提高。 -中国内地 +文本生成、深度思考 -qwen-vl-plus-2025-05-07 +2025-07-30 -视觉理解模型。模型在数学、推理、监控视频内容的理解方面的能力有显著提升。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +`qwen-plus-2025-07-28` -文字提取 +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。相较上个版本在中英文能力、工具调用上进行了专用增强。本版本为2025年7月28日快照，首次支持1M上下文长度，按照上下文长度进行阶梯计费。 -2025-04-30 +文本生成 -中国内地 +2025-07-29 -qwen-vl-ocr-2025-04-13 +`qwen3-30b-a3b-instruct-2507` -基于Qwen-VL训练的OCR专有大模型，提供强大的图文识别能力；该模型新增六种内置的OCR任务，分别是：通用识别、文档解析、表格解析、信息抽取、多语言识别、公式识别。[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr) +基于Qwen3的非思考模式开源模型，相较上一版本（千问3-30B-A3B）中英文和多语言整体通用能力有大幅提升。主观开放类任务专项优化，显著更加符合用户偏好，能够提供更有帮助性的回复。 -推理模型 +视频生成 -2025-04-28 +2025-07-28 -中国内地 +`wan2.2-t2v-plus` -**Qwen3商业版模型** +全新升级的万相2.2文生视频，视频品质更高。可稳定生成大幅度复杂运动，支持影视级画面表现与控制，更强大的指令遵循能力，实现物理世界还原。 -qwen-plus-2025-04-28、qwen-turbo-2025-04-28 +图片生成 -**Qwen3开源版模型** - -qwen3-235b-a22b、qwen3-30b-a3b、qwen3-32b、qwen3-14b、qwen3-8b、qwen3-4b、qwen3-1.7b、qwen3-0.6b +2025-07-28 -Qwen3 模型支持思考模式和非思考模式，您可以通过 `enable_thinking` 参数实现两种模式的切换。除此之外，Qwen3 模型的能力得到了大幅提升： +`wan2.2-t2i-plus` -1. **推理能力**：在数学、代码和逻辑推理等评测中，显著超过 QwQ 和同尺寸的非推理模型，达到同规模业界顶尖水平。 - -2. **人类偏好能力**：创意写作、角色扮演、多轮对话、指令遵循能力均大幅提升，通用能力显著超过同尺寸模型。 - -3. **Agent 能力**：在推理、非推理两种模式下都达到业界领先水平，能够精准地调用外部工具。 - -4. **多语言能力**：支持100多种语言和方言，多语言翻译、指令理解、常识推理能力都明显提升。 - -5. **回复格式问题修复**：修复了之前版本存在的回复格式的问题，如异常 Markdown、中间截断、错误输出 boxed 等问题。 - +全新升级的万相2.2文生图，更丰富的画面细节。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。 -思考模式请参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)，非思考模式请参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +图片生成 -声音复刻 +2025-07-28 -2025-04-24 +`wan2.2-t2i-flash` -中国内地 +全新升级的万相2.2文生图，更快的生成速度。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。 -cosyvoice-v2 +视频生成 -用户仅需提供10~20秒的音频，即可迅速生成高度相似且听感自然的定制声音。[声音复刻/设计](https://help.aliyun.com/zh/model-studio/voice-replica-1/) +2025-07-28 -视觉理解 +`wan2.2-i2v-plus` -2025-04-18 +全新升级的万相2.2图生视频，视频品质更高。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。 -中国内地 +文本生成、深度思考 -qwen-vl-max-2025-04-08 +2025-07-25 -视觉理解模型。数学和推理能力有所提升，回复风格面向人类偏好进行调整，模型回复详实程度和格式清晰度明显改善。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +`qwen3-235b-a22b-thinking-2507` -图生视频 +基于Qwen3的思考模式开源模型，相较上一版本（千问3-235B-A22B）逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景。 -2025-04-18 +文本生成 -中国内地 +2025-07-23 -wanx2.1-kf2v-plus +`qwen-doc-turbo` -基于输入的首帧和尾帧图片，模型能够根据提示词生成一段丝滑流畅的动态视频。[首尾帧生视频](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-by-first-and-last-frame-api-reference) +快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。 -视觉理解 +文本生成 -2025-04-04 +2025-07-22 -中国内地 +`qwen3-coder-plus` -qwen-vl-max-2025-04-02 +`qwen3-coder-plus-2025-07-22` -视觉理解模型。在解决复杂数学问题方面，准确性显著提高，回复风格面向人类偏好进行大幅调整，尤其是数学、逻辑推理、知识问答等客观类问题，模型回复详实程度和格式清晰度明显改善。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。 -视觉推理 +文本生成 -2025-03-28 +2025-07-22 -中国内地 +`qwen3-coder-480b-a35b-instruct` -qvq-max、qvq-max-latest、qvq-max-2025-03-25 +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。 -视觉推理模型。支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现出更强的能力。[视觉推理](https://help.aliyun.com/zh/model-studio/visual-reasoning) +文本生成 -全模态 +2025-07-22 -2025-03-26 +`qwen3-235b-a22b-instruct-2507` -中国内地 +基于Qwen3的非思考模式开源模型，相较上一版本（千问3-235B-A22B）主观创作能力与模型安全性均有小幅度提升。 -qwen-omni-turbo-2025-03-26 +文本生成 -千问全新多模态理解生成大模型，支持文本、图像、语音与视频输入，并输出文本与音频，提供了4种自然对话音色。使用方法请参见[非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni)。 +2025-07-22 -全模态 +`qwen-mt-turbo` -2025-03-26 +基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 -中国内地 +文本生成 -qwen2.5-omni-7b +2025-07-22 -千问全新多模态理解生成大模型，支持文本、图像、语音与视频输入，并输出文本与音频，提供了2种自然对话音色。使用方法请参见[非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni)。 +`qwen-mt-plus` -图像编辑 +基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 -2025-03-25 +实时语音合成 -中国内地 +2025-07-16 -wanx2.1-imageedit +`qwen-tts-realtime` -通用图像编辑模型。通过一句话指令实现多样化的图像编辑功能，如扩图、去文字水印、图像修复、图像风格迁移等。[万相-通用图像编辑2.1](https://help.aliyun.com/zh/model-studio/wanx-image-edit-api-reference) +`qwen-tts-realtime-latest` -视觉理解 +`qwen-tts-realtime-2025-07-15` -2025-03-24 +Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成模型。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。 -中国内地 +文本生成、深度思考 -qwen2.5-vl-32b-instruct  +2025-07-16 -视觉理解模型。在数学问题的解答方面达到了接近Qwen2.5VL-72B的水平，回复风格面向人类偏好进行大幅调整，尤其是数学、逻辑推理、知识问答等客观类问题，模型回复详实程度和格式清晰度明显改善。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +`qwen-plus-2025-07-14` -推理模型 +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。此版本为2025年7月14日快照模型，相较上个版本在非thinking模式下中英文能力均有大幅提升，工具调用能力专项增强。 -2025-03-06 +文本生成 -中国内地 +2025-07-16 -qwq-plus、qwq-plus-latest、qwq-plus-2025-03-05 +`Moonshot-Kimi-K2-Instruct` -基于 Qwen2.5 模型训练的 QwQ 推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、LiveCodeBench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking) +Kimi-K2是月之暗面提供的国内首个开源万亿参数MoE模型，激活参数达 320 亿，具有卓越的编码和工具调用能力。 -开源推理模型 +语音合成 -2025-03-06 +2025-06-26 -中国内地 +`qwen-tts-latest` -qwq-32b +`qwen-tts-2025-05-22` -基于 Qwen2.5-32B 模型训练的 QwQ 推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、LiveCodeBench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平，各指标均显著超过同样基于 Qwen2.5-32B 的 DeepSeek-R1-Distill-Qwen-32B。[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking) +模型是动态更新版本，等同于最新版本快照模型，模型更新时不会提前通知。 -语音识别/翻译 +文本生成、深度思考 -2025-03-03 +2025-06-24 -中国内地 +`qwen-turbo` -gummy-realtime-v1 +Qwen3系列Turbo模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-Turbo，达到同规模业界SOTA水平。 -gummy-chat-v1 +文本生成、深度思考 -Gummy大模型支持实时语音识别与翻译，能够精准识别中、英、日、韩等10种语言。此外，它还支持中、英、日、韩之间的互译，以及其他6种语言单向翻译成中文或英文。使用方法请参见[实时语音识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition)。 +2025-06-24 -图生视频 +`qwen-plus` -2025-02-25 +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-Plus，达到同规模业界SOTA水平。 -中国内地 +深度思考 -wanx2.1-i2v-turbo +2025-06-18 -相较于wanx2.1-i2v-plus模型，turbo模型生成速度更快，耗时仅为plus模型的三分之一，性价比更高。使用方法请参见[首帧生视频](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/)。 +`deepseek-r1` -全模态 +DeepSeek-R1 在后训练阶段大规模使用了强化学习技术，在仅有极少标注数据的情况下，极大提升了模型推理能力。在数学、代码、自然语言推理等任务上，性能较高，能力较强。 -2025-02-14 +视觉理解 -中国内地 +2025-06-13 -qwen-omni-turbo +`qwen-vl-plus` -qwen-omni-turbo-latest +千问VL-Plus（qwen-vl-plus），即千问大规模视觉语言模型增强版。大幅提升细节识别能力和文字识别能力，支持超百万像素分辨率和任意长宽比规格的图像。在广泛的视觉任务上提供卓越的性能。 -qwen-omni-turbo-2025-01-19 +文本向量 -Qwen-Omni 系列模型支持输入多种模态的数据，包括视频、音频、图片、文本，并输出文本。使用方法请参见[非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni)。 +2025-06-05 -视觉理解 +`text-embedding-v4` -2025-02-07 +通用文本向量V4版本，是通义实验室基于Qwen3训练的多语言文本统一向量模型，相较V3版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升15%~40%；支持64~2048维用户自定义向量维度。 -中国内地 +深度思考 -qwen-vl-max-2025-01-25 +2025-06-04 -视觉理解模型。属于[Qwen2.5-VL](https://qwenlm.github.io/blog/qwen2.5-vl/)系列模型，相较于上一版模型，扩展上下文至128k，显著增强图像和视频的理解能力。 +`deepseek-r1-0528` -视觉理解 +0528为R1模型的小版本升级，相较于旧版 R1，新版在复杂推理任务中的表现有了显著提升。在数学、编程与通用逻辑等多个基准测评中取得了优异成绩。 -2025-02-07 +深度思考、视觉理解 -中国内地 +2025-06-03 -qwen-vl-plus-2025-01-25 +`qvq-plus` -视觉理解模型。属于[Qwen2.5-VL](https://qwenlm.github.io/blog/qwen2.5-vl/)系列模型，相较于上一版模型，扩展上下文至128k，显著增强图像和视频的理解能力。 +千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力 -文生文 +语音合成 -2025-01-27 +2025-05-27 -中国内地 +`cosyvoice-v2` -deepseek-v3 +cosyvoice-V2是通义实验室依托大规模预训练语言模型，在深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。 -deepseek-r1 +视觉理解 -DeepSeek系列模型是由深度求索（DeepSeek）公司推出的大语言模型。 +2025-05-26 -- DeepSeek-V3 为 MoE 模型，671B 参数，激活 37B，在 14.8T Token 上进行了预训练，在长文本、代码、数学、百科、中文能力上表现优秀。 - -- DeepSeek-R1 在后训练阶段大规模使用了强化学习技术，在仅有极少标注数据的情况下，极大提升了模型推理能力，尤其在数学、代码、自然语言推理等任务上。 - -- 具体请参见[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api)。 - +`qwen-vl-max` -视觉理解 +千问VL-Max（qwen-vl-max），即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。 -2025-01-27 +视频生成 -中国内地 +2025-05-14 -qwen2.5-vl-3b-instruct +`wanx2.1-vace-plus` -qwen2.5-vl-7b-instruct +万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。 -qwen2.5-vl-72b-instruct +实时全模态 -- 相对于Qwen2-VL大模型有如下改进： - - - 在指令跟随、数学计算、代码生成、结构化输出（JSON输出）等方面的能力有显著提升。 - - - 支持对图像中的文字、图表、布局等视觉内容进行统一解析，并增加了精准定位视觉元素的能力，支持检测框和坐标点的表示方式。 - - - 支持对长视频文件（最长10分钟）进行理解，具备秒级别的事件时刻定位能力，能理解时间先后和快慢。 - -- 使用方法请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +2025-05-08 -文生文 +`qwen-omni-turbo-realtime` -2025-01-27 +`qwen-omni-turbo-realtime-latest` -中国内地 +`qwen-omni-turbo-realtime-2025-05-08` -qwen-max-2025-01-25 +千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。 -qwen2.5-14b-instruct-1m +文本生成、深度思考 -qwen2.5-7b-instruct-1m +2025-04-29 -- qwen-max-2025-01-25模型（又称为[Qwen2.5-Max](https://qwenlm.github.io/zh/blog/qwen2.5-max/)）：千问系列效果最好的模型，代码编写与理解能力、逻辑能力、多语言能力显著提升，回复风格面向人类偏好进行大幅调整，模型回复详实程度和格式清晰度明显改善，内容创作、JSON格式遵循、角色扮演能力定向提升。使用方法请参见：[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 - -- qwen2.5-14b-instruct-1m、qwen2.5-7b-instruct-1m模型：相比于qwen2.5-14b-instruct与qwen2.5-7b-instruct模型，将上下文长度提高到了1,000,000。使用方法请参见：[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 - +`qwen3-8b` -图生视频 +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-7B。 -2025-01-22 +文本生成、深度思考 -中国内地 +2025-04-29 -emoji-detect-v1 +`qwen3-32b` -emoji-v1 +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-32B-Instruct，达到同规模业界SOTA水平。 -- 基于人脸图片和预设的人脸动态模板，生成人脸动态视频。该模型可用于表情包制作、视频素材生成等场景。使用方法请参见[表情包Emoji快速开始](https://help.aliyun.com/zh/model-studio/emoji-quick-start/)。 - +文本生成、深度思考 -文生文 +2025-04-29 -2025-01-17 +`qwen3-30b-a3b` -中国内地 +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-14B，达到同规模业界SOTA水平。 -qwen-plus-2025-01-12 +文本生成、深度思考 -- 相对于qwen-plus-2024-12-20模型，中英文整体能力有提升，中英常识、阅读理解能力提升较为显著，在不同语言、方言、风格之间自然切换的能力有显著改善，中文指令遵循能力显著提升。使用方法请参见[qwen-plus-2025-01-12](https://bailian.console.aliyun.com/#/model-market/detail/qwen-plus-0112)。 - +2025-04-29 -图生视频 +`qwen3-235b-a22b` -2025-01-17 +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-72B-Instruct，达到同规模业界SOTA水平。 -中国内地 +文本生成、深度思考 -wanx2.1-i2v-plus +2025-04-29 -- 输入图片作为视频首帧，再根据提示词生成视频。使用方法请参见[首帧生视频](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/)。 - +`qwen-plus-2025-04-28` -文生图 +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-Plus，达到同规模业界SOTA水平。此版本为2025年4月28日快照模型。 -2025-01-17 +文本生成、深度思考 -中国内地 +2025-04-28 -wanx2.0-t2i-turbo +`qwen3-14b` -- 擅长质感人像与创意设计，速度中等，性价比高。使用方法请参见[文生图V2系列模型](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference)。 - +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-14B。 -视觉理解 +图片生成 -2025-01-13 +2025-04-28 -中国内地 +`aitryon-plus` -qwen-vl-plus-2025-01-02 +aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。 -- 相较于qwen-vl-plus-0809模型，大幅提升指令跟随、图像理解和数学能力。使用方法请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +视觉理解 -文生视频 +2025-04-23 -2025-01-08 +`qwen-vl-ocr-2025-04-13` -中国内地 +千问VL-OCR（qwen-vl-ocr-2025-04-13），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。模型内置六种识别任务分别是：通用识别、文档解析、表格解析、信息抽取、多语言识别、公式识别。本模型为2025年04月13日的快照版本。 -wanx2.1-t2v-turbo +视频生成 -wanx2.1-t2v-plus +2025-04-21 -- 一句话生成视频。 - -- 具备强大的指令遵循能力，支持大幅度复杂运动、现实物理规律还原，生成的视频呈现丰富的艺术风格及影视级画面质感。使用方法请参见[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference)。 - +`wanx2.1-kf2v-plus` -文生图 +万相2.1-首尾帧-Plus，两张图片生成丝滑过度视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成画面细节更丰富。 -2025-01-08 +语音合成 -中国内地 +2025-04-20 -wanx2.1-t2i-turbo +`qwen-tts` -wanx2.1-t2i-plus +`qwen-tts-2025-04-10` -- [文生图V2系列模型](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference)为全面升级的万相文生图模型，推荐体验。 - +千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持流式输出。 -视觉理解 +全模态 -2025-01-07 +2025-03-27 -中国内地 +`qwen-omni-turbo-2025-03-26` -qwen-vl-max-2024-12-30 +千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色，此版本为2025年3月26日的快照版本，相比1月19日的快照版本在视觉能力上有大幅提升。 -- 该模型丰富了知识库，图像识别和理解能力进一步提升，能够解析复杂的视觉内容。使用方法请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +全模态 -语音识别 +2025-03-26 -2025-01-02 +`qwen2.5-omni-7b` -中国内地 +基于Qwen2.5训练的全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 -qwen-audio-asr +深度思考、视觉理解 -qwen-audio-asr-latest +2025-03-26 -qwen-audio-asr-2024-12-04 +`qvq-max` -- 千问ASR是基于Qwen-Audio训练，专用于语音识别的模型，目前支持的语言有：中文、英文。使用方法请参见[实时语音识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition)。 - +千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。 -视觉推理 +图片生成 -2024-12-25 +2025-03-25 -中国内地 +`wanx2.1-imageedit` -qvq-72b-preview +万相-通义图像编辑，支持预设编辑任务与指令式编辑，包含多种局部/全图编辑能力，如图像风格化、线稿生图、局部重绘、参考图生成、图像外扩、图像超分等。 -- 专注于提升视觉推理能力，尤其在数学推理领域。使用方法请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +语音合成 -多语言翻译 +2025-03-20 -2024-12-25 +`sambert-zhiyuan-v1` -中国内地 +提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。 -qwen-mt-plus +文本生成 -qwen-mt-turbo +2025-03-20 -- Qwen-MT模型是基于千问模型优化的机器翻译大语言模型，擅长中英互译、中文与小语种互译、英文与小语种互译，小语种包括日、韩、法、西、德、葡（巴西）、泰、印尼、越、阿等26种。在多语言互译的基础上，提供术语干预、领域提示、记忆库等能力，提升模型在复杂应用场景下的翻译效果。详情请参见[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation)。 - +`qwen-plus-character` -视觉理解 +千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 -2024-12-18 +文本向量 -中国内地 +2025-03-20 -qwen2-vl-72b-instruct +`gte-rerank-v2` -- 在多个视觉理解基准测试中取得了最先进的成绩，显著增强多模态任务的处理能力。使用方法请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +gte-rerank-v2是通义实验室研发的多语言文本统一排序模型，面向全球多个主流语种，提供高水平的文本排序服务。通常用于语义检索、RAG等场景，可以简单、有效地提升文本检索的效果。给定查询 (Query) 和一系列候选文本 (documents)，模型会根据与查询的语义相关性从高到低对候选文本进行排序。 -意图理解 +文本生成 -2024-12-12 +2025-03-19 -中国内地 +`qwen-long-latest` -tongyi-intent-detect-v3 +`qwen-long-2025-01-25` -- 千问意图理解模型，能够在百毫秒级时间内快速、准确地解析用户意图，并选择合适工具来解决用户问题。详情请参见[意图理解](https://help.aliyun.com/zh/model-studio/intent-detect-capability)。 - +千问系列上下文窗口最长，能力均衡且成本较低的模型，适合长文本分析、信息抽取、总结摘要和分类打标等任务。 -声动人像 +语音合成 -2024-12-10 +2025-03-18 -中国内地 +`sambert-eva-v1` -videoretalk +`sambert-beth-v1` -- 支持根据人物视频和音频生成对口型视频，详情请参见[快速开始](https://help.aliyun.com/zh/model-studio/videos/videoretalk-quick-start)。 - +`sambert-waan-v1` -舞动人像 +`sambert-betty-v1` -2024-12-10 +`sambert-brian-v1` -中国内地 +`sambert-cally-v1` -animate-anyone-gen2 +`sambert-cindy-v1` -animate-anyone-detect-gen2 +`sambert-clara-v1` -animate-anyone-template-gen2 +`sambert-donna-v1` -- 分别提供人物图片合规检测、人物动作模板生成与人物视频生成能力，依次调用这三个模型可生成人物舞蹈视频。详情请参见[快速开始](https://help.aliyun.com/zh/model-studio/videos/animateanyone-quick-start)。 - +`sambert-hanna-v1` -语音合成 +`sambert-indah-v1` -2024-12-10 +`sambert-perla-v1` -中国内地 +`sambert-zhida-v1` -cosyvoice-v1 +`sambert-zhide-v1` -- 用户仅需提供10~20秒的音频，即可迅速生成高度相似且听感自然的定制声音。详情请参见[声音复刻/设计](https://help.aliyun.com/zh/model-studio/voice-replica-1/)。 - +`sambert-zhimo-v1` -llama系列 +`sambert-zhina-v1` -2024-12-09 +`sambert-zhiqi-v1` -中国内地 +`sambert-zhiru-v1` -llama3.3-70b-instruct +`sambert-zhiya-v1` -- 新增第三方大模型llama3.3系列中70B参数的模型。详情请参见[Llama（输入文本和图像）](https://help.aliyun.com/zh/model-studio/api-details-of-llama3-2-text-image)。 - +`sambert-zhiye-v1` -音频理解 +`sambert-camila-v1` -2024-12-09 +`sambert-zhichu-v1` -中国内地 +`sambert-zhifei-v1` -qwen-audio-turbo-latest +`sambert-zhigui-v1` -qwen-audio-turbo-2024-12-04 +`sambert-zhihao-v1` -- 新增的模型，相较于qwen-audio-turbo-2024-08-07版本，大幅提升语音识别准确率，并新增了语音聊天能力。详情请参见[音频理解-Qwen-Audio](https://help.aliyun.com/zh/model-studio/audio-language-model)。 - +`sambert-zhijia-v1` -文生文 +`sambert-zhilun-v1` -2024-11-28 +`sambert-zhimao-v1` -中国内地 +`sambert-zhinan-v1` -qwq-32b-preview +`sambert-zhishu-v1` -- 本模型专注于增强 AI 推理能力。详情请参见QWQ。 - +`sambert-zhiwei-v1` -2024-11-28 +`sambert-zhiyue-v1` -中国内地 +`sambert-zhijing-v1` -qwen-plus-2024-11-25 +`sambert-zhiming-v1` -qwen-plus-2024-11-27 +`sambert-zhiqian-v1` -- 相较于qwen-plus-0919模型中英文回复详实程度显著提升，更加符合用户偏好；模型角色扮演能力显著增强；模型中文的文本创作能力显著提升；中英文指令遵循能力提升；修复了RAG场景下引用角标的生成问题。详情请参见[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)。 - +`sambert-zhishuo-v1` -2024-11-15 +`sambert-zhiting-v1` -中国内地 +`sambert-zhixiao-v1` -qwen-turbo-2024-11-01 +`sambert-zhiying-v1` -- 上下文长度扩展至一百万Token。详情请参见[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)。 - +`sambert-zhixiang-v1` -视觉理解 +`sambert-zhistella-v1` -2024-11-15 +`sambert-zhimiao-emo-v1` -中国内地 +提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。 -qwen-vl-max-2024-11-19 +语音合成 -- 增强了图像理解效果，改善生成重复文本的情况。详情请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +2025-03-18 -中国内地 +`cosyvoice-v1` -qwen-vl-max-2024-10-30 +CosyVoice 是通义实验室依托大规模预训练语言模型，深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。 -- 加强了多语言理解能力。详情请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +全模态 -中国内地 +2025-03-17 -- qwen-vl-ocr - -- qwen-vl-ocr-2024-10-28 - -- qwen-vl-ocr-latest - +`qwen-omni-turbo-2025-01-19` -- qwen-vl-ocr是专用于OCR的模型；在表格、试题等类型图像的文字提取能力大幅提升。详情请参见[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr)。 - +千问全模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色，此版本为2025年1月19日的快照版本，预计维护至下一个快照发布前的一个月左右。 -文生文 +深度思考 -2024-11-12 +2025-03-05 -中国内地 +`qwq-plus` -- 闭源版本qwen-coder-plus等模型 - -- 开源版本qwen2.5-coder-32b-instruct等模型 - +千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。 -- 这些模型在代码生成、代码修复及代码推理能力上具备业界领先水平。详情请参见[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder)。 - +实时语音翻译 -灵动人像 +2025-03-04 -2024-11-7 +`gummy-realtime-v1` -中国内地 +多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。 -- liveportrait-detect - -- liveportrait - +语音识别 -- 基于人物肖像图片和人声音频文件，**快速、轻量**地生成人物肖像动态视频。详情请参见[灵动人像LivePortrait 快速开始](https://help.aliyun.com/zh/model-studio/liveportrait-quick-start/)。 - +2025-03-04 -悦动人像 +`gummy-chat-v1` -2024-11-7 +多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。 -中国内地 +视频生成 -- emo-detect-v1 - -- emo-v1 - +2025-02-27 -- 基于人物肖像图片和人声音频文件，生成人物肖像动态视频，不需要部署，可直接调用。与旧版的EMO模型（emo-detect、emo）相比，在调用方式及计费方式上有区别。详情请参见[悦动人像EMO 快速开始](https://help.aliyun.com/zh/model-studio/emo-quick-start/)。 - +`wanx2.1-i2v-turbo` -llama系列 +万相2.1-图生视频-Turbo，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成速度更快。 -2024-11-5 +全模态 -中国内地 +2025-02-14 -- llama3.2-90b-vision-instruct - -- llama3.2-11b-vision - +`qwen-omni-turbo` -- 新增第三方大模型llama3.2系列中11B和90B参数的模型，这两个模型加入了视觉理解的功能。详情请参见[Llama模型API参考](https://help.aliyun.com/zh/model-studio/api-details-of-llama3-2-text-image)。 - +`qwen-omni-turbo-latest` -视觉理解 +千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 -2024-10-29 +文本生成 -中国内地 +2025-02-05 -qwen2-vl-2b-instruct +`deepseek-r1-distill-qwen-7b` -- 扩展上下文至32k，大幅提升图像理解能力。详情请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +DeepSeek-R1-Distill-Qwen-7B是一个基于Qwen2.5-Math-7B的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。 -视觉理解 +文本生成 -2024-10-23 +2025-02-05 -中国内地 +`deepseek-r1-distill-qwen-32b` -qwen2-vl-7b-instruct +DeepSeek-R1-Distill-Qwen-32B是一个基于Qwen2.5-32B的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。 -- 新增qwen2-vl-7b-instruct模型，详情请参见[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision)。 - +文本生成 -文生文 +2025-02-05 -2024-09-18 +`deepseek-r1-distill-qwen-14b` -中国内地 +DeepSeek-R1-Distill-Qwen-14B是一个基于Qwen2.5-14B的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。 -qwen-long +文本生成 -- 新增qwen-long模型，详情请参见[长上下文（Qwen-Long）](https://help.aliyun.com/zh/model-studio/long-context-qwen-long) - +2025-02-05 -视觉理解 +`deepseek-r1-distill-qwen-1.5b` -2024-08-27 +DeepSeek-R1-Distill-Qwen-1.5B是一个基于Qwen2.5-Math-1.5B的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。 -中国内地 +文本生成 -qwen-vl-max-0809 +2025-02-03 -- 本模型为qwen-vl-max的2024年8月9日快照版本，将在9月9日更新至qwen-vl-max主版本，快照版本维护到下个快照版本发布时间（待定）后一个月。点击[千问VL API详情](https://help.aliyun.com/zh/document_detail/2712587.html)查看完整信息。 - +`qwen-plus-2025-01-25` -文生文 +千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。相对于通义千问-Plus-2025-0112版本，整体中英文能力都有综合能力升级，中英文code能力、逻辑能力、多语言能力显著提升，回复风格面向人类偏好进行大幅调整，尤其是数学、逻辑推理、知识问答等客观类query，模型回复详实程度和格式清晰度明显改善，创作类专项、json格式遵循专项、角色扮演专项能力均定向提升，预计维护至下一个快照上线前一个月。 -2024-08-16 +文本生成、深度思考 -中国内地 +2025-01-27 -Qwen2-Math系列模型 +`deepseek-v3` -- 具有强大的数学解题能力，点击[通过API调用Qwen2-Math模型](https://help.aliyun.com/zh/document_detail/2844170.html)进行体验。 - +DeepSeek-V3 为自研 MoE 模型，671B 参数，激活 37B，在 14.8T token 上进行了预训练，在长文本、代码、数学、百科、中文 能力上表现优秀。 -文生图 +视频生成 -2024-08-07 +2025-01-20 -中国内地 +`wanx2.1-i2v-plus` -FLUX文生图模型 +万相2.1-图生视频-Plus，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，视频质量更高。 -- FLUX文生图模型是由 Black Forest Labs 开源的高质量文本到图像生成模型，它在多个维度上展现了卓越性能，尤其在文本引导的图像生成、多主体场景构建、以及精细的手部细节生成等方面，实现了显著的提升，为文生图领域设定了新的技术标杆。点击[文生图FLUX](https://help.aliyun.com/zh/model-studio/videos/flux)进行了解。 - +图片生成 -llama系列 +2025-01-20 -2024-07-23 +`wanx2.0-t2i-turbo` -中国内地 +Wan2.0-T2I-Turbo，更擅长质感人像和创意设计画作生成，在图像美观度、真实感、艺术性上全面升级，支持最大200万像素生成，支持智能提示词改写等。 -- Llama3.1-8b-instruct - -- Llama3.1-70b-instruct - -- Llama3.1-405b-instruct - +视频生成 -- 新增Llama-3.1系列模型版本，点击[Llama （仅文本输入）](https://help.aliyun.com/zh/model-studio/api-details-of-llama-llm)查看。 - +2025-01-16 -文本向量 +`emoji-v1` -2024-07-10 +表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。 -中国内地 +视频生成 -text-embedding-v3 +2025-01-16 -- **text-embedding-v3**模型是**text-embedding-v2**模型的升级版本，主打**高性能、低成本、支持50+多语言、超长文本**。 - +`emoji-detect-v1` -> 更新内容主要包括： +表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 -- > **语种扩充**：**text-embedding-v3**模型对比**text-embedding-v2**模型扩展了意大利语、波兰语、越南语、泰语等语种，支持语种数量增加到50+。 - -- > **输入长度扩展**：支持编码的输入长度从2048扩展至8192, 对text-embedding-v3, 8192指文本输入的最大token长度 - -- > **可变输出连续向量维度**: 相比text-embedding-v2模型的固定1536向量维度, text-embedding-v3支持用户自定义连续向量的维度, 目前可以选择512,768和1024维度；同时为了进一步节省下游任务的使用成本，text-embedding-v3模型在不衰减效果的前提下将最大的向量维度降低至1024维。 - -- > **不再区分Query/Document类型：**text-embedding-v3模型在不降低模型效果的前提下不再区分输入文本的类型，text\_type参数无需指定输入的文本是Query还是Document类型。 - -- > **Sparse向量支持**: text-embedding-v3模型同时支持连续向量表示（dense vector）和离散向量表示模型（sparse vector）, 用户可以在接口参数中指定输出连续向量、离散向量或者同时输出。 - -- > **效果提升**：预训练模型底座和SFT策略优化提升embedding模型整体效果，公开数据评测结果。 - +文本生成 -文生文 +2025-01-15 -2024-07-05 +`qwen-plus-0112` -中国内地 +千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。相对于千问-Plus-2024-1220版本，中英文整体能力有提升，中英常识知识类、阅读理解能力提升较为显著，codeswtich现象相比上一版有显著改善，中文指令遵循能力显著提升。 -- qwen-turbo-0206 - -- qwen-turbo-0624 - +图片生成 -- 新增qwen-turbo模型的快照版本，点击[文本生成](https://help.aliyun.com/zh/model-studio/qwen-api-reference/)查看详情。 - +2025-01-15 -文生文 +`aitryon-parsing-v1` -2024-07-05 +图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。 -中国内地 +视频生成 -- qwen-plus-0206 - -- qwen-plus-0624 - +2025-01-09 -- 新增qwen-plus模型的快照版本，点击[文本生成](https://help.aliyun.com/zh/model-studio/qwen-api-reference/)查看详情。 - +`wanx2.1-t2v-turbo` -文生文 +万相2.1-文生视频-Turbo，一句话生成视频。生成速度更快，支持大幅度复杂运动、现实物理规律还原、丰富的艺术风格和影视级画面质感，指令遵循能力进一步提升。 -2024-07-02 +视频生成 -中国内地 +2025-01-09 -Minimax大语言模型 +`wanx2.1-t2v-plus` -- 新增Minimax大语言模型，点击[MiniMax大语言模型](https://help.aliyun.com/zh/model-studio/videos/minimax-llm/)查看详情。 - +万相2.1-文生视频-Plus，一句话生成视频。视频品质更高，支持大幅度复杂运动、现实物理规律还原、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升。 -文生文 +图片生成 -2024-06-07 +2025-01-09 -中国内地 +`wanx2.1-t2i-turbo` -qwen2系列开源大语言模型 +万相2.1-文生图-Turbo，更快的生成速度，在图像美观度、真实感、艺术性上全面升级，更强的语义理解能力、丰富的风格泛化性、支持最大200万像素生成，支持智能提示词改写等。 -- 与qwen1.5相比，qwen2在语言理解、语言生成、多语言能力、编码、推理等基准测试中超越了大多数开源模型。点击[模型介绍](https://help.aliyun.com/zh/document_detail/2804542.html)查看详情。 - +图片生成 -文生文 +2025-01-09 -2024-06-03 +`wanx2.1-t2i-plus` -中国内地 +万相2.1-文生图-Plus，更丰富的画面细节，在图像美观度、真实感、艺术性上全面升级，更强的语义理解能力、丰富的风格泛化性、支持最大200万像素生成，支持智能提示词改写等。 -零一万物大语言模型 +实时语音识别 -零一万物大语言模型是千亿参数大语言模型，是LM SYS榜单TOP10上唯一国产大模型。具备超强的问答、推理及文本生能力，完整内容请点击[零一万物大语言模型](https://help.aliyun.com/zh/model-studio/videos/yi-large-llm/)查看。 +2024-12-31 -文生文 +`paraformer-realtime-8k-v2` -2024-04-28 +推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language\_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服等场景下的实时语音识别。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 -中国内地 +文本生成 -qwen1.5-110b-chat +2024-12-26 -- qwen1.5-110b-chat模型时千问1.5对外开源的110B规模参数量的经过人类指令对齐的chat模型。点击[通义千问开源模型](https://help.aliyun.com/zh/document_detail/2713159.html)查看详情。 - +`qwen-plus-1220` -## 新加坡 +千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。相对于千问-Plus-1125版本，中英文整体能力有提升，中英常识知识类、阅读理解能力提升较为显著，codeswtich现象相比上一版有显著改善，中文指令遵循能力显著提升。 -**模型类型** +多模态向量 -**时间** +2024-12-23 -**服务部署范围** +`multimodal-embedding-v1` -**模型规格** +通义实验室基于预训练多模态大模型构建的多模态向量模型。该模型根据用户的输入生成高维连续向量，这些输入可以是文本、图片或视频。多模态向量在可应用于图片搜索、文搜图、视频搜索、图片分类和视频内容审核等下游任务中。 -**功能说明** +语音识别 -语音合成 +2024-12-19 -2026-07-14 +`paraformer-8k-v2` -国际 +Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。 -qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash +文本生成 -Qwen-Audio-TTS语音合成模型上线，新增更多小语种和中文方言支持，增强了指令遵循与细粒度标签控制能力，音质和表现力全面提升。其中 Plus 版本面向高品质专业场景，Flash 版本面向低延迟实时交互场景，首包延时控制在 200ms 以内。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +2024-12-12 -参考生视频 +`tongyi-intent-detect-v3` -2026-07-01 +意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。 -国际 +视频生成 -wan2.7-r2v-2026-06-12 +2024-12-10 -万相2.7参考生视频模型快照版本，支持主体参考和音色定制，并可输入单张多宫格故事板直接生成剧本化视频。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +`videoretalk` -推理模型 +VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。 -2026-06-26 +视频生成 -国际 +2024-12-10 -kimi-k2.7-code +`animate-anyone-template-gen2` -Kimi K2.7 Code 模型新加坡地域上线。以编码为中心的智能体模型，专为长程软件工程任务优化，仅支持思考模式。[](#) +AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。 -图像生成 +视频生成 -2026-06-25 +2024-12-10 -国际 +`animate-anyone-gen2` -qwen-image-2.0-pro-2026-06-22 +AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。 -Qwen-Image-2.0系列模型最新快照，融合图片生成与编辑能力。相较于4月22日快照，文字渲染能力进一步增强，支持最长1k token的指令输入；真实质感与写实场景细节刻画更加细腻；语义遵循能力更强。 +视频生成 -[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api)、[千问-图像编辑](https://help.aliyun.com/zh/model-studio/qwen-image-edit-api) +2024-12-10 -文生视频 +`animate-anyone-detect-gen2` -2026-06-22 +AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 -国际 +视觉理解 -happyhorse-1.1-t2v +2024-11-14 -HappyHorse 1.1系列文生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-文生视频](https://help.aliyun.com/zh/model-studio/happyhorse-text-to-video-api-reference) +`qwen-vl-ocr-1028` -图生视频 +千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。本模型为2024年10月28日的快照版本。 -2026-06-22 +文本生成 -国际 +2024-11-12 -happyhorse-1.1-i2v +`qwen-coder-plus` -HappyHorse 1.1系列图生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/happyhorse-image-to-video-api-reference) +千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。 -参考生视频 +视频生成 -2026-06-22 +2024-11-07 -国际 +`liveportrait-detect` -happyhorse-1.1-r2v +LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 -HappyHorse 1.1系列参考生视频模型，支持多参考图输入生成有声视频，可生成3~15秒、720P/1080P视频。[HappyHorse-参考生视频](https://help.aliyun.com/zh/model-studio/happyhorse-reference-to-video-api-reference) +视频生成 -推理模型 +2024-11-07 -2026-06-10 +`liveportrait` -国际 +LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。 -qwen3.7-max-2026-06-08 +视频生成 -Qwen3.7系列中规模最大、综合能力最强的Max模型，相较于5月20日快照增加了视觉模态理解能力，能够感知真实世界场景，具备多模态交互混合智能体能力。 +2024-11-07 -推理模型 +`emo-v1` -2026-06-01 +EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。 -国际 +视频生成 -qwen3.7-plus、qwen3.7-plus-2026-05-26 +2024-11-07 -千问3.7Plus系列，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 +`emo-detect-v1` -推理模型 +EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 -2026-05-27 +文本生成 -国际 +2024-10-15 -glm-5.1 +`qwen-max` -智谱GLM-5.1模型，专为长程任务设计，支持 200K 上下文，最大输出可达 128K Token。通过强大的逻辑推理、长文本理解及代码生成能力，在多项基准测试中表现优异，适用于智能交互、企业应用及开发辅助等场景。[GLM-阿里云](https://help.aliyun.com/zh/model-studio/glm) +千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。 -推理模型 +文本生成 -2026-05-21 +2024-09-19 -国际 +`qwen-math-turbo` -qwen3.7-max、qwen3.7-max-2026-05-20 +千问系列数学模型是专门用于数学解题的语言模型，推理速度快，成本低。 -Qwen Max 系列新一代旗舰模型。仅支持纯文本输入，默认开启思考模式，支持显式缓存，在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 +文本生成 -音视频翻译 +2024-09-19 -2026-05-19 +`qwen-math-plus-0919` -国际 +`qwen-math-plus-latest` -qwen3.5-livetranslate-flash-realtime、qwen3.5-livetranslate-flash-realtime-2026-05-19 +千问系列数学模型是专门用于数学解题的语言模型，推理效果好，模型性能优秀本模型为2024年9月19日快照版本，预计维护到下个版本发布后一个月（待定）。 -一款多语言音视频实时翻译模型，可识别 60 种语言，并实时翻译为 29 种语言的音频。[实时语音/音视频翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-5-livetranslate-flash-realtime) +文本生成 -文生视频 +2024-09-19 -2026-04-27 +`qwen-coder-turbo` -国际 +千问系列代码及编程模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。 -happyhorse-1.0-t2v +语音合成 -HappyHorse系列文生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-文生视频](https://help.aliyun.com/zh/model-studio/happyhorse-text-to-video-api-reference) +2024-09-13 -图生视频 +`voice-enrollment` -2026-04-27 +大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。 大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。 -国际 +视频生成 -happyhorse-1.0-i2v +2024-09-13 -HappyHorse系列图生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/happyhorse-image-to-video-api-reference) +`video-style-transform` -参考生视频 +视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。 -2026-04-27 +语音识别 -国际 +2024-09-13 -happyhorse-1.0-r2v +`speech-biasing` -HappyHorse系列参考生视频模型，支持多参考图输入生成有声视频，可生成3~15秒、720P/1080P视频。[HappyHorse-参考生视频](https://help.aliyun.com/zh/model-studio/happyhorse-reference-to-video-api-reference) +热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。 -视频编辑 +文本生成 -2026-04-27 +2024-09-13 -国际 +`qwen-math-plus` -happyhorse-1.0-video-edit +`qwen-math-plus-0816` -HappyHorse系列视频编辑模型，支持对视频进行编辑处理。[HappyHorse-视频编辑](https://help.aliyun.com/zh/model-studio/happyhorse-video-edit-api-reference) +千问数学模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。 -文生视频 +语音合成 -2026-07-01 +2024-09-13 -国际 +`cosyvoice-clone-v1` -wan2.7-t2v-2026-06-12 +声音复刻Cosyvoice大模型，依托先进的大模型技术进行特征提取，从而完成声音的复刻，且无需训练过程。仅需提供时长较短的音频，即可迅速生成高度相似且听感自然的定制声音。 -万相2.7-文生视频模型快照版本，模型能力与wan2.7-t2v一致。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +语音识别 -文生视频 +2024-09-06 -2026-04-26 +`paraformer-v2` -国际 +推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language\_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。可支持热词。 -wan2.7-t2v-2026-04-25 +实时语音识别 -万相2.7-文生视频模型快照版本，模型能力与wan2.7-t2v一致。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +2024-09-06 -图生视频 +`paraformer-realtime-v2` -2026-04-26 +推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language\_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 可支持热词。 -国际 +图片生成 -wan2.7-i2v-2026-04-25 +2024-08-19 -万相2.7-图生视频模型快照版本，模型能力与wan2.7-i2v一致。[万相2.7-图生视频](https://help.aliyun.com/zh/model-studio/image-to-video-general-api-reference) +`image-instance-segmentation` -图像生成 +人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。 -2026-04-23 +图片生成 -国际 +2024-08-19 -qwen-image-2.0-pro-2026-04-22 +`image-erase-completion` -Qwen-Image-2.0系列模型，实现了图片生成和图片编辑的融合。相较于3月3日快照，本模型在画面质感，尤其是纹理细节、光影、材质上有明显跃升；支持多语言的图内文字生成；艺术风格表现更加均衡。 +图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。 -[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api)、[千问-图像编辑](https://help.aliyun.com/zh/model-studio/qwen-image-edit-api) +文本向量 -推理模型 +2024-07-12 -2026-04-23 +`text-embedding-v3` -国际 +通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。 -qwen3.5-plus-2026-04-20 +图片生成 -Qwen3.5原生视觉语言系列Plus模型新快照，相较2月15日快照Agentic coding能力大幅提升，推理速度显著提升；知识、推理与长上下文能力保持较高水准，适合编码智能体、生产工作流和高吞吐场景。 +2024-06-28 -推理模型 +`aitryon-refiner` -2026-04-23 +图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。 -国际 +图片生成 -qwen3.6-27b +2024-06-25 -Qwen3.6系列27B原生视觉语言Dense模型，相较3.5-27B重点提升Agentic coding能力，STEM与推理能力进一步增强；视觉模态方面空间智能、物体定位与检测能力显著增强，视频理解、文档OCR及视觉Agent能力稳步提升。 +`wanx-virtualmodel` -推理模型 +虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。 -2026-04-20 +图片生成 -国际 +2024-06-25 -qwen3.6-max-preview +`virtualmodel-v2` -Qwen3.6系列规模最大的闭源模型，Coding能力进一步提升、Agent执行更加高效。仅支持纯文本输入，支持思考模式（默认开启），支持显式缓存和Function Calling。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。 -> 不支持图像与视频输入。 +图片生成 -推理模型 +2024-06-21 -2026-04-16 +`wanx-poster-generation-v1` -国际 +创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。 -qwen3.6-flash、qwen3.6-flash-2026-04-16、qwen3.6-35b-a3b +图片生成 -Qwen3.6 原生视觉语言 Flash 系列模型，在整体性能上较 Qwen3.5-Flash 显著提升。重点增强了智能体编程能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；在视觉能力方面，空间智能显著增强，其中物体定位和目标检测表现尤为突出。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +2024-06-21 -语音识别 +`shoemodel-v1` -2026-04-16 +鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。 -国际 +图片生成 -fun-asr、fun-asr-2025-11-07 +2024-06-11 -Fun-ASR 实时语音识别能力升级： +`wanx-sketch-to-image-lite` -- 方言支持：覆盖汉语七大方言（官话/吴/湘/赣/客/闽/粤）及 20+ 地区口音 - -- 古诗词优化：提升古诗词识别准确率，适用于文化教育、有声读物等场景 - -- 文本优化：标点预测与文本归一化增强，数字/日期/金额自动转标准格式 - -- 多语种扩展：支持中、英、日、韩、法、德、西班牙等共 30 个语种 - +万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。 -详情请参见[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition) +实时语音识别 -视频编辑 +2024-06-06 -2026-04-03 +`paraformer-realtime-v1` -国际 +Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。 -wan2.7-videoedit +实时语音识别 -万相2.7-视频编辑模型，支持指令编辑与视频迁移任务，可修改视频局部或整体画面，同时支持多图参考替换及动作、特效、运镜的复刻。[万相2.7-视频编辑](https://help.aliyun.com/zh/model-studio/wan-video-editing-api-reference) +2024-06-06 -图生视频 +`paraformer-realtime-8k-v1` -2026-04-03 +Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。 -国际 +图片生成 -wan2.7-i2v +2024-05-28 -万相2.7-图生视频模型，支持多模态输入（文本/图像/音频/视频），可完成首帧生视频、首尾帧生视频、视频续写三大任务。[万相2.7-图生视频](https://help.aliyun.com/zh/model-studio/image-to-video-general-api-reference) +`wanx-x-painting` -文生视频 +万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。 -2026-04-03 +图片生成 -国际 +2024-05-24 -wan2.7-t2v +`image-out-painting` -万相2.7-文生视频模型，新增分辨率档位与宽高比自定义设置，支持灵活适配不同创作场景与平台发布需求。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。 -参考生视频 +图片生成 -2026-04-03 +2024-05-24 -国际 +`aitryon` -wan2.7-r2v +aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。 -万相2.7参考生视频模型，支持主体参考和音色定制，并可输入单张多宫格故事板直接生成剧本化视频。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +文本生成 -推理模型 +2024-05-20 -2026-04-02 +`qwen-long` -国际 +Qwen-Long是在通义千问针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服务，可支持文本文件（ TXT、DOCX、PDF、XLSX、EPUB、MOBI、MD、CSV）和图片文件（BMP、PNG、JPG/JPEG、GIF 以及PDF扫描件）的解析和对话。说明：通过HTTP直接提交请求，支持1M tokens长度，超过此长度建议通过文件方式提交。 -qwen3.6-plus、qwen3.6-plus-2026-04-02 +文本生成 -千问3.6-Plus，代码开发能力重点升级（Agentic Coding、前端编程等），Vibe Coding体验显著提升；泛化场景推理能力进一步增强；多模态方面，万物识别、OCR、物体定位等能力显著提升；同时修复了Qwen3.5-Plus上线后的已知问题。使用方法与qwen3.5-plus一致。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +2024-05-14 -图像生成与编辑 +`farui-plus` -2026-04-01 +通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分析、生成法律文书、检索法律知识、审查合同条款等功能 -国际 +图片生成 -wan2.7-image-pro、wan2.7-image +2024-04-09 -万相2.7-图像生成与编辑模型，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循表现更优。Pro系列支持4K输出；加速版兼顾效果与响应速度。[万相-图像生成与编辑2.7](https://help.aliyun.com/zh/model-studio/wan-image-generation-and-editing-api-reference) +`wordart-texture` -全模态 +WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。 -2026-03-30 +图片生成 -国际 +2024-04-09 -qwen3.5-omni-plus、qwen3.5-omni-plus-2026-03-15、qwen3.5-omni-flash、qwen3.5-omni-flash-2026-03-15 +`wordart-semantic` -最新一代全模态大模型，支持长视频分析、会议纪要、字幕输出、安全审核、音视频交互；支持音视频内容的深度理解与生成描述，支持 113 种语言识别和 36 种语言的音频生成，可处理 3 小时音频及1 小时视频输入，支持联网搜索及指令来控制输出音频的音量、语速、情绪。[非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni) +WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。 -全模态 +文本向量 -2026-03-30 +2024-04-09 -国际 +`text-embedding-v2` -qwen3.5-omni-plus-realtime、qwen3.5-omni-plus-realtime-2026-03-15、qwen3.5-omni-flash-realtime、qwen3.5-omni-flash-realtime-2026-03-15 +通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。 -千问最新推出的实时多模态模型，相比于上一代的 Qwen3-Omni-Flash-Realtime：模型智力大幅提升，与 Qwen3.5-Plus 智能水平相当。原生支持联网搜索（WebSearch），支持语音打断和控制；支持 113 种语种和方言的语音识别，以及 36 种语种和方言的语音生成。[实时（Qwen-Omni-Realtime）](https://help.aliyun.com/zh/model-studio/realtime) +文本向量 -推理模型 +2024-04-09 -2026-03-20 +`text-embedding-v1` -国际 +通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。 -deepseek-v3.2 +文本向量 -DeepSeek-V3.2是引入DeepSeek Sparse Attention（一种稀疏注意力机制）的正式版模型，也是DeepSeek推出的首个将思考融入工具使用的模型，同时支持思考模式与非思考模式的工具调用。 +2024-04-09 -[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api) +`text-embedding-async-v2` -图像生成与编辑 +通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。 -2026-03-03 +文本向量 -国际 +2024-04-09 -qwen-image-2.0、qwen-image-2.0-2026-03-03、qwen-image-2.0-pro、qwen-image-2.0-pro-2026-03-03 +`text-embedding-async-v1` -千问-Image2.0系列，同时支持图像生成和编辑。Pro系列文字渲染、真实质感、语义遵循能力更强。；加速版兼顾效果与响应速度。[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api)、[千问-图像编辑](https://help.aliyun.com/zh/model-studio/qwen-image-edit-api) +通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。 语音识别 -2026-03-03 - -国际 - -qwen3-asr-flash-2026-02-10 +2024-04-09 -千问录音文件识别新增快照模型，较 qwen3-asr-flash-2025-09-08 效果更优。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +`paraformer-v1` -推理模型 +Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。 -2026-02-24 +语音识别 -国际 +2024-04-09 -qwen3.5-flash、qwen3.5-flash-2026-02-23、qwen3.5-122b-a10b、qwen3.5-27b、qwen3.5-35b-a3b +`paraformer-mtl-v1` -阿里巴巴推出的最新模型千问3.5-Flash和开源模型，支持文本、图像和视频输入，响应速度快，综合表现接近qwen3.5-plus，支持内置[工具调用](https://help.aliyun.com/zh/model-studio/tool-calls/)。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。 支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话、宁夏话、山西话、陕西话、山东话、四川话、天津话）、英语、日语、韩语、西班牙语、印尼语、法语、德语、意大利语、马来语。 -代码模型 +语音识别 -2026-02-20 +2024-04-09 -国际 +`paraformer-8k-v1` -qwen3-coder-next +Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。 -Qwen3系列新一代开源代码生成模型，支持多轮工具交互，提升了对仓库级别代码的理解能力和对AI编程工具的适配性。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +图片生成 -推理模型 +2024-04-09 -2026-02-16 +`facechain-generation` -国际 +基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。 -qwen3.5-plus、qwen3.5-plus-2026-02-15、qwen3.5-397b-a17b +图片生成 -阿里巴巴推出的最新模型千问3.5-Plus和开源模型，支持文本、图像和视频输入，在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中表现卓越，支持内置工具调用。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +2024-04-09 -语音识别 +`facechain-facedetect` -2026-02-13 +对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。 -国际 +图片生成 -qwen3-asr-flash-realtime-2026-02-10 +2024-03-22 -千问实时语音识别新增最新快照模型，较 qwen3-asr-flash-realtime-2025-10-27 效果更优。[实时语音识别](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition-user-guide) +`wanx-style-repaint-v1` -语音合成 +人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。 -2026-02-10 +图片生成 -国际 +2024-03-22 -cosyvoice-v3-plus、cosyvoice-v3-flash +`wanx-background-generation-v2` -语音合成 CosyVoice 新增 v3 模型，支持使用系统音色和复刻音色进行语音合成。[实时语音合成-CosyVoice /Sambert](https://help.aliyun.com/zh/model-studio/text-to-speech) +图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。 -语音合成 +图片生成 -2026-02-10 +2024-01-05 -国际 +`wanx-v1` -qwen3-tts-instruct-flash、qwen3-tts-instruct-flash-2026-01-26 +万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通 -千问语音合成上线Instruct（指令控制）模型，支持通过自然语言指令精准控制合成效果。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +## 新加坡 -语音合成 +**模型类型** -2026-02-10 +**时间** -国际 +**服务部署范围** -qwen3-tts-vd-2026-01-26 +**模型ID** -千问语音合成上线声音设计模型，可通过文本描述创建定制化音色。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +**功能说明** -语音合成 +文本生成、深度思考、视觉理解 -2026-02-10 +2026-07-21 国际 -qwen3-tts-vc-2026-01-22 +`qwen3.7-flash` -千问语音合成上线声音复刻模型，可基于真实音频样本快速复刻音色。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +`qwen3.7-flash-2026-07-15` -语音合成 +Qwen3.7原生视觉语言系列Flash模型，相较3.6-Flash全面提升多模态理解与Agent执行能力。重点强化多模态基础能力、万物识别能力更强，真实世界感知与空间智能进一步提升，Search Agent、CI Agent等多模态Agent场景能力显著升级、端到端任务执行更稳定，多模态Coding能力优化、vibe coding 体验更加流畅。 -2026-02-04 +图片生成 + +2026-07-20 国际 -qwen3-tts-instruct-flash-realtime、qwen3-tts-instruct-flash-realtime-2026-01-22 +`qwen-image-3.0-pro` -千问实时语音合成新增Instruct（指令控制）模型，支持通过自然语言指令精准控制合成效果。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。 细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。 知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。 Qwen-Image-3.0-Pro 不只是在追求"好看"，更在追求“好用”——让图像生成真正成为可落地的生产力工具。 -参考生视频 +实时语音合成 -2026-02-02 +2026-07-14 国际 -wan2.6-r2v-flash +`qwen-audio-3.0-tts-plus` -基于参考视频和图像的角色形象，生成多镜头视频，支持自动配音。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +qwen-audio-3.0-tts-plus是面向高质量语音生成场景打造的高性能语音合成大模型。相比前一版本，模型支持更多小语种和中文方言，显著提升方言发音的正宗程度，并增强了 free-style 指令遵循能力和细粒度标签控制能力，可更准确地控制情绪、语气、角色、语速、音量和合成风格。同时，模型在噪声、混响等复杂声学条件下具备更强鲁棒性，进一步提升了音质、清晰度、分辨率和整体表现力。Plus 版本更强调合成效果和细节表现，适用于有更高音质、自然度和表现力要求的专业场景，如内容创作、有声书、影视配音、品牌声音设计和高品质语音服务。 -视觉理解 +实时语音合成 -2026-01-28 +2026-07-14 国际 -qwen3-vl-flash-2026-01-22 +`qwen-audio-3.0-tts-flash` -千问VL的全新快照版模型，有效融合了思考模式与非思考模式，相较于 2025 年 10 月 15 日的快照版本，显著提升了模型的整体性能，在通用视觉识别、安防、巡店、巡检、拍照解题等业务场景中实现了更高准确率的推理。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +qwen-audio-3.0-tts-flash是面向实时交互场景优化的高性能语音合成大模型。相比前一版本，模型支持更多小语种和中文方言，提升了方言发音的正宗程度，并增强了 free-style 指令遵循能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。同时，模型在噪声、混响等复杂声学条件下具备更强鲁棒性，提升了音质、清晰度和整体表现力。Flash 版本重点优化实时合成体验，首包延时控制在 200ms 以内，适用于语音助手、实时对话、智能客服等低延迟交互场景。 -语音识别 +文本生成 -2026-01-28 +2026-07-10 国际 -qwen3-asr-flash-filetrans、qwen3-asr-flash-filetrans-2025-11-17 +`glm-5.2-fast-preview` -千问3-ASR-Flash-Filetrans系列模型现已支持词级别时间戳，通过设置新参数 `enable_words`，获取毫秒级的词/字对齐信息，并体验更符合语义的精细化断句。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。 -推理模型 +视频生成 -2026-01-27 +2026-07-01 国际 -qwen3-max-2026-01-23 +`wan2.7-t2v-2026-06-12` -相较于 2025 年 9 月 23 日的快照版本，有效融合了思考模式与非思考模式，显著提升了模型的整体性能。在思考模式下，模型集成了 Web 搜索、网页信息提取和代码解释器三项工具，通过在思考过程中引入外部工具，在复杂问题上实现更高的准确率。[OpenAI兼容-Responses](https://help.aliyun.com/zh/model-studio/compatibility-with-openai-responses-api) +万相2.7-文生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。该版本为2026年6月12日快照。 -图生视频 +视频生成 -2026-01-18 +2026-07-01 国际 -wan2.6-i2v-flash +`wan2.7-r2v-2026-06-12` -支持生成有声与无声视频，两类视频按各自计费规则独立计费；同时具备多镜头叙事能力与音频处理能力。[万相-图生视频-基于首帧（2.1-2.6）](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +万相2.7-参考生视频，更加稳定的角色、道具与场景参考，支持最大5个图/视频混合参考，支持音频音色参考，搭配基础能力升级实现更强表演能力。该版本为2026年6月12日快照。 -图像编辑 +图片生成 -2026-01-18 +2026-06-25 国际 -qwen-image-edit-max、qwen-image-edit-max-2026-01-16 +`qwen-image-2.0-pro-2026-06-22` -千问图像编辑模型Max系列，具备更稳定、丰富的编辑能力，增强了工业设计与几何推理能力，并提升了角色一致性与编辑的精准度。[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide) +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。 -语音合成 +文本生成、深度思考、视觉理解 -2026-01-16 +2026-06-25 国际 -qwen3-tts-vc-realtime-2026-01-15 +`kimi-k2.7-code` -千问实时语音合成新增最新快照模型，[声音复刻（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning)效果进一步优化，较 qwen3-tts-vc-realtime-2025-11-27 更自然、更贴近原声。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +kimi-k2.7-code是 kimi 迄今最智能的coding模型，在长上下文中更可靠地遵循指令，能以更高的成功率完成编程任务，同时支持文本、图片与视频输入，思考模式，对话与 Agent 任务。 -文生图 +文本生成、深度思考 -2026-01-12 +2026-06-25 国际 -qwen-image-plus-2026-01-09 +`glm-5.2` -千问图像生成的全新快照版模型，为qwen-image-max的蒸馏加速版，支持快速生成高质量图像。[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api) +GLM-5.2是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 语音识别 -2026-01-06 +2026-06-17 国际 -qwen3-asr-flash、qwen3-asr-flash-2025-09-08 +`fun-asr-flash-2026-06-15` -千问3-ASR-Flash支持OpenAI兼容模式。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。 -文生图 +视频生成 -2025-12-31 +2026-06-16 国际 -qwen-image-max、qwen-image-max-2025-12-30 +`happyhorse-1.1-t2v` -千问图像生成模型Max系列，相较于Plus系列提升了图像的真实感与自然度，有效降低了AI合成痕迹，在人物质感、纹理细节和文字渲染等方面表现突出。[千问-文生图](https://help.aliyun.com/zh/model-studio/qwen-image-api) +HappyHorse-1.1-T2V支持文生视频，进一步提升文本语义理解、镜头调度与动态生成表现。模型能够更精准地还原创作意图，在人物动作、场景氛围、视觉美感和物理运动上生成更流畅自然、细节丰富且一致性更高的高质量视频。 -图像编辑 +视频生成 -2025-12-23 +2026-06-16 国际 -qwen-image-edit-plus-2025-12-15 +`happyhorse-1.1-r2v` -千问图像编辑发布的最新快照模型，相较于上一版本提升了角色一致性、工业设计能力和几何推理能力，并优化了编辑后的图片与原图在空间布局、纹理和风格上的匹配度，编辑效果更精准。[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide) +HappyHorse-1.1-R2V支持参考生视频，进一步提升主体、场景风格与画面一致性的稳定保持。模型最多支持9张参考图片输入，能够更精准理解并延续创作意图，在人物、场景、风格和镜头表现上带来更强的可控性与表现力。 -文生图 +视频生成 -2025-12-22 +2026-06-16 国际 -z-image-turbo +`happyhorse-1.1-i2v` -轻量级文生图模型，可快速生成高质量图像，支持中英双语渲染、复杂语义理解和多风格题材，并可灵活适配多种分辨率与宽高比。[文生图Z-Image](https://help.aliyun.com/zh/model-studio/z-image-api-reference) +HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。 -视觉理解 +文本生成、深度思考、视觉理解 -2025-12-19 +2026-06-10 国际 -qwen3-vl-plus-2025-12-19 +`qwen3.7-max-2026-06-08` -千问VL的全新快照版模型，指令遵循能力更强，具有更低的延迟。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +Qwen3.7系列中规模最大、综合能力最强的Max模型，相较于5月20日快照增加了视觉模态理解能力，能够感知真实世界场景，具备多模态交互混合智能体能力。该版本为2026年6月8日快照。 -语音识别 +文本生成、深度思考、视觉理解 -2025-12-19 +2026-06-01 国际 -qwen3-asr-flash-filetrans、qwen3-asr-flash-filetrans-2025-11-17、qwen3-asr-flash、qwen3-asr-flash-2025-09-08 +`qwen3.7-plus` -新增捷克语、丹麦语等共 9 种语言的语音识别支持。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +`qwen3.7-plus-2026-05-26` -语音识别 +Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 + +文本生成、深度思考 -2025-12-17 +2026-05-21 国际 -qwen3-asr-flash-realtime、qwen3-asr-flash-realtime-2025-10-27 +`qwen3.7-max` -新增捷克语、丹麦语等共 9 种语言的语音识别支持。[实时语音识别](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition-user-guide) +`qwen3.7-max-2026-05-20` -语音识别 +Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 -2025-12-17 +实时语音翻译 + +2026-05-19 国际 -qwen3-asr-flash、qwen3-asr-flash-2025-09-08 +`qwen3.5-livetranslate-flash-realtime` -支持任意采样率和声道的音频。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +`qwen3.5-livetranslate-flash-realtime-2026-05-19` -语音识别 +Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。 + +文本生成、深度思考 -2025-12-17 +2026-05-11 国际 -fun-asr-mtl、fun-asr-mtl-2025-08-25 +`deepseek-v4-pro` -支持对中、英、日、韩等共 31 种语言的语音识别，尤其适合东南亚出海场景。[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition) +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 -声音设计 +文本生成、深度思考 -2025-12-16 +2026-05-11 国际 -qwen-voice-design +`deepseek-v4-flash` -千问发布声音设计模型，通过文本描述生成定制化音色。结合qwen3-tts-vd-realtime-2025-12-16模型使用生成语音，覆盖 10 种语言。[声音设计（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design) +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 -语音合成 +视频生成 -2025-12-16 +2026-04-26 国际 -qwen3-tts-vd-realtime-2025-12-16（快照版） +`wan2.7-t2v-2026-04-25` -千问实时语音合成发布全新快照版模型，可使用[声音设计（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-design)生成的音色进行低延迟、高稳定性的实时合成；支持多语言输出；能根据文本自动调节语气，并优化复杂文本的合成表现。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +万相2.7-文生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。该版本为2026年4月25日快照。 -文生图 +视频生成 -2025-12-16 +2026-04-26 国际 -wan2.6-t2i +`wan2.7-i2v-2026-04-25` -新增同步接口。支持在总像素面积与宽高比约束内，自由选尺寸。[万相-文生图V2](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference) +万相2.7-图生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。该版本为2026年4月25日快照。 -图像生成与编辑 +视频生成 -2025-12-16 +2026-04-26 国际 -wan2.6-image +`happyhorse-1.0-video-edit` -支持图像编辑和图文混合输出。[万相-图像生成与编辑2.6](https://help.aliyun.com/zh/model-studio/wan-image-generation-api-reference) +HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。 -图生视频-基于首帧 +视频生成 -2025-12-16 +2026-04-26 国际 -wan2.6-i2v +`happyhorse-1.0-r2v` -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[万相-图生视频-基于首帧（2.1-2.6）](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +HappyHorse-1.0-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。 -参考生视频 +文本生成、深度思考、视觉理解 -2025-12-16 +2026-04-23 国际 -wan2.6-r2v +`qwen3.5-plus-2026-04-20` -基于参考视频的角色形象和音色，生成多镜头视频，支持自动配音。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +Qwen3.5原生视觉语言系列Plus模型，相较于2月15日快照，本模型在Agentic coding能力上大幅提升；推理速度显著提升；知识、推理与长上下文能力保持较高水准，满足复杂Agent任务的需求，适合应用于编码智能体、生产工作流和高吞吐场景。该版本为2026年4月20日快照。 -文生视频 +图片生成 -2025-12-16 +2026-04-23 国际 -wan2.6-t2v +`qwen-image-2.0-pro` -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +`qwen-image-2.0-pro-2026-04-22` -语音识别 +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。 -2025-12-12 +文本生成、深度思考、视觉理解 + +2026-04-22 + +国际 + +`qwen3.6-27b` + +Qwen3.6系列27B原生视觉语言Dense模型，模型效果相较3.5-27B重点提升了Agentic coding能力、模型STEM与推理能力进一步增强；视觉模态方面在空间智能、物体定位与检测能力上显著增强，视频理解、文档OCR及视觉Agent能力稳步提升。 + +视频生成 + +2026-04-22 + +国际 + +`happyhorse-1.0-t2v` + +HappyHorse-1.0-T2V支持文生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 + +视频生成 + +2026-04-22 + +国际 + +`happyhorse-1.0-i2v` + +HappyHorse-1.0-I2V支持图生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 + +文本生成、深度思考、视觉理解 + +2026-04-17 + +国际 + +`qwen3.6-flash` + +`qwen3.6-flash-2026-04-16` + +Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。 + +文本生成、深度思考、视觉理解 + +2026-04-17 + +国际 + +`qwen3.6-35b-a3b` + +Qwen3.6系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果相较3.5-35B-A3B显著提升了agentic coding能力、数学推理和代码推理能力、空间智能能力、物体定位与目标检测能力。 + +语音合成 + +2026-04-15 + +国际 + +`voice-enrollment` + +大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。 大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。 + +文本生成、深度思考 + +2026-04-14 + +国际 + +`qwen3.6-max-preview` + +Qwen3.6系列中规模最大、综合能力最强的Max模型Preview版本，当前开放纯文本模型能力供体验。相较于此前发布的Qwen3-Max和Qwen3.6-Plus，本模型在vibe coding能力上进一步提升、coding agent执行更加高效、前端编程开发能力显著提升；长尾知识能力进一步升级。 + +文本生成、深度思考 + +2026-04-14 + +国际 + +`glm-5.1` + +GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 + +视频生成 + +2026-04-03 + +国际 + +`wan2.7-videoedit` + +Wan2.7-VideoEdit，自然语言指令编辑视频，支持局部或全局编辑，可参考图像替换视频元素，支持复刻视频动作、特效、运镜等动态过程。 + +视频生成 + +2026-04-03 + +国际 + +`wan2.7-t2v` + +Wan2.7-T2V，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。 + +视频生成 + +2026-04-03 + +国际 + +`wan2.7-r2v` + +Wan2.7-R2V，更加稳定的角色、道具与场景参考，支持最大5个图/视频混合参考，支持音频音色参考，搭配基础能力升级实现更强表演能力。 + +视频生成 + +2026-04-03 + +国际 + +`wan2.7-i2v` + +万相2.7-图生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。 + +图片生成 + +2026-04-01 + +国际 + +`wan2.7-image-pro` + +万相2.7-图像生成与编辑旗舰版模型，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现。 + +图片生成 + +2026-04-01 + +国际 + +`wan2.7-image` + +万相2.7-图像生成与编辑，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现 + +文本生成、深度思考、视觉理解 + +2026-04-01 国际 -fun-asr、fun-asr-2025-11-07 +`qwen3.6-plus` + +`qwen3.6-plus-2026-04-02` + +Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。 + +实时全模态 + +2026-03-26 + +国际 -录音文件识别-Fun-ASR功能更新： +`qwen3.5-omni-plus-realtime` -- 支持歌唱识别，能实现整首歌曲的转写，详情请参见[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition)。 - +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。 全模态 -2025-12-04 +2026-03-26 国际 -qwen3-omni-flash-2025-12-01 +`qwen3.5-omni-plus` -千问Omni发布的最新快照模型，支持的音色增加至49种，模型的指令跟随能力大幅升级，能高效理解文本、图像、音频、视频。[非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni) +`qwen3.5-omni-plus-2026-03-15` -实时多模态 +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。 -2025-12-04 +实时全模态 + +2026-03-26 国际 -qwen3-omni-flash-realtime**\-**2025-12-01 +`qwen3.5-omni-flash-realtime` -千问Omni 实时版发布的最新快照模型，提供了低延迟的多模态交互能力，支持的音色增加至49种，模型的指令跟随能力和交互体验大幅升级。[实时（Qwen-Omni-Realtime）](https://help.aliyun.com/zh/model-studio/realtime) +`qwen3.5-omni-flash-realtime-2026-03-15` -语音翻译 +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。 -2025-12-04 +全模态 + +2026-03-26 国际 -qwen3-livetranslate-flash、qwen3-livetranslate-flash-2025-12-01 +`qwen3.5-omni-flash` -千问3-LiveTranslate-Flash 是音视频翻译模型，支持 18 种语言（包括中文、英文、俄文、法文等）互译，可结合视觉上下文提升翻译准确性，并输出文本与语音。[音视频文件翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-livetranslate-flash) +`qwen3.5-omni-flash-2026-03-15` -多语言翻译 +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。 -2025-12-02 +实时全模态 + +2026-03-25 国际 -qwen-mt-lite +`qwen3.5-omni-plus-realtime-2026-03-15` -千问基础级文本翻译大模型，支持31个语种互译，相较于qwen-mt-flash响应更快，成本更低，适用于等对延迟敏感的场景。[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation) +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。该版本为2026年3月15日快照。 -声音复刻 +文本生成、深度思考 -2025-11-27 +2026-03-20 + +国际 + +`deepseek-v3.2` + +DeepSeek-V3.2是引入DeepSeek Sparse Attention（一种稀疏注意力机制）的正式版模型，也是DeepSeek推出的首个将思考融入工具使用的模型，同时支持思考模式与非思考模式的工具调用。 + +图片生成 + +2026-03-03 + +国际 + +`qwen-image-2.0-pro-2026-03-03` + +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。该版本为2026年3月3日快照。 + +图片生成 + +2026-03-03 + +国际 + +`qwen-image-2.0` + +`qwen-image-2.0-2026-03-03` + +Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。 + +语音识别 + +2026-03-02 + +国际 + +`qwen3-asr-flash-2026-02-10` + +通义千问3-ASR-Flash，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2026年2月10日的快照版本。 + +文本生成、深度思考、视觉理解 + +2026-02-23 + +国际 + +`qwen3.5-flash` + +`qwen3.5-flash-2026-02-23` + +Qwen3.5原生视觉语言系列Flash模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步；响应速度快，兼具推理速度和性能。 + +文本生成、深度思考、视觉理解 + +2026-02-23 + +国际 + +`qwen3.5-35b-a3b` + +Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。 + +文本生成、深度思考、视觉理解 + +2026-02-23 + +国际 + +`qwen3.5-27b` + +Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。 + +文本生成、深度思考、视觉理解 + +2026-02-23 + +国际 + +`qwen3.5-122b-a10b` + +Qwen3.5系列122B-A10B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现仅次于Qwen3.5-397B-A17B，文本能力显著优于Qwen3-235B-2507，视觉能力优于Qwen3-VL-235B。 + +文本生成 + +2026-02-20 + +国际 + +`qwen3-coder-next` + +Qwen3系列新一代代码生成模型，效果接近Qwen3-Coder-Plus兼具更优性能。模型重点优化仓库级别理解、支持多轮工具交互、提升对于agentic coding类工具的适配能力。 + +文本生成、深度思考、视觉理解 + +2026-02-15 + +国际 + +`qwen3.5-plus` + +`qwen3.5-plus-2026-02-15` + +Qwen3.5原生视觉语言系列Plus模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在多项任务评测中，3.5系列均展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 + +文本生成、深度思考、视觉理解 + +2026-02-15 + +国际 + +`qwen3.5-397b-a17b` + +Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。 + +语音识别 + +2026-02-13 国际 -qwen-voice-enrollment +`qwen3-asr-flash-realtime-2026-02-10` -千问发布声音复刻模型，仅需 5 秒以上音频即可快速生成高相似度声音。结合qwen3-tts-vc-realtime-2025-11-27模型使用，可高保真复刻并实时输出某人的声音，覆盖 11 种语言。[声音复刻（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning) +Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2026年2月10日的快照版本。 语音合成 -2025-11-27 +2026-02-10 国际 -qwen3-tts-vc-realtime-2025-11-27（快照版） +`qwen3-tts-vd-2026-01-26` -千问实时语音合成发布全新快照版模型，可使用[声音复刻（Qwen）](https://help.aliyun.com/zh/model-studio/qwen-tts-voice-cloning)生成的音色进行低延迟、高稳定性的实时合成；支持多语言输出；能根据文本自动调节语气，并优化复杂文本的合成表现。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +Qwen3-TTS-VD模型是通义最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月26日快照版本模型。 语音合成 -2025-11-27 +2026-02-10 国际 -qwen3-tts-flash-realtime-2025-11-27（快照版） +`qwen3-tts-vc-2026-01-22` -千问实时语音合成发布全新快照版模型，低延迟且稳定性高；音色更丰富，同一音色支持多语言输出；能根据文本自动调节语气，并提升复杂文本的合成表现。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +Qwen3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月22日快照版本模型。 语音合成 -2025-11-27 +2026-02-10 国际 -qwen3-tts-flash-2025-11-27（快照版） +`qwen3-tts-instruct-flash` -千问语音合成发布全新快照版模型，音色更丰富；同一音色支持多语言输出；可自适应文本调节语气，并优化复杂文本的合成能力。[非实时语音合成](https://help.aliyun.com/zh/model-studio/non-realtime-tts-user-guide) +`qwen3-tts-instruct-flash-2026-01-26` -文字提取 +Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。 -2025-11-21 +语音合成 + +2026-02-10 国际 -qwen-vl-ocr-2025-11-20（快照版） +`cosyvoice-v3-plus` -千问文字提取模型，该快照版基于Qwen3-VL架构，大幅提升文档解析、文字定位能力。[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr) +克隆能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音克隆大模型，具有更好的音质和复刻相似度，适用于更专业的场景。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。合成能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音合成大模型，具有更好的音质和表现力，适用于更专业的场景。该模型支持文本至语音的实时流式合成。 -语音识别 +语音合成 -2025-11-20 +2026-02-09 国际 -qwen3-asr-flash-filetrans、qwen3-asr-flash-filetrans-2025-11-17（快照版） +`cosyvoice-v3-flash` -千问录音文件识别发布了新模型，专为音频文件的异步转写设计，支持最长12小时录音。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +合成能力：CosyVoice-v3-Flash是通义实验室CosyVoice系列最新版高性能的语音合成大模型，较之前版本在自然度、音质、韵律、情感表现力上有更好的表现。该模型支持文本至语音的实时流式合成。克隆能力：CosyVoice-v3-Flash也是通义实验室CosyVoice系列最新版的语音克隆大模型，较之前版本提升了发音准确性、音色相似度，并且增加了更多小语种支持（德、西、法、意、俄）。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。 -语音识别 +文本生成 -2025-11-19 +2026-01-30 国际 -fun-asr-2025-11-07（快照版） +`qwen-plus-character` -Fun-ASR录音文件识别发布了全新快照版模型，优化远场语音活动检测（VAD）以提升识别准确率与稳定性，并在原有中英文识别基础上新增支持中文多地方言及日语。[录音文件识别-Fun-ASR/Paraformer](https://help.aliyun.com/zh/model-studio/recording-file-recognition) +千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 -多语言翻译 +视频生成 -2025-11-11 +2026-01-29 + +国际 + +`wan2.6-r2v-flash` + +万相2.6-参考生视频-Flash，生成更快性价比更高。支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍 + +文本向量 + +2026-01-27 国际 -qwen-mt-flash +`qwen3-rerank` -相较于qwen-mt-turbo支持流式增量输出，整体性能表现有所提升。[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation) +基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐开源Qwen3-Rerank系列模型 -图生视频 +文本生成、深度思考 -2025-11-10 +2026-01-23 + +国际 + +`qwen3-max-2026-01-23` + +千问3系列Max模型，相较2025年9月23日快照，此版本实现思考模式和非思考模式的有效融合，模型整体效果得到全方位的大幅度提升。在思考模式下，同时发布Web搜索、Web信息提取和代码解释器工具能力，使得模型在慢思考的同时，能够通过引入外部工具，以更高的准确性解决更有难度的问题。此版本为2026年1月23日快照。 + +视觉理解 + +2026-01-22 + +国际 + +`qwen3-vl-flash-2026-01-22` + +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于2025年10月15日快照，模型整体效果有大幅提升：在通用视觉识别与推理方面有增强；在安防、巡店、巡检、拍照解题等业务场景识别效果上提升显著。此版本为2026年1月22日快照版本。 + +语音合成 + +2026-01-21 + +国际 + +`qwen3-tts-instruct-flash-realtime` + +`qwen3-tts-instruct-flash-realtime-2026-01-22` + +千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。 + +视频生成 + +2026-01-15 + +国际 + +`wan2.6-i2v-flash` + +万相2.6-图生视频-Flash，生成更快更高性价比。智能分镜调度支持多镜头叙事，多人稳定对话，更自然真实音色，最高支持15秒时长生成 + +图片生成 + +2026-01-15 + +国际 + +`qwen-image-edit-max` + +`qwen-image-edit-max-2026-01-16` + +千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。 + +语音合成 + +2026-01-14 + +国际 + +`qwen3-tts-vd-realtime-2026-01-15` + +千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。 + +语音合成 + +2026-01-14 + +国际 + +`qwen3-tts-vc-realtime-2026-01-15` + +千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。 + +文本生成 + +2026-01-13 + +国际 + +`qwen-flash-character` + +千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 + +图片生成 + +2026-01-09 + +国际 + +`qwen-image-plus-2026-01-09` + +千问系列图像生成模型，具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色。此版本为2026年1月9日快照，为Qwen-Image-Max的蒸馏加速版，可以更快速地生成高质量图片。 + +图片生成 + +2025-12-30 + +国际 + +`qwen-image-max` + +`qwen-image-max-2025-12-30` + +千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。 + +图片生成 + +2025-12-22 + +国际 + +`z-image-turbo` + +Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。 + +深度思考、视觉理解 + +2025-12-18 + +国际 + +`qwen3-vl-plus-2025-12-19` + +Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合。相较于9月23日快照，在推理及分析任务、风格控制上表现更优；同时拥有更低的延时和更快的响应速度。此版本为2025年12月19日快照版本。 + +视频生成 + +2025-12-16 + +国际 + +`wan2.6-r2v` + +万相2.6-参考生视频，支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍。提醒：当使用视频进行参考时，输入视频也会计入费用，详见模型计费文档。 + +图片生成 + +2025-12-15 + +国际 + +`wan2.6-t2i` + +万相2.6-文生图，画面质感、美学表现、指令遵循升级，在艺术风格精准控制、真实感人像、长文本生图及广泛历史文化IP覆盖上均表现出卓越能力，可生成高质量且富有表现力的视觉内容。 + +图片生成 + +2025-12-15 + +国际 + +`wan2.6-image` + +万相2.6-图像生成，全能图像生成模型，支持图文一体化推理生成，具备多图创意融合、商用级一致性、美学要素迁移与镜头光影精确控制，全面提升图像生成的一致性、可控性和表现力。 + +图片生成 + +2025-12-15 + +国际 + +`qwen-image-edit-plus-2025-12-15` + +千问系列图像编辑Plus模型，相较10月30日快照提升角色一致性、工业设计能力、几何推理能力；同时集成例如打光等Lora能力、减轻偏移问题。此版本为2025年12月15日快照。 + +语音合成 + +2025-12-12 + +国际 + +`qwen3-tts-vd-realtime-2025-12-16` + +千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年12月16日快照版本模型。 + +语音合成 + +2025-12-12 + +国际 + +`qwen-voice-design` + +千问voice-design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 + +实时全模态 + +2025-12-04 + +国际 + +`qwen3-omni-flash-realtime` + +`qwen3-omni-flash-realtime-2025-12-01` + +千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。 + +全模态 + +2025-12-04 + +国际 + +`qwen3-omni-flash` + +`qwen3-omni-flash-2025-12-01` + +千问3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。 + +实时语音识别 + +2025-12-04 + +国际 + +`qwen3-livetranslate-flash` + +`qwen3-livetranslate-flash-2025-12-01` + +Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。 + +视频生成 + +2025-12-03 + +国际 + +`wan2.6-t2v` + +万相2.6-文生视频，智能分镜调度支持多镜头叙事，能够生成主体、场景和氛围一致的多镜头叙事视频，最高支持15秒时长，更高品质的声音生成，更好的指令遵循和视觉质量 + +视频生成 + +2025-12-03 + +国际 + +`wan2.6-i2v` + +万相2.6-图生视频，智能分镜调度支持多镜头叙事，更高品质的声音生成，多人稳定对话，更自然真实音色，最高支持15秒时长生成 + +— + +2025-12-01 + +国际 + +`qwen-plus-2025-12-01` + +本版本为2025年12月1日快照，相较7月28日快照在推理能力上有提升；智能体能力、多轮工具调用能力进一步增强；主观创作类任务表现更优。支持1M上下文长度，按照上下文长度进行阶梯计费。 + +语音合成 + +2025-11-27 + +国际 + +`qwen3-tts-vc-realtime-2025-11-27` + +千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年11月27日快照版本模型。 + +语音合成 + +2025-11-27 + +国际 + +`qwen3-tts-flash-realtime` + +`qwen3-tts-flash-realtime-2025-11-27` + +千问3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 + +语音合成 + +2025-11-27 + +国际 + +`qwen3-tts-flash` + +`qwen3-tts-flash-2025-11-27` + +Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 + +语音合成 + +2025-11-27 + +国际 + +`qwen-voice-enrollment` + +千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 + +视觉理解 + +2025-11-21 + +国际 + +`qwen-vl-ocr-2025-11-20` + +本模型为2025年11月20日的快照版本，基于最新Qwen-VL3架构全面升级，在文档解析，文字定位能力全面升级，端到端时延，幻觉大幅降低。 + +语音识别 + +2025-11-21 + +国际 + +`fun-asr-2025-11-07` + +百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本为2025年11月7日的快照版本。 + +视觉理解 + +2025-11-20 + +国际 + +`qwen-vl-ocr` + +千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。 + +文本生成 + +2025-11-19 + +国际 + +`qwen-mt-lite` + +基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + +语音识别 + +2025-11-18 + +国际 + +`qwen3-asr-flash-filetrans` + +`qwen3-asr-flash-filetrans-2025-11-17` + +千问3-ASR-Flash的大文件转录版本，千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。 + +实时语音识别 + +2025-11-17 + +国际 + +`fun-asr-realtime-2025-11-07` + +通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。此版本为2025年11月7日的快照版本。 + +文本生成 + +2025-11-11 + +国际 + +`qwen-mt-flash` + +基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + +视频生成 + +2025-11-10 + +国际 + +`wan2.2-animate-move` + +wan2.2-animate-move图生动作是一款角色动画生成模型，用户只需上传一张角色照片和一段参考表演视频，即可将视频中的表情和动作迁移到图片角色上，生成高保真的动画视频。 + +视频生成 + +2025-11-10 + +国际 + +`wan2.2-animate-mix` + +wan2.2-animate-mix视频换人是一款角色替换的模型产品，上传一张角色照片与一段表演视频，即可将原视频中的角色精准替换为照片中的角色，完整保留原始视频的场景、光照和色调等环境细节。 + +多模态向量 + +2025-10-31 + +国际 + +`tongyi-embedding-vision-plus` + +Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。 + +多模态向量 + +2025-10-31 + +国际 + +`tongyi-embedding-vision-flash` + +Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，在视觉向量化上具备极高性价比。 + +图片生成 + +2025-10-31 + +国际 + +`qwen-image-edit-plus` + +`qwen-image-edit-plus-2025-10-30` + +千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。 + +实时语音识别 + +2025-10-29 + +国际 + +`qwen3-asr-flash-realtime` + +`qwen3-asr-flash-realtime-2025-10-27` + +千问3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。 + +深度思考、视觉理解 + +2025-10-21 + +国际 + +`qwen3-vl-32b-thinking` + +Qwen3-VL系列最大尺寸Dense模型的推理版本，多模态推理能力仅次于Qwen3-VL-235B-Thinking，STEM&数学类解题能力、通用图像和视频理解能力出众，多模态Agent能力达到SOTA，适合做复杂多模态推理任务。 + +视觉理解 + +2025-10-21 + +国际 + +`qwen3-vl-32b-instruct` + +Qwen3-VL系列最大尺寸Dense模型的非推理版本，综合表现仅次于Qwen3-VL-235B-Instruct，文档识别和理解能力出色，空间感知与万物识别能力强，视觉2D检测/空间推理能力达到SOTA，适合通用场景下的复杂感知任务。 + +深度思考、视觉理解 + +2025-10-15 + +国际 + +`qwen3-vl-flash` + +`qwen3-vl-flash-2025-10-15` + +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。 + +深度思考、视觉理解 + +2025-10-03 + +国际 + +`qwen3-vl-30b-a3b-thinking` + +Qwen3-VL系列第二大MoE模型的Thinking版本，响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级图像/视频理解、空间感知与万物识别能力，胜任复杂现实任务。 + +视觉理解 + +2025-10-03 + +国际 + +`qwen3-vl-30b-a3b-instruct` + +Qwen3-VL系列第二大MoE模型的Instruct版本，响应速度快，支持长视频长文档等超长上下文；全面升级图像/视频理解、空间感知与万物识别能力；具备视觉2D/3D定位能力，胜任复杂现实任务。 + +深度思考、视觉理解 + +2025-09-30 + +国际 + +`qwen3-vl-8b-thinking` + +Qwen3-VL系列8B Dense模型的Thinking版本，占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面升级图像/视频理解、空间感知与万物识别能力。 + +视觉理解 + +2025-09-30 + +国际 + +`qwen3-vl-8b-instruct` + +Qwen3-VL系列8B Dense模型的Instruct版本，占用显存更低，全面升级图像/视频理解、长视频长文档等超长上下文支持、空间感知与万物识别能力，胜任复杂现实任务。 + +语音识别 + +2025-09-25 + +国际 + +`fun-asr-mtl` + +`fun-asr-mtl-2025-08-25` + +百聆多语言语音识别大模型，支持超过31种语言，支持语种自由切换，出海用户首推，尤其东南亚出海。fun-asr为该模型的升级版本，建议切换使用fun-asr。 + +图片生成 + +2025-09-24 + +国际 + +`wan2.5-t2i-preview` + +万相2.5-文生图-Preview，全新升级模型架构。画面美学、设计感、真实质感显著提升，精准指令遵循，擅长中英文和小语种文字生成，支持复杂结构化长文本和图表、架构图等内容生成。 + +图片生成 + +2025-09-24 + +国际 + +`wan2.5-i2i-preview` + +万相2.5-图像编辑-Preview，全新升级模型架构。支持指令控制实现丰富的图像编辑能力，指令遵循能力进一步提升，支持高一致性保持的多图参考生成，文字生成表现优异。 + +文本生成、深度思考 + +2025-09-24 + +国际 + +`qwen3-max` + +`qwen3-max-2025-09-23` + +千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。 + +实时语音翻译 + +2025-09-24 + +国际 + +`qwen3-livetranslate-flash-realtime` + +Qwen3-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。 + +语音识别 + +2025-09-24 + +国际 + +`fun-asr` + +`fun-asr-2025-08-25` + +百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本等同于2025年11月7日的快照版本。 + +视频生成 + +2025-09-23 + +国际 + +`wan2.5-t2v-preview` + +万相2.5-文生视频-Preview，全新升级模型架构，支持与画面同步的声音生成，支持10秒长视频生成，更强的指令遵循能力，运动能力、画面质感进一步提升。 + +视频生成 + +2025-09-23 + +国际 + +`wan2.5-i2v-preview` + +万相2.5-图生视频-Preview，全新升级技术架构，支持与画面同步的声音生成，支持10秒长视频生成，更强的指令遵循能力，运动能力、画面质感进一步提升。 + +视觉理解 + +2025-09-23 + +国际 + +`qwen3-vl-plus` + +`qwen3-vl-plus-2025-09-23` + +Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。 + +深度思考、视觉理解 + +2025-09-23 + +国际 + +`qwen3-vl-235b-a22b-thinking` + +Qwen3系列视觉理解模型，多模态思考能力显著增强，模型在STEM与数学推理方面进行了重点优化；视觉感知与识别能力全面提升、OCR能力迎来重大升级。 + +视觉理解 + +2025-09-23 + +国际 + +`qwen3-vl-235b-a22b-instruct` + +Qwen3系列视觉理解模型，在视觉coding、空间感知等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解，OCR能力迎来重大升级。 + +实时语音翻译 + +2025-09-23 + +国际 + +`qwen3-livetranslate-flash-realtime-2025-09-22` + +千问3-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，千问3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。此版本为2025年9月22日的快照版本。 + +文本生成 + +2025-09-23 + +国际 + +`qwen3-coder-plus-2025-09-23` + +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。本版本为2025年9月23日快照，相较上一版本（7月22日快照）在下游任务效果和工具调用方面鲁棒性有所提升；代码安全性增强。 + +图片生成 + +2025-09-23 + +国际 + +`qwen-image-plus` + +千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。 + +实时语音识别 + +2025-09-23 + +国际 + +`fun-asr-realtime` + +通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。 + +语音识别 + +2025-09-19 + +国际 + +`qwen3-omni-30b-a3b-captioner` + +千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。 + +实时全模态 + +2025-09-17 + +国际 + +`qwen3-omni-flash-realtime-2025-09-15` + +千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。此版本为2025年9月15日的快照版本。 + +全模态 + +2025-09-17 + +国际 + +`qwen3-omni-flash-2025-09-15` + +Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。此版本为2025年9月15日的快照版本。 + +语音合成 + +2025-09-16 + +国际 + +`qwen3-tts-flash-realtime-2025-09-18` + +Qwen3-TTS-Flash-Realtime-2025-09-18模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年9月18日快照版本模型。 + +语音合成 + +2025-09-16 + +国际 + +`qwen3-tts-flash-2025-09-18` + +千问3-TTS-Flash-2025-09-18模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年9月18日快照版本模型。 + +语音识别 + +2025-09-16 + +国际 + +`qwen3-asr-flash` + +`qwen3-asr-flash-2025-09-08` + +千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。 + +视觉理解 + +2025-09-16 + +国际 + +`qwen-vl-plus` + +千问VL-Plus（qwen-vl-plus），即千问大规模视觉语言模型增强版。大幅提升细节识别能力和文字识别能力，支持超百万像素分辨率和任意长宽比规格的图像。在广泛的视觉任务上提供卓越的性能。 + +视觉理解 + +2025-09-16 + +国际 + +`qwen-vl-max` + +千问VL-Max（qwen-vl-max），即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。 + +文本生成、深度思考 + +2025-09-16 + +国际 + +`qwen-plus` + +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-Plus，达到同规模业界SOTA水平。 + +视觉理解 + +2025-09-12 + +国际 + +`wan2.2-kf2v-flash` + +全新升级的万相2.2-首尾帧生视频，生成速度更快。优化视频动态稳定性与成功率，更强大的指令遵循能力，两张图片生成丝滑过度视频。 + +文本生成、深度思考 + +2025-09-11 + +国际 + +`qwen3-next-80b-a3b-thinking` + +基于Qwen3的新一代思考模式开源模型，相较上一版本（千问3-235B-A22B-Thinking-2507指令遵循能力有提升、模型总结回复更加精简。 + +文本生成 + +2025-09-11 + +国际 + +`qwen3-next-80b-a3b-instruct` + +基于Qwen3的新一代非思考模式开源模型，相较上一版本（千问3-235B-A22B-Instruct-2507）中文文本理解能力更佳、逻辑推理能力有增强、文本生成类任务表现更好。 + +文本生成、深度思考 + +2025-09-11 + +国际 + +`qwen-plus-2025-09-11` + +本版本为2025年9月11日快照，相较7月28日快照在思考模式下指令遵循能力有提升、模型总结回复更加精简；在非思考模式下中文文本理解能力更佳、逻辑推理能力有增强。支持1M上下文长度，按照上下文长度进行阶梯计费。 + +文本生成、深度思考 + +2025-09-05 + +国际 + +`qwen3-max-preview` + +Qwen3系列Max模型Preview版本，实现思考模式和非思考模式的有效融合。思考模式下在智能体编程能力、常识知识推理能力、数学/科学/通用类推理等能力上均有显著增强。 + +文本向量 + +2025-08-25 + +国际 + +`text-embedding-v4` + +通用文本向量V4版本，是通义实验室基于Qwen3训练的多语言文本统一向量模型，相较V3版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升15%~40%；支持64~2048维用户自定义向量维度。 + +图片生成 + +2025-08-18 + +国际 + +`qwen-image-edit` + +千问系列首个图像编辑模型，成功将Qwen-Image的文本渲染能力拓展到编辑任务上。支持精准的中英双语文字编辑、视觉外观与语义双重编辑、具备强大的跨基准性能表现。 + +视频生成 + +2025-08-15 + +国际 + +`wan2.2-i2v-flash` + +全新升级的万相2.2图生视频，生成速度更快。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。 + +实时全模态 + +2025-08-14 + +国际 + +`qwen-omni-turbo-realtime` + +`qwen-omni-turbo-realtime-latest` + +`qwen-omni-turbo-realtime-2025-05-08` + +千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。 + +图片生成 + +2025-08-14 + +国际 + +`qwen-image` + +千问系列首个图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA。 + +文本生成、深度思考 + +2025-08-01 + +国际 + +`qwen-flash` + +`qwen-flash-2025-07-28` + +Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。 + +文本生成 + +2025-07-31 + +国际 + +`qwen3-coder-30b-a3b-instruct` + +基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。 + +文本生成、深度思考 + +2025-07-31 + +国际 + +`qwen-plus-latest` + +千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。本模型是动态更新版本，模型更新不会提前通知。 + +文本生成、深度思考 + +2025-07-30 + +国际 + +`qwen3-30b-a3b-thinking-2507` + +基于Qwen3的思考模式开源模型，相较上一版本（千问3-30B-A3B）复杂推理类任务性能优秀，包括逻辑推理、数学、科学、代码类等具有一定难度的任务场景，指令遵循、文本理解、多语言翻译等能力显著提高。 + +文本生成 + +2025-07-29 + +国际 + +`qwen3-coder-flash` + +`qwen3-coder-flash-2025-07-28` + +基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。 + +文本生成 + +2025-07-29 + +国际 + +`qwen3-30b-a3b-instruct-2507` + +基于Qwen3的非思考模式开源模型，相较上一版本（千问3-30B-A3B）中英文和多语言整体通用能力有大幅提升。主观开放类任务专项优化，显著更加符合用户偏好，能够提供更有帮助性的回复。 + +文本生成、深度思考 + +2025-07-29 + +国际 + +`qwen-plus-2025-07-28` + +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。相较上个版本在中英文能力、工具调用上进行了专用增强。本版本为2025年7月28日快照，首次支持1M上下文长度，按照上下文长度进行阶梯计费。 + +视频生成 + +2025-07-28 + +国际 + +`wan2.2-t2v-plus` + +全新升级的万相2.2文生视频，视频品质更高。可稳定生成大幅度复杂运动，支持影视级画面表现与控制，更强大的指令遵循能力，实现物理世界还原。 + +图片生成 + +2025-07-28 + +国际 + +`wan2.2-t2i-plus` + +全新升级的万相2.2文生图，更丰富的画面细节。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。 + +图片生成 + +2025-07-28 + +国际 + +`wan2.2-t2i-flash` + +全新升级的万相2.2文生图，更快的生成速度。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。 + +视频生成 + +2025-07-28 + +国际 + +`wan2.2-i2v-plus` + +全新升级的万相2.2图生视频，视频品质更高。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。 + +文本生成、深度思考 + +2025-07-25 + +国际 + +`qwen3-235b-a22b-thinking-2507` + +基于Qwen3的思考模式开源模型，相较上一版本（千问3-235B-A22B）逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景。 + +文本生成 + +2025-07-24 + +国际 + +`qwen-mt-turbo` + +基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + +文本生成 + +2025-07-24 + +国际 + +`qwen-mt-plus` + +基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + +文本生成 + +2025-07-23 + +国际 + +`qwen3-coder-plus` + +`qwen3-coder-plus-2025-07-22` + +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。 + +文本生成 + +2025-07-23 + +国际 + +`qwen3-coder-480b-a35b-instruct` + +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。 + +文本生成 + +2025-07-23 + +国际 + +`qwen3-235b-a22b-instruct-2507` + +基于Qwen3的非思考模式开源模型，相较上一版本（千问3-235B-A22B）主观创作能力与模型安全性均有小幅度提升。 + +文本生成、深度思考 + +2025-07-23 + +国际 + +`qwen-plus-2025-07-14` + +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。此版本为2025年7月14日快照模型，相较上个版本在非thinking模式下中英文能力均有大幅提升，工具调用能力专项增强。 + +文本生成 + +2025-07-21 + +国际 + +`qwen-plus-character-ja` + +千问系列角色扮演模型，针对日语拟人化交互场景专项优化。在角色一致性保持、上下文感知的话题推进、倾听共情等方面表现出色，可精准还原个性化角色。本版本显著增强了日语本地化表达（含方言与敬语）、拟人化角色扮演真实度、叙事连贯性控制以及场景化认知能力。 + +全模态 + +2025-07-18 + +国际 + +`qwen-omni-turbo` + +`qwen-omni-turbo-latest` + +`qwen-omni-turbo-2025-03-26` + +千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 + +图片生成 + +2025-05-22 + +国际 + +`wan2.1-t2i-turbo` + +万相2.1-文生图Turbo版，生成速度更快。在图像美感、真实感与艺术性上全面升级，语义理解能力更强，风格泛化能力丰富，支持最高200万像素生成，支持智能改写提示词。 + +图片生成 + +2025-05-22 + +国际 + +`wan2.1-t2i-plus` + +万相2.1-文生图Plus版，生成图像细节更丰富。在图像美感、真实感与艺术性上全面升级，语义理解能力更强，风格泛化能力丰富，支持最高200万像素生成，支持智能改写提示词。 + +视频生成 + +2025-05-14 + +国际 + +`wan2.1-vace-plus` + +万相2.1-VACE-Plus 视频创作与编辑一体化模型。支持局部编辑、视频重绘、背景扩展、时长延展、图像参考等视频编辑与生成任务，并支持通过文本、图像、视频进行多模态条件控制。 + +文本生成、深度思考 + +2025-05-12 + +国际 + +`qwen3-8b` + +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-7B。 + +文本生成、深度思考 + +2025-05-12 + +国际 + +`qwen3-32b` + +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-32B-Instruct，达到同规模业界SOTA水平。 + +文本生成、深度思考 + +2025-05-12 + +国际 + +`qwen3-30b-a3b` + +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-14B，达到同规模业界SOTA水平。 + +文本生成、深度思考 + +2025-05-12 + +国际 + +`qwen3-235b-a22b` + +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-72B-Instruct，达到同规模业界SOTA水平。 + +文本生成、深度思考 + +2025-05-12 + +国际 + +`qwen3-14b` + +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-14B。 + +文本生成、深度思考 + +2025-04-29 + +国际 + +`qwen-plus-2025-04-28` + +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-Plus，达到同规模业界SOTA水平。此版本为2025年4月28日快照模型。 + +视频生成 + +2025-04-07 + +国际 + +`wan2.1-kf2v-plus` + +万相2.1-首尾帧生视频Plus版，为两张图片生成流畅的过渡视频。支持大幅度复杂运动、遵循物理规律、丰富的艺术风格与影视级画面质感，指令遵循能力进一步增强，生成视频细节更丰富。 + +全模态 + +2025-03-26 + +国际 + +`qwen2.5-omni-7b` + +基于Qwen2.5训练的全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 + +深度思考、视觉理解 + +2025-03-26 + +国际 + +`qvq-max` + +千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。 + +深度思考 + +2025-03-05 + +国际 + +`qwq-plus` + +千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。 + +视频生成 + +2025-02-27 + +国际 + +`wan2.1-i2v-turbo` + +万相2.1-图生视频Turbo版，让静态图片生成视频。支持大幅度复杂运动、遵循物理规律、电影级艺术风格与画面质感，指令遵循能力进一步提升，生成速度更快。 + +文本生成、深度思考 + +2025-01-30 + +国际 + +`qwen-turbo` + +Qwen3系列Turbo模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-Turbo，达到同规模业界SOTA水平。 + +文本生成 + +2025-01-30 + +国际 + +`qwen-plus-2025-01-25` + +千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。相对于通义千问-Plus-2025-0112版本，整体中英文能力都有综合能力升级，中英文code能力、逻辑能力、多语言能力显著提升，回复风格面向人类偏好进行大幅调整，尤其是数学、逻辑推理、知识问答等客观类query，模型回复详实程度和格式清晰度明显改善，创作类专项、json格式遵循专项、角色扮演专项能力均定向提升，预计维护至下一个快照上线前一个月。 + +文本生成 + +2025-01-30 + +国际 + +`qwen-max` + +千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。 + +视频生成 + +2025-01-20 + +国际 + +`wan2.1-i2v-plus` + +万相2.1-图生视频Plus版，让静态图片生成视频。支持大幅度复杂运动、遵循物理规律、电影级艺术风格与画面质感，指令遵循能力进一步提升，视频质量更佳。 + +视频生成 + +2025-01-09 + +国际 + +`wan2.1-t2v-turbo` + +万相2.1-文生视频Turbo版，一句话生成视频。支持大幅度复杂运动、遵循物理规律、电影级艺术风格与画面质感，指令遵循能力进一步提升，生成速度更快。 + +视频生成 + +2025-01-09 + +国际 + +`wan2.1-t2v-plus` + +万相2.1-文生视频Plus版，一句话生成视频。支持大幅度复杂运动、遵循物理规律、电影级艺术风格与画面质感，指令遵循能力进一步提升，视频质量更佳。 + +文本向量 + +2024-07-12 + +国际 + +`text-embedding-v3` + +通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。 + +## 美国（弗吉尼亚） + +**模型类型** + +**时间** + +**服务部署范围** + +**模型ID** + +**功能说明** + +文本生成、深度思考 + +2026-07-07 + +美国 + +`glm-5.2-us` + +GLM-5.2-US是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 + +文本生成、深度思考 + +2026-07-07 + +全球 + +`glm-5.2-us` + +GLM-5.2-US是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 + +文本生成、深度思考、视觉理解 + +2026-07-03 + +美国 + +`qwen3.6-flash-us` + +Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。 + +文本生成、深度思考、视觉理解 + +2026-07-03 + +全球 + +`qwen3.6-flash-us` + +Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。 + +文本生成 + +2026-07-02 + +全球 + +`qwen-plus-character` + +千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 + +文本生成、深度思考、视觉理解 + +2026-06-26 + +美国 + +`qwen3.7-plus-us` + +Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 + +文本生成、深度思考、视觉理解 + +2026-06-26 + +全球 + +`qwen3.7-plus-us` + +Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 + +文本生成、深度思考 + +2026-06-26 + +美国 + +`qwen3.7-max-us` + +Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 + +文本生成、深度思考 + +2026-06-26 + +全球 + +`qwen3.7-max-us` + +Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 + +视频生成 + +2026-06-16 + +全球 + +`happyhorse-1.1-t2v` + +HappyHorse-1.1-T2V支持文生视频，进一步提升文本语义理解、镜头调度与动态生成表现。模型能够更精准地还原创作意图，在人物动作、场景氛围、视觉美感和物理运动上生成更流畅自然、细节丰富且一致性更高的高质量视频。 + +视频生成 + +2026-06-16 + +全球 + +`happyhorse-1.1-r2v` + +HappyHorse-1.1-R2V支持参考生视频，进一步提升主体、场景风格与画面一致性的稳定保持。模型最多支持9张参考图片输入，能够更精准理解并延续创作意图，在人物、场景、风格和镜头表现上带来更强的可控性与表现力。 + +视频生成 + +2026-06-16 + +全球 + +`happyhorse-1.1-i2v` + +HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。 + +文本生成、深度思考 + +2026-06-16 + +全球 + +`glm-5.2` + +GLM-5.2是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 + +文本生成、深度思考、视觉理解 + +2026-06-15 + +全球 + +`kimi-k2.7-code` + +kimi-k2.7-code是 kimi 迄今最智能的coding模型，在长上下文中更可靠地遵循指令，能以更高的成功率完成编程任务，同时支持文本、图片与视频输入，思考模式，对话与 Agent 任务。 + +文本生成、深度思考、视觉理解 + +2026-06-09 + +全球 + +`qwen3.7-max-2026-06-08` + +Qwen3.7系列中规模最大、综合能力最强的Max模型，相较于5月20日快照增加了视觉模态理解能力，能够感知真实世界场景，具备多模态交互混合智能体能力。该版本为2026年6月8日快照。 + +文本生成、深度思考、视觉理解 + +2026-06-01 + +全球 + +`qwen3.7-plus` + +`qwen3.7-plus-2026-05-26` + +Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 + +文本生成、深度思考 + +2026-05-20 + +全球 + +`qwen3.7-max` + +`qwen3.7-max-2026-05-20` + +Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 + +文本生成、深度思考 + +2026-05-11 + +美国 + +`deepseek-v4-pro-us` + +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 + +文本生成、深度思考 + +2026-05-11 + +全球 + +`deepseek-v4-pro-us` + +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 + +文本生成、深度思考 + +2026-05-11 + +美国 + +`deepseek-v4-flash-us` + +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 + +文本生成、深度思考 + +2026-05-11 + +全球 + +`deepseek-v4-flash-us` + +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 + +文本生成、深度思考、视觉理解 + +2026-04-29 + +全球 + +`kimi-k2.5` + +kimi-k2.5是月之暗面迄今发布最全能的模型，原生多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与Agent任务。 + +视频生成 + +2026-04-26 + +全球 + +`happyhorse-1.0-video-edit` + +HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。 + +视频生成 + +2026-04-26 + +全球 + +`happyhorse-1.0-r2v` + +HappyHorse-1.0-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。 + +文本生成、深度思考 + +2026-04-24 + +全球 + +`deepseek-v4-pro` + +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 + +文本生成、深度思考 + +2026-04-24 + +全球 + +`deepseek-v4-flash` + +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 + +视频生成 + +2026-04-22 + +全球 + +`happyhorse-1.0-t2v` + +HappyHorse-1.0-T2V支持文生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 + +视频生成 + +2026-04-22 + +全球 + +`happyhorse-1.0-i2v` + +HappyHorse-1.0-I2V支持图生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 + +文本生成、深度思考、视觉理解 + +2026-04-17 + +全球 + +`qwen3.6-flash` + +`qwen3.6-flash-2026-04-16` + +Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。 + +文本生成、深度思考、视觉理解 + +2026-04-17 + +全球 + +`qwen3.6-35b-a3b` + +Qwen3.6系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果相较3.5-35B-A3B显著提升了agentic coding能力、数学推理和代码推理能力、空间智能能力、物体定位与目标检测能力。 + +文本生成、深度思考 + +2026-04-14 + +全球 + +`glm-5.1` + +GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 + +文本生成、深度思考、视觉理解 + +2026-04-01 + +全球 + +`qwen3.6-plus` + +`qwen3.6-plus-2026-04-02` + +Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。 + +文本生成 + +2026-03-30 + +美国 + +`qwen-mt-lite-us` + +基于Qwen3全面升级的基础级文本翻译大模型，支持31个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + +文本生成 + +2026-03-30 + +全球 + +`qwen-mt-lite-us` + +基于Qwen3全面升级的基础级文本翻译大模型，支持31个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + +视觉理解 + +2026-03-14 + +美国 + +`qwen3-vl-flash-2026-01-22-us` + +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于2025年10月15日快照，模型整体效果有大幅提升：在通用视觉识别与推理方面有增强；在安防、巡店、巡检、拍照解题等业务场景识别效果上提升显著。此版本为2026年1月22日快照版本。 + +视觉理解 + +2026-03-14 + +全球 + +`qwen3-vl-flash-2026-01-22-us` + +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于2025年10月15日快照，模型整体效果有大幅提升：在通用视觉识别与推理方面有增强；在安防、巡店、巡检、拍照解题等业务场景识别效果上提升显著。此版本为2026年1月22日快照版本。 + +文本生成、深度思考、视觉理解 + +2026-02-23 + +全球 + +`qwen3.5-flash` + +`qwen3.5-flash-2026-02-23` + +Qwen3.5原生视觉语言系列Flash模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步；响应速度快，兼具推理速度和性能。 + +文本生成、深度思考、视觉理解 + +2026-02-23 + +全球 + +`qwen3.5-35b-a3b` + +Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。 + +文本生成、深度思考、视觉理解 + +2026-02-23 + +全球 + +`qwen3.5-27b` + +Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。 + +文本生成、深度思考、视觉理解 + +2026-02-23 + +全球 + +`qwen3.5-122b-a10b` + +Qwen3.5系列122B-A10B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现仅次于Qwen3.5-397B-A17B，文本能力显著优于Qwen3-235B-2507，视觉能力优于Qwen3-VL-235B。 + +文本生成、深度思考、视觉理解 + +2026-02-15 + +全球 + +`qwen3.5-plus` + +`qwen3.5-plus-2026-02-15` + +Qwen3.5原生视觉语言系列Plus模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在多项任务评测中，3.5系列均展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 + +文本生成、深度思考、视觉理解 + +2026-02-15 + +全球 + +`qwen3.5-397b-a17b` + +Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。 + +视频生成 + +2025-12-16 + +全球 + +`wan2.6-r2v` + +万相2.6-参考生视频，支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍 + +图片生成 + +2025-12-15 + +全球 + +`wan2.6-t2i` + +万相2.6-文生图，画面质感、美学表现、指令遵循升级，在艺术风格精准控制、真实感人像、长文本生图及广泛历史文化IP覆盖上均表现出卓越能力，可生成高质量且富有表现力的视觉内容。 + +图片生成 + +2025-12-15 + +全球 + +`wan2.6-image` + +万相2.6-图像生成，全能图像生成模型，支持图文一体化推理生成，具备多图创意融合、商用级一致性、美学要素迁移与镜头光影精确控制，全面提升图像生成的一致性、可控性和表现力。 + +视频生成 + +2025-12-03 + +美国 + +`wan2.6-t2v-us` + +万相2.6-文生视频，智能分镜调度支持多镜头叙事，能够生成主体、场景和氛围一致的多镜头叙事视频，最高支持15秒时长，更高品质的声音生成，更好的指令遵循和视觉质量 + +视频生成 + +2025-12-03 + +全球 + +`wan2.6-t2v-us` + +万相2.6-文生视频，智能分镜调度支持多镜头叙事，能够生成主体、场景和氛围一致的多镜头叙事视频，最高支持15秒时长，更高品质的声音生成，更好的指令遵循和视觉质量 + +视频生成 + +2025-12-03 + +全球 + +`wan2.6-t2v` + +万相2.6-文生视频，智能分镜调度支持多镜头叙事，能够生成主体、场景和氛围一致的多镜头叙事视频，最高支持15秒时长，更高品质的声音生成，更好的指令遵循和视觉质量 + +视频生成 + +2025-12-03 + +美国 + +`wan2.6-i2v-us` + +万相2.6-图生视频，智能分镜调度支持多镜头叙事，更高品质的声音生成，多人稳定对话，更自然真实音色，最高支持15秒时长生成 + +视频生成 + +2025-12-03 + +全球 + +`wan2.6-i2v-us` + +万相2.6-图生视频，智能分镜调度支持多镜头叙事，更高品质的声音生成，多人稳定对话，更自然真实音色，最高支持15秒时长生成 + +视频生成 + +2025-12-03 + +全球 + +`wan2.6-i2v` + +万相2.6-图生视频，智能分镜调度支持多镜头叙事，更高品质的声音生成，多人稳定对话，更自然真实音色，最高支持15秒时长生成 + +文本生成、深度思考 + +2025-12-01 + +美国 + +`qwen-plus-2025-12-01-us` + +本版本为2025年12月1日快照，相较7月28日快照在推理能力上有提升；智能体能力、多轮工具调用能力进一步增强；主观创作类任务表现更优。支持1M上下文长度，按照上下文长度进行阶梯计费。 + +文本生成、深度思考 + +2025-12-01 + +全球 + +`qwen-plus-2025-12-01-us` + +本版本为2025年12月1日快照，相较7月28日快照在推理能力上有提升；智能体能力、多轮工具调用能力进一步增强；主观创作类任务表现更优。支持1M上下文长度，按照上下文长度进行阶梯计费。 + +文本生成、深度思考 + +2025-12-01 + +全球 + +`qwen-plus-2025-12-01` + +本版本为2025年12月1日快照，相较7月28日快照在推理能力上有提升；智能体能力、多轮工具调用能力进一步增强；主观创作类任务表现更优。支持1M上下文长度，按照上下文长度进行阶梯计费。 + +视觉理解 + +2025-11-21 + +全球 + +`qwen-vl-ocr-2025-11-20` + +本模型为2025年11月20日的快照版本，基于最新Qwen-VL3架构全面升级，在文档解析，文字定位能力全面升级，端到端时延、幻觉大幅降低。 + +文本生成 + +2025-11-19 + +全球 + +`qwen-mt-lite` + +基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + +文本生成 + +2025-11-11 + +全球 + +`qwen-mt-flash` + +基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + +深度思考、视觉理解 + +2025-10-21 + +全球 + +`qwen3-vl-32b-thinking` + +Qwen3-VL系列最大尺寸Dense模型的推理版本，多模态推理能力仅次于Qwen3-VL-235B-Thinking，STEM&数学类解题能力、通用图像和视频理解能力出众，多模态Agent能力达到SOTA，适合做复杂多模态推理任务。 + +视觉理解 + +2025-10-21 + +全球 + +`qwen3-vl-32b-instruct` + +Qwen3-VL系列最大尺寸Dense模型的非推理版本，综合表现仅次于Qwen3-VL-235B-Instruct，文档识别和理解能力出色，空间感知与万物识别能力强，视觉2D检测/空间推理能力达到SOTA，适合通用场景下的复杂感知任务。 + +视觉理解 + +2025-10-15 + +美国 + +`qwen3-vl-flash-us` + +`qwen3-vl-flash-2025-10-15-us` + +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。 + +视觉理解 + +2025-10-15 + +全球 + +`qwen3-vl-flash-us` + +`qwen3-vl-flash-2025-10-15-us` + +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。 + +深度思考、视觉理解 + +2025-10-15 + +全球 + +`qwen3-vl-flash` + +`qwen3-vl-flash-2025-10-15` + +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。 + +深度思考、视觉理解 + +2025-10-03 + +全球 + +`qwen3-vl-30b-a3b-thinking` + +Qwen3-VL系列第二大MoE模型的Thinking版本，响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级图像/视频理解、空间感知与万物识别能力，胜任复杂现实任务。 + +视觉理解 + +2025-10-03 + +全球 + +`qwen3-vl-30b-a3b-instruct` + +Qwen3-VL系列第二大MoE模型的Instruct版本，响应速度快，支持长视频长文档等超长上下文；全面升级图像/视频理解、空间感知与万物识别能力；具备视觉2D/3D定位能力，胜任复杂现实任务。 + +深度思考、视觉理解 + +2025-09-30 + +全球 + +`qwen3-vl-8b-thinking` + +Qwen3-VL系列8B Dense模型的Thinking版本，占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面升级图像/视频理解、空间感知与万物识别能力。 + +视觉理解 + +2025-09-30 + +全球 + +`qwen3-vl-8b-instruct` + +Qwen3-VL系列8B Dense模型的Instruct版本，占用显存更低，全面升级图像/视频理解、长视频长文档等超长上下文支持、空间感知与万物识别能力，胜任复杂现实任务。 + +文本生成、深度思考 + +2025-09-24 + +全球 + +`qwen3-max` + +`qwen3-max-2025-09-23` + +千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。 + +视觉理解 + +2025-09-23 + +全球 + +`qwen3-vl-plus` + +`qwen3-vl-plus-2025-09-23` + +Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。 + +深度思考、视觉理解 + +2025-09-23 + +全球 + +`qwen3-vl-235b-a22b-thinking` + +Qwen3系列视觉理解模型，多模态思考能力显著增强，模型在STEM与数学推理方面进行了重点优化；视觉感知与识别能力全面提升、OCR能力迎来重大升级。 + +视觉理解 + +2025-09-23 + +全球 + +`qwen3-vl-235b-a22b-instruct` + +Qwen3系列视觉理解模型，在视觉coding、空间感知等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解，OCR能力迎来重大升级。 + +文本生成 + +2025-09-23 + +全球 -国际 +`qwen3-coder-plus-2025-09-23` -wan2.2-animate-move +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。本版本为2025年9月23日快照，相较上一版本（7月22日快照）在下游任务效果和工具调用方面鲁棒性有所提升；代码安全性增强。 -支持将模板视频中角色的动作和表情，迁移至单张静态人物图片上，生成人物动作视频。[万相-图生动作](https://help.aliyun.com/zh/model-studio/wan-animate-move-api) +语音识别 -图生视频 +2025-09-16 -2025-11-10 +美国 -国际 +`qwen3-asr-flash-us` -wan2.2-animate-mix +`qwen3-asr-flash-2025-09-08-us` -能够依据人物图片和参考视频，将视频中的主角替换为图片中的角色，同时保留原视频的场景、光照和色调，实现无缝人物替换。[万相-视频换人](https://help.aliyun.com/zh/model-studio/wan-animate-mix-api) +Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 11 个语种的语音，在复杂的音频环境下能够保证精确转录。 -推理模型 +语音识别 -2025-11-03 +2025-09-16 -国际 +全球 -qwen3-max-preview +`qwen3-asr-flash-us` -qwen3-max-preview 模型的思考模式：在整体推理能力上显著提升，尤其在智能体编程、常识推理，以及数学、科学和通用任务方面表现更优。[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking) +`qwen3-asr-flash-2025-09-08-us` -图像编辑 +Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 11 个语种的语音，在复杂的音频环境下能够保证精确转录。 -2025-10-31 +文本生成、深度思考 -国际 +2025-09-16 -qwen-image-edit-plus、qwen-image-edit-plus-2025-10-30 +美国 -在qwen-image-edit的基础上优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间，且支持单次请求返回多张图片。[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide) +`qwen-plus-us` -推理模型 +千问超大规模语言模型增强版，支持中文英文等不同语言输入。相对于之前版本，中英文code能力、逻辑能力、多语言能力显著提升，回复风格面向人类偏好进行大幅调整，模型回复详实程度和格式清晰度明显改善，创作类专项、json格式遵循专项、角色扮演专项能力定向提升。 -2025-10-31 +文本生成、深度思考 -国际 +2025-09-16 -qwen3-next-80b-a3b-thinking、qwen3-next-80b-a3b-instruct +全球 -基于Qwen3的新一代开源模型，thinking模型相较于qwen3-235b-a22b-thinking-2507提升了指令遵循能力，总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型相较于qwen3-235b-a22b-instruct-2507增强了中文理解、逻辑推理及文本生成能力，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +`qwen-plus-us` -实时语音识别 +千问超大规模语言模型增强版，支持中文英文等不同语言输入。相对于之前版本，中英文code能力、逻辑能力、多语言能力显著提升，回复风格面向人类偏好进行大幅调整，模型回复详实程度和格式清晰度明显改善，创作类专项、json格式遵循专项、角色扮演专项能力定向提升。 -2025-10-27 +文本生成、深度思考 -国际 +2025-09-16 -qwen3-asr-flash-realtime、qwen3-asr-flash-realtime-2025-10-27 +全球 -千问实时语音识别大模型具备自动语种识别功能，可识别 11 种语音类型，并能在复杂音频环境下较为准确地转录。[实时语音识别](https://help.aliyun.com/zh/model-studio/real-time-speech-recognition-user-guide) +`qwen-plus` -文本、图像、视频、语音等 +千问超大规模语言模型增强版，支持中文英文等不同语言输入。相对于之前版本，中英文code能力、逻辑能力、多语言能力显著提升，回复风格面向人类偏好进行大幅调整，模型回复详实程度和格式清晰度明显改善，创作类专项、json格式遵循专项、角色扮演专项能力定向提升。 -2025-09-23 +文本生成、深度思考 -国际 +2025-09-11 -模型列表 +全球 -新加坡地域首次上线。 +`qwen3-next-80b-a3b-thinking` -## 美国（弗吉尼亚） +基于Qwen3的新一代思考模式开源模型，相较上一版本（千问3-235B-A22B-Thinking-2507指令遵循能力有提升、模型总结回复更加精简。 -**模型类型** +文本生成 -**时间** +2025-09-11 -**服务部署范围** +全球 -**模型规格** +`qwen3-next-80b-a3b-instruct` -**功能说明** +基于Qwen3的新一代非思考模式开源模型，相较上一版本（千问3-235B-A22B-Instruct-2507）中文文本理解能力更佳、逻辑推理能力有增强、文本生成类任务表现更好。 -文生文 +文本生成、深度思考 -2026-07-14 +2025-09-11 全球 -qwen-plus-character +`qwen-plus-2025-09-11` -千问系列角色扮演模型，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。[角色扮演（Qwen-Character）](https://help.aliyun.com/zh/model-studio/role-play) +本版本为2025年9月11日快照，相较7月28日快照在思考模式下指令遵循能力有提升、模型总结回复更加精简；在非思考模式下中文文本理解能力更佳、逻辑推理能力有增强。支持1M上下文长度，按照上下文长度进行阶梯计费。 -推理模型 +文本生成、深度思考 -2026-07-13 +2025-09-05 -美国 +全球 -glm-5.2-us +`qwen3-max-preview` -GLM-5.2是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 +Qwen3系列Max模型Preview版本，实现思考模式和非思考模式的有效融合。思考模式下在智能体编程能力、常识知识推理能力、数学/科学/通用类推理等能力上均有显著增强。 -推理模型 +文本生成、深度思考 -2026-07-06 +2025-08-01 美国 -qwen3.7-plus-us +`qwen-flash-us` -千问3.7Plus系列，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 +`qwen-flash-2025-07-28-us` -推理模型 +Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。 -2026-07-03 +文本生成、深度思考 -美国 +2025-08-01 + +全球 -qwen3.7-max-us +`qwen-flash-us` -Qwen Max 系列新一代旗舰模型。仅支持纯文本输入，默认开启思考模式，支持显式缓存，在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 +`qwen-flash-2025-07-28-us` -文生视频 +Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。 -2026-06-26 +文本生成、深度思考 + +2025-08-01 全球 -happyhorse-1.1-t2v +`qwen-flash` -HappyHorse 1.1系列文生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-文生视频](https://help.aliyun.com/zh/model-studio/happyhorse-text-to-video-api-reference) +`qwen-flash-2025-07-28` -图生视频 +Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。 -2026-06-26 +文本生成 + +2025-07-31 全球 -happyhorse-1.1-i2v +`qwen3-coder-30b-a3b-instruct` -HappyHorse 1.1系列图生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/happyhorse-image-to-video-api-reference) +基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。 -参考生视频 +文本生成、深度思考 -2026-06-26 +2025-07-30 全球 -happyhorse-1.1-r2v +`qwen3-30b-a3b-thinking-2507` -HappyHorse 1.1系列参考生视频模型，支持多参考图输入生成有声视频，可生成3~15秒、720P/1080P视频。[HappyHorse-参考生视频](https://help.aliyun.com/zh/model-studio/happyhorse-reference-to-video-api-reference) +基于Qwen3的思考模式开源模型，相较上一版本（千问3-30B-A3B）复杂推理类任务性能优秀，包括逻辑推理、数学、科学、代码类等具有一定难度的任务场景，指令遵循、文本理解、多语言翻译等能力显著提高。 -推理模型 +文本生成 -2026-06-10 +2025-07-29 全球 -qwen3.7-max-2026-06-08 +`qwen3-coder-flash` -Qwen3.7系列中规模最大、综合能力最强的Max模型，相较于5月20日快照增加了视觉模态理解能力，能够感知真实世界场景，具备多模态交互混合智能体能力。 +`qwen3-coder-flash-2025-07-28` -推理模型 +基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。 -2026-06-02 +文本生成 + +2025-07-29 全球 -qwen3.7-plus、qwen3.7-plus-2026-05-26 +`qwen3-30b-a3b-instruct-2507` -千问3.7Plus系列，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 +基于Qwen3的非思考模式开源模型，相较上一版本（千问3-30B-A3B）中英文和多语言整体通用能力有大幅提升。主观开放类任务专项优化，显著更加符合用户偏好，能够提供更有帮助性的回复。 -推理模型 +文本生成、深度思考 -2026-05-21 +2025-07-29 全球 -qwen3.7-max、qwen3.7-max-2026-05-20 +`qwen-plus-2025-07-28` -Qwen Max 系列新一代旗舰模型。仅支持纯文本输入，默认开启思考模式，支持显式缓存，在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。相较上个版本在中英文能力、工具调用上进行了专用增强。本版本为2025年7月28日快照，首次支持1M上下文长度，按照上下文长度进行阶梯计费。 -推理模型 +文本生成、深度思考 -2026-05-20 +2025-07-25 全球 -glm-5.1 +`qwen3-235b-a22b-thinking-2507` -智谱GLM-5.1模型，专为长程任务设计，支持 200K 上下文，最大输出可达 128K Token。通过强大的逻辑推理、长文本理解及代码生成能力，在多项基准测试中表现优异，适用于智能交互、企业应用及开发辅助等场景。[GLM-阿里云](https://help.aliyun.com/zh/model-studio/glm) +基于Qwen3的思考模式开源模型，相较上一版本（千问3-235B-A22B）逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景。 -文生视频 +文本生成 -2026-05-06 +2025-07-24 全球 -happyhorse-1.0-t2v +`qwen-mt-plus` -HappyHorse系列文生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-文生视频](https://help.aliyun.com/zh/model-studio/happyhorse-text-to-video-api-reference) +基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 -图生视频 +文本生成 -2026-05-06 +2025-07-23 全球 -happyhorse-1.0-i2v +`qwen3-coder-plus` -HappyHorse系列图生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/happyhorse-image-to-video-api-reference) +`qwen3-coder-plus-2025-07-22` -视频编辑 +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。 -2026-05-06 +文本生成 + +2025-07-23 全球 -happyhorse-1.0-video-edit +`qwen3-coder-480b-a35b-instruct` -HappyHorse系列视频编辑模型，支持对视频进行编辑处理。[HappyHorse-视频编辑](https://help.aliyun.com/zh/model-studio/happyhorse-video-edit-api-reference) +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。 -参考生视频 +文本生成 -2026-05-06 +2025-07-23 全球 -happyhorse-1.0-r2v +`qwen3-235b-a22b-instruct-2507` -HappyHorse系列参考生视频模型，支持多参考图输入生成有声视频，可生成3~15秒、720P/1080P视频。[HappyHorse-参考生视频](https://help.aliyun.com/zh/model-studio/happyhorse-reference-to-video-api-reference) +基于Qwen3的非思考模式开源模型，相较上一版本（千问3-235B-A22B）主观创作能力与模型安全性均有小幅度提升。 -文生文与视觉理解 +视觉理解 -2026-04-29 +2025-07-07 全球 -kimi-k2.5 +`qwen-vl-ocr` -由月之暗面（Moonshot AI）公司推出的视觉理解模型，在代码生成、视觉理解等通用智能任务中表现突出。同时支持图像、视频与文本输入、对话与 Agent 任务。[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api) +千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。 -推理模型 +文本生成、深度思考 -2026-04-29 +2025-05-12 全球 -deepseek-v4-pro +`qwen3-8b` -deepseek-v4-pro `cached_token` 单价调整为 **1 元/百万 token**，标准 `input_token` 单价不变。详见[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)。 +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-7B。 -多语言翻译 +文本生成、深度思考 -2026-04-10 +2025-05-12 -美国 +全球 -qwen-mt-lite-us +`qwen3-32b` -千问基础级文本翻译大模型，支持31个语种互译，相较于qwen-mt-flash响应更快，成本更低，适用于对延迟敏感的场景。[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation) +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-32B-Instruct，达到同规模业界SOTA水平。 -推理模型 +文本生成、深度思考 -2026-04-02 +2025-05-12 全球 -qwen3.6-plus、qwen3.6-plus-2026-04-02 +`qwen3-30b-a3b` -千问3.6-Plus，代码开发能力重点升级（Agentic Coding、前端编程等），Vibe Coding体验显著提升；泛化场景推理能力进一步增强；多模态方面，万物识别、OCR、物体定位等能力显著提升；同时修复了Qwen3.5-Plus上线后的已知问题。使用方法与qwen3.5-plus一致。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-14B，达到同规模业界SOTA水平。 -推理模型 +文本生成、深度思考 -2026-03-04 +2025-05-12 全球 -qwen3.5-flash、qwen3.5-flash-2026-02-23、qwen3.5-122b-a10b、qwen3.5-27b、qwen3.5-35b-a3b +`qwen3-235b-a22b` -阿里巴巴推出的最新模型千问3.5-Flash和开源模型，支持文本、图像和视频输入，响应速度快，综合表现接近qwen3.5-plus，支持内置[工具调用](https://help.aliyun.com/zh/model-studio/tool-calls/)。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-72B-Instruct，达到同规模业界SOTA水平。 -推理模型 +文本生成、深度思考 -2026-03-04 +2025-05-12 全球 -qwen3.5-plus、qwen3.5-plus-2026-02-15、qwen3.5-397b-a17b +`qwen3-14b` + +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-14B。 -阿里巴巴推出的最新模型千问3.5-Plus和开源模型，支持文本、图像和视频输入，在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中表现卓越，支持内置[工具调用](https://help.aliyun.com/zh/model-studio/tool-calls/)。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +## 德国（法兰克福） -图生视频-基于首帧 +**模型类型** -2026-01-12 +**时间** -美国 +**服务部署范围** -wan2.6-i2v-us +**模型ID** -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[万相-图生视频-基于首帧（2.1-2.6）](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +**功能说明** -文生视频 +文本生成 -2026-01-12 +2026-07-02 -美国 +全球 -wan2.6-t2v-us +`qwen-plus-character` -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 -语音识别 +视频生成 -2026-01-12 +2026-06-16 -美国 +全球 -qwen3-asr-flash-us、qwen3-asr-flash-2025-09-08-us +`happyhorse-1.1-t2v` -支持任意采样率和声道的音频。[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) +HappyHorse-1.1-T2V支持文生视频，进一步提升文本语义理解、镜头调度与动态生成表现。模型能够更精准地还原创作意图，在人物动作、场景氛围、视觉美感和物理运动上生成更流畅自然、细节丰富且一致性更高的高质量视频。 -视觉理解 +视频生成 -2026-01-12 +2026-06-16 -美国 +全球 -qwen3-vl-flash-us、qwen3-vl-flash-2025-10-15-us +`happyhorse-1.1-r2v` -Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于开源版Qwen3-VL-30B-A3B，效果更优，响应速度更快。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +HappyHorse-1.1-R2V支持参考生视频，进一步提升主体、场景风格与画面一致性的稳定保持。模型最多支持9张参考图片输入，能够更精准理解并延续创作意图，在人物、场景、风格和镜头表现上带来更强的可控性与表现力。 -推理模型 +视频生成 -2026-01-12 +2026-06-16 -美国 +全球 -qwen-plus-2025-12-01-us +`happyhorse-1.1-i2v` -属于 Qwen3 系列模型，相较于qwen-plus-2025-07-28，在思考模式下提升了指令遵循能力、总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。在非思考模式下中文理解与逻辑推理能力得到增强，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。 -推理模型 +文本生成、深度思考 -2026-01-12 +2026-06-16 -美国 +全球 -qwen-plus-us +`glm-5.2` -能力均衡，推理效果、成本和速度介于千问Max和千问Flash之间，适合中等复杂任务。模型列表 +GLM-5.2是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 -文生文 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-06-15 -美国 +全球 -qwen-flash-us、qwen-flash-2025-07-28-us +`kimi-k2.7-code` -千问系列速度最快、成本极低的模型，适合简单任务。模型列表 +kimi-k2.7-code是 kimi 迄今最智能的coding模型，在长上下文中更可靠地遵循指令，能以更高的成功率完成编程任务，同时支持文本、图片与视频输入，思考模式，对话与 Agent 任务。 -文生图 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-06-09 全球 -wan2.6-t2i +`qwen3.7-max-2026-06-08` -新增同步接口。支持在总像素面积与宽高比约束内，自由选尺寸。[万相-文生图V2](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference) +Qwen3.7系列中规模最大、综合能力最强的Max模型，相较于5月20日快照增加了视觉模态理解能力，能够感知真实世界场景，具备多模态交互混合智能体能力。该版本为2026年6月8日快照。 -图像生成与编辑 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-06-01 全球 -wan2.6-image +`qwen3.7-plus` -支持图像编辑和图文混合输出。[万相-图像生成与编辑2.6](https://help.aliyun.com/zh/model-studio/wan-image-generation-api-reference) +`qwen3.7-plus-2026-05-26` -图生视频-基于首帧 +Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 -2026-01-12 +文本生成、深度思考 + +2026-05-20 全球 -wan2.6-i2v +`qwen3.7-max` -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[万相-图生视频-基于首帧（2.1-2.6）](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +`qwen3.7-max-2026-05-20` -参考生视频 +Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 -2026-01-12 +文本生成、深度思考、视觉理解 + +2026-04-29 全球 -wan2.6-r2v +`kimi-k2.5` -基于参考视频的角色形象和音色，生成多镜头视频，支持自动配音。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +kimi-k2.5是月之暗面迄今发布最全能的模型，原生多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与Agent任务。 -文生视频 +视频生成 -2026-01-12 +2026-04-26 全球 -wan2.6-t2v +`happyhorse-1.0-video-edit` -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。 -视觉理解 +视频生成 -2026-01-12 +2026-04-26 全球 -qwen3-vl-flash、qwen3-vl-flash-2025-10-15 +`happyhorse-1.0-r2v` -Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于开源版Qwen3-VL-30B-A3B，效果更优，响应速度更快。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +HappyHorse-1.0-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。 -视觉理解 +文本生成、深度思考 -2026-01-12 +2026-04-24 全球 -qwen3-vl-30b-a3b-thinking、qwen3-vl-30b-a3b-instruct +`deepseek-v4-pro` -基于Qwen3-VL新一代开源模型，提供思考和非思考两个版本。响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级空间感知与万物识别能力，胜任复杂现实任务。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 -视觉理解 +文本生成、深度思考 -2026-01-12 +2026-04-24 全球 -qwen3-vl-8b-thinking、qwen3-vl-8b-instruct +`deepseek-v4-flash` -Qwen3-VL系列 8B 的Dense开源模型，提供思考和非思考两个版本。占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面空间感知与万物识别能力。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 -视觉理解 +视频生成 -2026-01-12 +2026-04-22 全球 -qwen3-vl-32b-thinking、qwen3-vl-32b-instruct +`happyhorse-1.0-t2v` -Qwen3-VL系列 32B 的Dense模型，综合表现仅次于Qwen3-VL-235B模型，文档识别和理解、空间感知与万物识别、视觉2D检测/空间推理能力均表现出色，适合通用场景下的复杂感知任务。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +HappyHorse-1.0-T2V支持文生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 -视觉理解 +视频生成 -2026-01-12 +2026-04-22 全球 -qwen3-vl-plus、qwen3-vl-plus-2025-09-23、qwen3-vl-235b-a22b-thinking、qwen3-vl-235b-a22b-instruct +`happyhorse-1.0-i2v` -Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力达到世界顶尖水平。此版本在视觉编码、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +HappyHorse-1.0-I2V支持图生视频，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 -推理模型 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-04-17 全球 -qwen3-next-80b-a3b-thinking、qwen3-next-80b-a3b-instruct +`qwen3.6-flash` + +`qwen3.6-flash-2026-04-16` -基于Qwen3的新一代开源模型，thinking模型相较于qwen3-235b-a22b-thinking-2507提升了指令遵循能力，总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型相较于qwen3-235b-a22b-instruct-2507增强了中文理解、逻辑推理及文本生成能力，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。 -推理模型 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-04-17 全球 -qwen3-max、qwen3-max-2025-09-23 +`qwen3.6-35b-a3b` -相较qwen3-max-preview版本，在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。模型列表 +Qwen3.6系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果相较3.5-35B-A3B显著提升了agentic coding能力、数学推理和代码推理能力、空间智能能力、物体定位与目标检测能力。 -推理模型 +文本生成、深度思考 -2026-01-12 +2026-04-14 全球 -qwen3-max-preview +`glm-5.1` -基于Qwen3的Qwen-Max模型（预览版），相较Qwen 2.5系列整体通用能力有大幅度提升，中英文通用文本理解能力、复杂指令遵循能力、主观开放任务能力、多语言能力、工具调用能力均显著增强；模型知识幻觉更少。千问 Max +GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 -代码模型 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-04-01 全球 -qwen3-coder-flash、qwen3-coder-flash-2025-07-28 +`qwen3.6-plus` -千问Coder系列速度最快、成本最低的模型。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder)。 +`qwen3.6-plus-2026-04-02` -代码模型 +Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。 -2026-01-12 +文本生成、深度思考、视觉理解 -全球 +2026-02-23 + +欧盟 + +`qwen3.5-flash` -qwen3-coder-plus-2025-09-23 +`qwen3.5-flash-2026-02-23` -相较上一版本（7月22日快照）在下游任务效果和工具调用方面鲁棒性有所提升，代码安全性增强。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +Qwen3.5原生视觉语言系列Flash模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步；响应速度快，兼具推理速度和性能。 -代码模型 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-02-23 全球 -qwen3-coder-plus、qwen3-coder-plus-2025-07-22、qwen3-coder-30b-a3b-instruct、qwen3-coder-480b-a35b-instruct +`qwen3.5-flash` -基于 Qwen3 的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，代码能力卓越的同时兼具通用能力。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +`qwen3.5-flash-2026-02-23` -推理模型 +Qwen3.5原生视觉语言系列Flash模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步；响应速度快，兼具推理速度和性能。 -2026-01-12 +文本生成、深度思考、视觉理解 + +2026-02-23 全球 -qwen3-30b-a3b-thinking-2507、qwen3-30b-a3b-instruct-2507 +`qwen3.5-35b-a3b` -是qwen3-30b-a3b的升级版。thinking模型逻辑能力、通用能力、知识增强及创作能力提升，参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型创作能力与模型安全性提升，参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。 -推理模型 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-02-23 全球 -qwen3-235b-a22b-thinking-2507、qwen3-235b-a22b-instruct-2507 +`qwen3.5-27b` -是qwen3-235b-a22b的升级版。thinking模型逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景，参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型创作能力与模型安全性均有提升，参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。 -推理模型 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-02-23 全球 -qwen3-235b-a22b、qwen3-30b-a3b、qwen3-32b、qwen3-14b、qwen3-8b +`qwen3.5-122b-a10b` + +Qwen3.5系列122B-A10B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现仅次于Qwen3.5-397B-A17B，文本能力显著优于Qwen3-235B-2507，视觉能力优于Qwen3-VL-235B。 -Qwen3 模型支持思考模式和非思考模式，您可以通过 `enable_thinking` 参数实现两种模式的切换。除此之外，Qwen3 模型的能力得到了大幅提升： +文本生成 -1. 推理能力：在数学、代码和逻辑推理等评测中，显著超过 QwQ 和同尺寸的非推理模型，达到同规模业界顶尖水平。 - -2. 人类偏好能力：创意写作、角色扮演、多轮对话、指令遵循能力均大幅提升，通用能力显著超过同尺寸模型。 - -3. Agent 能力：在推理、非推理两种模式下都达到业界领先水平，能够精准地调用外部工具。 - -4. 多语言能力：支持100多种语言和方言，多语言翻译、指令理解、常识推理能力都明显提升。 - -5. 回复格式问题修复：修复了之前版本存在的回复格式的问题，如异常 Markdown、中间截断、错误输出 boxed 等问题。 - +2026-02-20 + +欧盟 + +`qwen3-coder-next` -思考模式请参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)，非思考模式请参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +Qwen3系列新一代代码生成模型，效果接近Qwen3-Coder-Plus兼具更优性能。模型重点优化仓库级别理解、支持多轮工具交互、提升对于agentic coding类工具的适配能力。 -文字提取 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-02-15 全球 -qwen-vl-ocr-2025-11-20 +`qwen3.5-plus` -千问文字提取模型，该快照版基于Qwen3-VL架构，大幅提升文档解析、文字定位能力。[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr) +`qwen3.5-plus-2026-02-15` -文字提取 +Qwen3.5原生视觉语言系列Plus模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在多项任务评测中，3.5系列均展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 -2026-01-12 +文本生成、深度思考、视觉理解 + +2026-02-15 全球 -qwen-vl-ocr +`qwen3.5-397b-a17b` -qwen-vl-ocr是专用于OCR的模型；在表格、试题等类型图像的文字提取能力大幅提升。详情请参见[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr)。 +Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。 -推理模型 +文本生成、深度思考 -2026-01-12 +2026-01-23 -全球 +欧盟 -qwen-plus-2025-12-01、qwen-plus-2025-09-11 +`qwen3-max-2026-01-23` -属于 Qwen3 系列模型，相较于qwen-plus-2025-07-28，在思考模式下提升了指令遵循能力、总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。在非思考模式下中文理解与逻辑推理能力得到增强，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +千问3系列Max模型，相较2025年9月23日快照，此版本实现思考模式和非思考模式的有效融合，模型整体效果得到全方位的大幅度提升。在思考模式下，同时发布Web搜索、Web信息提取和代码解释器工具能力，使得模型在慢思考的同时，能够通过引入外部工具，以更高的准确性解决更有难度的问题。此版本为2026年1月23日快照。 -推理模型 +视觉理解 -2026-01-12 +2026-01-22 -全球 +欧盟 -qwen-plus-2025-07-28 +`qwen3-vl-flash-2026-01-22` -属于 Qwen3 系列模型，相较于上一版模型，将上下文长度提高到了1,000,000。思考模式请参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)，非思考模式请参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于2025年10月15日快照，模型整体效果有大幅提升：在通用视觉识别与推理方面有增强；在安防、巡店、巡检、拍照解题等业务场景识别效果上提升显著。此版本为2026年1月22日快照版本。 -推理模型 +视频生成 -2026-01-12 +2025-12-16 全球 -qwen-plus +`wan2.6-r2v` -能力均衡，推理效果、成本和速度介于千问Max和千问Flash之间，适合中等复杂任务。模型列表 +万相2.6-参考生视频，支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍 -多语言翻译 +图片生成 -2026-01-12 +2025-12-15 全球 -qwen-mt-lite +`wan2.6-t2i` -千问基础级文本翻译大模型，支持31个语种互译，相较于qwen-mt-flash响应更快，成本更低，适用于对延迟敏感的场景。[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation) +万相2.6-文生图，画面质感、美学表现、指令遵循升级，在艺术风格精准控制、真实感人像、长文本生图及广泛历史文化IP覆盖上均表现出卓越能力，可生成高质量且富有表现力的视觉内容。 -多语言翻译 +图片生成 -2026-01-12 +2025-12-15 全球 -qwen-mt-plus、qwen-mt-flash +`wan2.6-image` -Qwen-MT模型是基于千问模型优化的机器翻译大语言模型，擅长中英互译、中文与小语种互译、英文与小语种互译，小语种包括日、韩、法、西、德、葡（巴西）、泰、印尼、越、阿等26种。在多语言互译的基础上，提供术语干预、领域提示、记忆库等能力，提升模型在复杂应用场景下的翻译效果。详情请参见[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation)。 +万相2.6-图像生成，全能图像生成模型，支持图文一体化推理生成，具备多图创意融合、商用级一致性、美学要素迁移与镜头光影精确控制，全面提升图像生成的一致性、可控性和表现力。 -文生文 +视频生成 -2026-01-12 +2025-12-03 全球 -qwen-flash、qwen-flash-2025-07-28 +`wan2.6-t2v` -千问系列速度最快、成本极低的模型，适合简单任务。模型列表 - -## 德国（法兰克福） +Wan2.6-文生视频，智能分镜调度支持多镜头叙事，能够生成主体、场景和氛围一致的多镜头叙事视频，最高支持15秒时长，更高品质的声音生成，更好的指令遵循和视觉质量 -**模型类型** +视频生成 -**时间** +2025-12-03 -**服务部署范围** +全球 -**模型规格** +`wan2.6-i2v` -**功能说明** +万相2.6-图生视频，智能分镜调度支持多镜头叙事，更高品质的声音生成，多人稳定对话，更自然真实音色，最高支持15秒时长生成 -文生文 +— -2026-07-14 +2025-12-01 -全球 +欧盟 -qwen-plus-character +`qwen-plus-2025-12-01` -千问系列角色扮演模型，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。[角色扮演（Qwen-Character）](https://help.aliyun.com/zh/model-studio/role-play) +本版本为2025年12月1日快照，相较7月28日快照在推理能力上有提升；智能体能力、多轮工具调用能力进一步增强；主观创作类任务表现更优。支持1M上下文长度，按照上下文长度进行阶梯计费。 -文生视频 +— -2026-06-26 +2025-12-01 全球 -happyhorse-1.1-t2v +`qwen-plus-2025-12-01` -HappyHorse 1.1系列文生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-文生视频](https://help.aliyun.com/zh/model-studio/happyhorse-text-to-video-api-reference) +本版本为2025年12月1日快照，相较7月28日快照在推理能力上有提升；智能体能力、多轮工具调用能力进一步增强；主观创作类任务表现更优。支持1M上下文长度，按照上下文长度进行阶梯计费。 -图生视频 +视觉理解 -2026-06-26 +2025-11-21 全球 -happyhorse-1.1-i2v +`qwen-vl-ocr-2025-11-20` -HappyHorse 1.1系列图生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/happyhorse-image-to-video-api-reference) +本模型为2025年11月20日的快照版本，基于最新Qwen-VL3架构全面升级，在文档解析，文字定位能力全面升级，端到端时延、幻觉大幅降低。 -参考生视频 +视觉理解 -2026-06-26 +2025-11-20 全球 -happyhorse-1.1-r2v +`qwen-vl-ocr` -HappyHorse 1.1系列参考生视频模型，支持多参考图输入生成有声视频，可生成3~15秒、720P/1080P视频。[HappyHorse-参考生视频](https://help.aliyun.com/zh/model-studio/happyhorse-reference-to-video-api-reference) +千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。 -推理模型 +文本生成 -2026-06-10 +2025-11-19 全球 -qwen3.7-max-2026-06-08 +`qwen-mt-lite` -Qwen3.7系列中规模最大、综合能力最强的Max模型，相较于5月20日快照增加了视觉模态理解能力，能够感知真实世界场景，具备多模态交互混合智能体能力。 +基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 -推理模型 +文本生成 -2026-06-02 +2025-11-11 全球 -qwen3.7-plus、qwen3.7-plus-2026-05-26 +`qwen-mt-flash` -千问3.7Plus系列，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 +基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 -推理模型 +深度思考、视觉理解 -2026-05-21 +2025-10-21 全球 -qwen3.7-max、qwen3.7-max-2026-05-20 +`qwen3-vl-32b-thinking` -Qwen Max 系列新一代旗舰模型。仅支持纯文本输入，默认开启思考模式，支持显式缓存，在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 +Qwen3-VL系列最大尺寸Dense模型的推理版本，多模态推理能力仅次于Qwen3-VL-235B-Thinking，STEM&数学类解题能力、通用图像和视频理解能力出众，多模态Agent能力达到SOTA，适合做复杂多模态推理任务。 -推理模型 +视觉理解 -2026-05-20 +2025-10-21 全球 -glm-5.1 +`qwen3-vl-32b-instruct` -智谱GLM-5.1模型，专为长程任务设计，支持 200K 上下文，最大输出可达 128K Token。通过强大的逻辑推理、长文本理解及代码生成能力，在多项基准测试中表现优异，适用于智能交互、企业应用及开发辅助等场景。[GLM-阿里云](https://help.aliyun.com/zh/model-studio/glm) +Qwen3-VL系列最大尺寸Dense模型的非推理版本，综合表现仅次于Qwen3-VL-235B-Instruct，文档识别和理解能力出色，空间感知与万物识别能力强，视觉2D检测/空间推理能力达到SOTA，适合通用场景下的复杂感知任务。 -推理模型 +深度思考、视觉理解 -2026-05-07 +2025-10-15 欧盟 -qwen3-max、qwen3-max-2026-01-23 +`qwen3-vl-flash` -相较于 2025 年 9 月 23 日的快照版本，有效融合了思考模式与非思考模式，显著提升了模型的整体性能。 +`qwen3-vl-flash-2025-10-15` -推理模型 +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。 -2026-05-07 +深度思考、视觉理解 -欧盟 +2025-10-15 + +全球 -qwen-plus、qwen-plus-2025-12-01 +`qwen3-vl-flash` -属于 Qwen3 系列模型，相较于qwen-plus-2025-07-28，在思考模式下提升了指令遵循能力、总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。在非思考模式下中文理解与逻辑推理能力得到增强，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +`qwen3-vl-flash-2025-10-15` -推理模型 +Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。 -2026-05-07 +深度思考、视觉理解 -欧盟 +2025-10-03 + +全球 -qwen3.5-flash、qwen3.5-flash-2026-02-23 +`qwen3-vl-30b-a3b-thinking` -阿里巴巴推出的最新模型千问3.5-Flash和开源模型，支持文本、图像和视频输入，响应速度快。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +Qwen3-VL系列第二大MoE模型的Thinking版本，响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级图像/视频理解、空间感知与万物识别能力，胜任复杂现实任务。 视觉理解 -2026-05-07 +2025-10-03 -欧盟 +全球 -qwen3-vl-plus、qwen3-vl-flash、qwen3-vl-flash-2025-10-15 +`qwen3-vl-30b-a3b-instruct` -Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力达到世界顶尖水平。此版本在视觉编码、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，指令遵循能力更强，具有更低的延迟。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +Qwen3-VL系列第二大MoE模型的Instruct版本，响应速度快，支持长视频长文档等超长上下文；全面升级图像/视频理解、空间感知与万物识别能力；具备视觉2D/3D定位能力，胜任复杂现实任务。 -代码能力 +深度思考、视觉理解 -2026-05-07 +2025-09-30 -欧盟 +全球 -qwen3-coder-next +`qwen3-vl-8b-thinking` -Qwen3系列新一代开源代码生成模型，支持多轮工具交互，提升了对仓库级别代码的理解能力和对AI编程工具的适配性。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +Qwen3-VL系列8B Dense模型的Thinking版本，占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面升级图像/视频理解、空间感知与万物识别能力。 -文生视频 +视觉理解 -2026-05-06 +2025-09-30 全球 -happyhorse-1.0-t2v +`qwen3-vl-8b-instruct` -HappyHorse系列文生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-文生视频](https://help.aliyun.com/zh/model-studio/happyhorse-text-to-video-api-reference) +Qwen3-VL系列8B Dense模型的Instruct版本，占用显存更低，全面升级图像/视频理解、长视频长文档等超长上下文支持、空间感知与万物识别能力，胜任复杂现实任务。 -图生视频 +文本生成、深度思考 -2026-05-06 +2025-09-24 -全球 +欧盟 -happyhorse-1.0-i2v +`qwen3-max` -HappyHorse系列图生视频模型，支持有声视频生成，可生成3~15秒、720P/1080P视频。[HappyHorse-图生视频-基于首帧](https://help.aliyun.com/zh/model-studio/happyhorse-image-to-video-api-reference) +千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。 -视频编辑 +文本生成、深度思考 -2026-05-06 +2025-09-24 全球 -happyhorse-1.0-video-edit +`qwen3-max` -HappyHorse系列视频编辑模型，支持对视频进行编辑处理。[HappyHorse-视频编辑](https://help.aliyun.com/zh/model-studio/happyhorse-video-edit-api-reference) +`qwen3-max-2025-09-23` -参考生视频 +千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。 -2026-05-06 +视觉理解 -全球 +2025-09-23 -happyhorse-1.0-r2v +欧盟 -HappyHorse系列参考生视频模型，支持多参考图输入生成有声视频，可生成3~15秒、720P/1080P视频。[HappyHorse-参考生视频](https://help.aliyun.com/zh/model-studio/happyhorse-reference-to-video-api-reference) +`qwen3-vl-plus` -文生文与视觉理解 +Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。 -2026-04-29 +视觉理解 + +2025-09-23 全球 -kimi-k2.5 +`qwen3-vl-plus` -由月之暗面（Moonshot AI）公司推出的视觉理解模型，在代码生成、视觉理解等通用智能任务中表现突出。同时支持图像、视频与文本输入、对话与 Agent 任务。[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api) +`qwen3-vl-plus-2025-09-23` -推理模型 +Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。 -2026-04-29 +深度思考、视觉理解 + +2025-09-23 全球 -deepseek-v4-pro +`qwen3-vl-235b-a22b-thinking` -deepseek-v4-pro `cached_token` 单价调整为 **1 元/百万 token**，标准 `input_token` 单价不变。详见[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)。 +Qwen3系列视觉理解模型，多模态思考能力显著增强，模型在STEM与数学推理方面进行了重点优化；视觉感知与识别能力全面提升、OCR能力迎来重大升级。 -推理模型 +视觉理解 -2026-04-02 +2025-09-23 全球 -qwen3.6-plus、qwen3.6-plus-2026-04-02 +`qwen3-vl-235b-a22b-instruct` -千问3.6-Plus，代码开发能力重点升级（Agentic Coding、前端编程等），Vibe Coding体验显著提升；泛化场景推理能力进一步增强；多模态方面，万物识别、OCR、物体定位等能力显著提升；同时修复了Qwen3.5-Plus上线后的已知问题。使用方法与qwen3.5-plus一致。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +Qwen3系列视觉理解模型，在视觉coding、空间感知等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解，OCR能力迎来重大升级。 -推理模型 +文本生成 -2026-03-04 +2025-09-23 全球 -qwen3.5-flash、qwen3.5-flash-2026-02-23、qwen3.5-122b-a10b、qwen3.5-27b、qwen3.5-35b-a3b +`qwen3-coder-plus-2025-09-23` -阿里巴巴推出的最新模型千问3.5-Flash和开源模型，支持文本、图像和视频输入，响应速度快，综合表现接近qwen3.5-plus，支持内置[工具调用](https://help.aliyun.com/zh/model-studio/tool-calls/)。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。本版本为2025年9月23日快照，相较上一版本（7月22日快照）在下游任务效果和工具调用方面鲁棒性有所提升；代码安全性增强。 -推理模型 +文本生成、深度思考 -2026-03-04 +2025-09-16 -全球 +欧盟 -qwen3.5-plus、qwen3.5-plus-2026-02-15、qwen3.5-397b-a17b +`qwen-plus` -阿里巴巴推出的最新模型千问3.5-Plus和开源模型，支持文本、图像和视频输入，在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中表现卓越，支持内置[工具调用](https://help.aliyun.com/zh/model-studio/tool-calls/)。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +千问超大规模语言模型增强版，支持中文英文等不同语言输入。相对于之前版本，中英文code能力、逻辑能力、多语言能力显著提升，回复风格面向人类偏好进行大幅调整，模型回复详实程度和格式清晰度明显改善，创作类专项、json格式遵循专项、角色扮演专项能力定向提升。 -文生图 +文本生成、深度思考 -2026-01-12 +2025-09-16 全球 -wan2.6-t2i +`qwen-plus` -新增同步接口。支持在总像素面积与宽高比约束内，自由选尺寸。[万相-文生图V2](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference) +千问超大规模语言模型增强版，支持中文英文等不同语言输入。相对于之前版本，中英文code能力、逻辑能力、多语言能力显著提升，回复风格面向人类偏好进行大幅调整，模型回复详实程度和格式清晰度明显改善，创作类专项、json格式遵循专项、角色扮演专项能力定向提升。 -图像生成与编辑 +文本生成、深度思考 -2026-01-12 +2025-09-11 全球 -wan2.6-image +`qwen3-next-80b-a3b-thinking` -支持图像编辑和图文混合输出。[万相-图像生成与编辑2.6](https://help.aliyun.com/zh/model-studio/wan-image-generation-api-reference) +基于Qwen3的新一代思考模式开源模型，相较上一版本（千问3-235B-A22B-Thinking-2507指令遵循能力有提升、模型总结回复更加精简。 -图生视频-基于首帧 +文本生成 -2026-01-12 +2025-09-11 全球 -wan2.6-i2v +`qwen3-next-80b-a3b-instruct` -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[万相-图生视频-基于首帧（2.1-2.6）](https://help.aliyun.com/zh/model-studio/legacy-image-to-video-api-reference/) +基于Qwen3的新一代非思考模式开源模型，相较上一版本（千问3-235B-A22B-Instruct-2507）中文文本理解能力更佳、逻辑推理能力有增强、文本生成类任务表现更好。 -参考生视频 +文本生成、深度思考 -2026-01-12 +2025-09-11 全球 -wan2.6-r2v +`qwen-plus-2025-09-11` -基于参考视频的角色形象和音色，生成多镜头视频，支持自动配音。[万相2.7-参考生视频](https://help.aliyun.com/zh/model-studio/wan-video-to-video-api-reference) +本版本为2025年9月11日快照，相较7月28日快照在思考模式下指令遵循能力有提升、模型总结回复更加精简；在非思考模式下中文文本理解能力更佳、逻辑推理能力有增强。支持1M上下文长度，按照上下文长度进行阶梯计费。 -文生视频 +文本生成、深度思考 -2026-01-12 +2025-09-05 全球 -wan2.6-t2v +`qwen3-max-preview` -新增多镜头叙事能力，支持音频能力，支持自动配音，或传入自定义音频文件。[万相2.7-文生视频](https://help.aliyun.com/zh/model-studio/text-to-video-api-reference) +Qwen3系列Max模型Preview版本，实现思考模式和非思考模式的有效融合。思考模式下在智能体编程能力、常识知识推理能力、数学/科学/通用类推理等能力上均有显著增强。 -视觉理解 +文本生成、深度思考 -2026-01-12 +2025-08-01 全球 -qwen3-vl-flash、qwen3-vl-flash-2025-10-15 +`qwen-flash` -Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，相较于开源版Qwen3-VL-30B-A3B，效果更优，响应速度更快。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +`qwen-flash-2025-07-28` -视觉理解 +Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。 + +文本生成 -2026-01-12 +2025-07-31 全球 -qwen3-vl-30b-a3b-thinking、qwen3-vl-30b-a3b-instruct +`qwen3-coder-30b-a3b-instruct` -基于Qwen3-VL新一代开源模型，提供思考和非思考两个版本。响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级空间感知与万物识别能力，胜任复杂现实任务。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。 -视觉理解 +文本生成、深度思考 -2026-01-12 +2025-07-30 全球 -qwen3-vl-8b-thinking、qwen3-vl-8b-instruct +`qwen3-30b-a3b-thinking-2507` -Qwen3-VL系列 8B 的Dense开源模型，提供思考和非思考两个版本。占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面空间感知与万物识别能力。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +基于Qwen3的思考模式开源模型，相较上一版本（千问3-30B-A3B）复杂推理类任务性能优秀，包括逻辑推理、数学、科学、代码类等具有一定难度的任务场景，指令遵循、文本理解、多语言翻译等能力显著提高。 -视觉理解 +文本生成 -2026-01-12 +2025-07-29 全球 -qwen3-vl-32b-thinking、qwen3-vl-32b-instruct +`qwen3-coder-flash` -Qwen3-VL系列 32B 的Dense模型，综合表现仅次于Qwen3-VL-235B模型，文档识别和理解、空间感知与万物识别、视觉2D检测/空间推理能力均表现出色，适合通用场景下的复杂感知任务。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +`qwen3-coder-flash-2025-07-28` -视觉理解 +基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。 -2026-01-12 +文本生成 + +2025-07-29 全球 -qwen3-vl-plus、qwen3-vl-plus-2025-09-23、qwen3-vl-235b-a22b-thinking、qwen3-vl-235b-a22b-instruct +`qwen3-30b-a3b-instruct-2507` -Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力达到世界顶尖水平。此版本在视觉编码、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升。[图像与视频理解](https://help.aliyun.com/zh/model-studio/vision) +基于Qwen3的非思考模式开源模型，相较上一版本（千问3-30B-A3B）中英文和多语言整体通用能力有大幅提升。主观开放类任务专项优化，显著更加符合用户偏好，能够提供更有帮助性的回复。 -推理模型 +文本生成、深度思考 -2026-01-12 +2025-07-29 全球 -qwen3-next-80b-a3b-thinking、qwen3-next-80b-a3b-instruct +`qwen-plus-2025-07-28` -基于Qwen3的新一代开源模型，thinking模型相较于qwen3-235b-a22b-thinking-2507提升了指令遵循能力，总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型相较于qwen3-235b-a22b-instruct-2507增强了中文理解、逻辑推理及文本生成能力，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。相较上个版本在中英文能力、工具调用上进行了专用增强。本版本为2025年7月28日快照，首次支持1M上下文长度，按照上下文长度进行阶梯计费。 -推理模型 +文本生成、深度思考 -2026-01-12 +2025-07-25 全球 -qwen3-max、qwen3-max-2025-09-23 +`qwen3-235b-a22b-thinking-2507` -相较qwen3-max-preview版本，在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。模型列表 +基于Qwen3的思考模式开源模型，相较上一版本（千问3-235B-A22B）逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景。 -推理模型 +文本生成 -2026-01-12 +2025-07-24 全球 -qwen3-max-preview +`qwen-mt-plus` -基于Qwen3的Qwen-Max模型（预览版），相较Qwen 2.5系列整体通用能力有大幅度提升，中英文通用文本理解能力、复杂指令遵循能力、主观开放任务能力、多语言能力、工具调用能力均显著增强；模型知识幻觉更少。千问 Max +基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 -代码模型 +文本生成 -2026-01-12 +2025-07-23 全球 -qwen3-coder-flash、qwen3-coder-flash-2025-07-28 +`qwen3-coder-plus` + +`qwen3-coder-plus-2025-07-22` -千问Coder系列速度最快、成本最低的模型。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder)。 +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。 -代码模型 +文本生成 -2026-01-12 +2025-07-23 全球 -qwen3-coder-plus-2025-09-23 +`qwen3-coder-480b-a35b-instruct` -相较上一版本（7月22日快照）在下游任务效果和工具调用方面鲁棒性有所提升，代码安全性增强。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。 -代码模型 +文本生成 -2026-01-12 +2025-07-23 全球 -qwen3-coder-plus、qwen3-coder-plus-2025-07-22、qwen3-coder-30b-a3b-instruct、qwen3-coder-480b-a35b-instruct +`qwen3-235b-a22b-instruct-2507` -基于 Qwen3 的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，代码能力卓越的同时兼具通用能力。[代码能力（Qwen-Coder）](https://help.aliyun.com/zh/model-studio/qwen-coder) +基于Qwen3的非思考模式开源模型，相较上一版本（千问3-235B-A22B）主观创作能力与模型安全性均有小幅度提升。 -推理模型 +文本生成、深度思考 -2026-01-12 +2025-05-12 全球 -qwen3-30b-a3b-thinking-2507、qwen3-30b-a3b-instruct-2507 +`qwen3-8b` -是qwen3-30b-a3b的升级版。thinking模型逻辑能力、通用能力、知识增强及创作能力提升，参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型创作能力与模型安全性提升，参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-7B。 -推理模型 +文本生成、深度思考 -2026-01-12 +2025-05-12 全球 -qwen3-235b-a22b-thinking-2507、qwen3-235b-a22b-instruct-2507 +`qwen3-32b` -是qwen3-235b-a22b的升级版。thinking模型逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景，参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。instruct模型创作能力与模型安全性均有提升，参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-32B-Instruct，达到同规模业界SOTA水平。 -推理模型 +文本生成、深度思考 -2026-01-12 +2025-05-12 全球 -qwen3-235b-a22b、qwen3-30b-a3b、qwen3-32b、qwen3-14b、qwen3-8b - -Qwen3 模型支持思考模式和非思考模式，您可以通过 `enable_thinking` 参数实现两种模式的切换。除此之外，Qwen3 模型的能力得到了大幅提升： - -1. 推理能力：在数学、代码和逻辑推理等评测中，显著超过 QwQ 和同尺寸的非推理模型，达到同规模业界顶尖水平。 - -2. 人类偏好能力：创意写作、角色扮演、多轮对话、指令遵循能力均大幅提升，通用能力显著超过同尺寸模型。 - -3. Agent 能力：在推理、非推理两种模式下都达到业界领先水平，能够精准地调用外部工具。 - -4. 多语言能力：支持100多种语言和方言，多语言翻译、指令理解、常识推理能力都明显提升。 - -5. 回复格式问题修复：修复了之前版本存在的回复格式的问题，如异常 Markdown、中间截断、错误输出 boxed 等问题。 - +`qwen3-30b-a3b` -思考模式请参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)，非思考模式请参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-14B，达到同规模业界SOTA水平。 -文字提取 +文本生成、深度思考 -2026-01-12 +2025-05-12 全球 -qwen-vl-ocr-2025-11-20 +`qwen3-235b-a22b` -千问文字提取模型，该快照版基于Qwen3-VL架构，大幅提升文档解析、文字定位能力。[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr) +实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-72B-Instruct，达到同规模业界SOTA水平。 -文字提取 +## 日本（东京） -2026-01-12 +**模型类型** -全球 +**时间** -qwen-vl-ocr +**服务部署范围** + +**模型ID** -qwen-vl-ocr是专用于OCR的模型；在表格、试题等类型图像的文字提取能力大幅提升。详情请参见[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr)。 +**功能说明** -推理模型 +文本生成 -2026-01-12 +2026-07-02 全球 -qwen-plus-2025-12-01、qwen-plus-2025-09-11 +`qwen-plus-character` -属于 Qwen3 系列模型，相较于qwen-plus-2025-07-28，在思考模式下提升了指令遵循能力、总结回复更加精简，详见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)。在非思考模式下中文理解与逻辑推理能力得到增强，详见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 -推理模型 +视频生成 -2026-01-12 +2026-06-16 全球 -qwen-plus-2025-07-28 +`happyhorse-1.1-t2v` -属于 Qwen3 系列模型，相较于上一版模型，将上下文长度提高到了1,000,000。思考模式请参见[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)，非思考模式请参见[概述](https://help.aliyun.com/zh/model-studio/text-generation)。 +HappyHorse-1.1-T2V支持文生视频，进一步提升文本语义理解、镜头调度与动态生成表现。模型能够更精准地还原创作意图，在人物动作、场景氛围、视觉美感和物理运动上生成更流畅自然、细节丰富且一致性更高的高质量视频。 -推理模型 +视频生成 -2026-01-12 +2026-06-16 全球 -qwen-plus +`happyhorse-1.1-r2v` -能力均衡，推理效果、成本和速度介于千问Max和千问Flash之间，适合中等复杂任务。模型列表 +HappyHorse-1.1-R2V支持参考生视频，进一步提升主体、场景风格与画面一致性的稳定保持。模型最多支持9张参考图片输入，能够更精准理解并延续创作意图，在人物、场景、风格和镜头表现上带来更强的可控性与表现力。 -多语言翻译 +视频生成 -2026-01-12 +2026-06-16 全球 -qwen-mt-lite +`happyhorse-1.1-i2v` -千问基础级文本翻译大模型，支持31个语种互译，相较于qwen-mt-flash响应更快，成本更低，适用于对延迟敏感的场景。[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation) +HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。 -多语言翻译 +文本生成、深度思考、视觉理解 -2026-01-12 +2026-06-01 全球 -qwen-mt-plus、qwen-mt-flash +`qwen3.7-plus` -Qwen-MT模型是基于千问模型优化的机器翻译大语言模型，擅长中英互译、中文与小语种互译、英文与小语种互译，小语种包括日、韩、法、西、德、葡（巴西）、泰、印尼、越、阿等26种。在多语言互译的基础上，提供术语干预、领域提示、记忆库等能力，提升模型在复杂应用场景下的翻译效果。详情请参见[翻译能力（Qwen-MT）](https://help.aliyun.com/zh/model-studio/machine-translation)。 +`qwen3.7-plus-2026-05-26` -文生文 +Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 -2026-01-12 +文本生成、深度思考 + +2026-05-21 全球 -qwen-flash、qwen-flash-2025-07-28 +`qwen3.7-max` -千问系列速度最快、成本极低的模型，适合简单任务。模型列表 +`qwen3.7-max-2026-05-20` -## 日本（东京） +Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 -**模型类型** +文本生成、深度思考 -**时间** +2026-05-11 -**服务部署范围** +全球 -**模型规格** +`deepseek-v4-pro` -**功能说明** +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 -文生文 +文本生成、深度思考 -2026-07-14 +2026-05-11 全球 -qwen-plus-character +`deepseek-v4-flash` + +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 -千问系列角色扮演模型，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。[角色扮演（Qwen-Character）](https://help.aliyun.com/zh/model-studio/role-play) +视频生成 -推理模型 +2026-04-26 -2026-06-18 +全球 -日本 +`happyhorse-1.0-video-edit` -qwen3.7-plus、qwen3.7-plus-2026-05-26 +HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。 -千问3.7Plus系列，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 +文本生成、深度思考、视觉理解 -推理模型 +2026-04-17 -2026-06-18 +全球 -日本 +`qwen3.6-flash` -deepseek-v4-pro、deepseek-v4-flash +`qwen3.6-flash-2026-04-16` -DeepSeek-V4系列模型，阿里直供。deepseek-v4-pro为旗舰模型，deepseek-v4-flash为轻量级高速模型。[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api) +Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。 -推理模型 +文本生成、深度思考 -2026-06-18 +2026-04-14 全球 -qwen3.7-plus、qwen3.7-plus-2026-05-26 +`glm-5.1` -千问3.7Plus系列，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。 +GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 -推理模型 +视频生成 -2026-06-18 +2026-04-03 全球 -qwen3.7-max、qwen3.7-max-2026-05-20 +`wan2.7-videoedit` -Qwen Max 系列新一代旗舰模型。仅支持纯文本输入，默认开启思考模式，支持显式缓存，在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。 +Wan2.7-VideoEdit，自然语言指令编辑视频，支持局部或全局编辑，可参考图像替换视频元素，支持复刻视频动作、特效、运镜等动态过程。 -推理模型 +视频生成 -2026-06-18 +2026-04-03 全球 -glm-5.1 +`wan2.7-t2v` -智谱GLM-5.1模型，专为长程任务设计，支持 200K 上下文，最大输出可达 128K Token。通过强大的逻辑推理、长文本理解及代码生成能力，在多项基准测试中表现优异，适用于智能交互、企业应用及开发辅助等场景。[GLM-阿里云](https://help.aliyun.com/zh/model-studio/glm) +Wan2.7-T2V，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。 -文生文与视觉理解 +视频生成 -2026-06-18 +2026-04-03 全球 -kimi-k2.5 +`wan2.7-r2v` -由月之暗面（Moonshot AI）公司推出的视觉理解模型，在代码生成、视觉理解等通用智能任务中表现突出。同时支持图像、视频与文本输入、对话与 Agent 任务。[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api) +Wan2.7-R2V，更加稳定的角色、道具与场景参考，支持最大5个图/视频混合参考，支持音频音色参考，搭配基础能力升级实现更强表演能力。 -推理模型 +视频生成 -2026-06-18 +2026-04-03 全球 -deepseek-v4-pro、deepseek-v4-flash +`wan2.7-i2v` -DeepSeek-V4系列模型，阿里直供。deepseek-v4-pro为旗舰模型，deepseek-v4-flash为轻量级高速模型。[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api) +万相2.7-图生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。 -推理模型 +图片生成 -2026-06-18 +2026-04-01 全球 -qwen3.6-flash、qwen3.6-flash-2026-04-16 +`wan2.7-image` -Qwen3.6 原生视觉语言 Flash 系列模型，在整体性能上较 Qwen3.5-Flash 显著提升。重点增强了智能体编程能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；在视觉能力方面，空间智能显著增强，其中物体定位和目标检测表现尤为突出。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +万相2.7-图像生成与编辑，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现 -推理模型 +文本生成、深度思考、视觉理解 -2026-06-18 +2026-04-01 全球 -qwen3.6-plus、qwen3.6-plus-2026-04-02 +`qwen3.6-plus` + +`qwen3.6-plus-2026-04-02` -千问3.6-Plus，代码开发能力重点升级（Agentic Coding、前端编程等），Vibe Coding体验显著提升；泛化场景推理能力进一步增强；多模态方面，万物识别、OCR、物体定位等能力显著提升；同时修复了Qwen3.5-Plus上线后的已知问题。使用方法与qwen3.5-plus一致。[概述](https://help.aliyun.com/zh/model-studio/text-generation) +Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/content-security.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/content-security.md index 9244e72f..e7b5761d 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/content-security.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/content-security.md @@ -327,6 +327,41 @@ curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation 结果查询页面包含筛选栏（支持按条件、文本内容、时间范围搜索）和结果表格。表格列包括**文本内容**、**服务Service**、**风险等级**、**返回标签（释义）**、**反馈结果**、**请求时间**和**操作**。示例中两条记录分别对应`bailian_query_check`（请求检查）和`bailian_response_check`（响应检查）服务，均被标记为**高风险**，返回标签为`contraband_act（疑似违禁行为）:100`，每行可单击**详情**或**反馈**进行操作。 +## 知识库检索场景安全拦截 + +使用知识库检索问答功能时，上传到知识库的文档内容经检索后会作为上下文注入模型输入。如果知识库文档中包含违规或敏感内容，这些内容同样会触发 AI 安全护栏，导致请求被拦截并返回 `DataInspectionFailed` 错误。 + +如果您在知识库检索场景中遇到安全策略拦截，建议按以下步骤排查： + +1. 检查上传到知识库的文档内容，确认是否包含政治敏感、涉黄涉暴、违法不良、个人隐私数据或可能诱导违规的表述。 + +2. 简化或改写应用的提示词，减少可能触发安全拦截的表述。 + +3. 如问题仍未解决，收集完整报错信息（HTTP 状态码、错误码、错误消息），提交工单申请加白，具体流程见下方[内容加白申请流程](#section-whitelist-apply)。 + + +## 内容加白申请流程 + +如果您的业务场景需要豁免特定内容的安全拦截，可通过提交工单申请内容加白。安全部门审核通过后，指定内容将被添加至豁免名单。 + +申请加白前，请准备以下信息： + +- HTTP 状态码（如 `400`） + +- 错误码（如 `DataInspectionFailed` 或 `data_inspection_failed`） + +- 完整错误消息（如 `Input data may contain inappropriate content.`） + + +**申请步骤**： + +1. 收集上述完整报错信息及问题复现场景描述（包括知识库 ID、调用的模型名称、触发拦截的操作）。 + +2. 提交工单，在工单中提供报错信息、使用场景及需要加白的具体内容。 + +3. 等待安全部门完成审核，审核通过后加白处理生效。 + + ## 计费说明 在百炼控制台开通 AI 安全护栏产品的 SLR 授权，并通过百炼配置启用该产品策略后，系统将根据实际调用量计费。计费方式为按 Token 数量后付费，每日费用按当日实际使用量结算；未调用服务时不产生费用。计费规则详见[计费概述](https://help.aliyun.com/zh/document_detail/2872706.html#0529545b91g7c)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/permission-management-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/permission-management-overview.md index e0f5e9f2..9c33367e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/permission-management-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/permission-management-overview.md @@ -125,7 +125,7 @@ API Key 管理 ## **业务空间权限管理** -百炼按地理区域划分资源和业务空间，**单个业务空间不能跨地域存在。即使各个地域的默认业务空间，也是不同的空间**。点击前往全局管理菜单（[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management) | [新加坡](https://modelstudio.console.aliyun.com/?tab=globalset#/efm/business_management) | [弗吉尼亚](https://modelstudio.console.aliyun.com/us-east-1?tab=globalset#/efm/business_management) | [法兰克福](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management)）。 +百炼按地理区域划分资源和业务空间，**单个业务空间不能跨地域存在。即使各个地域的默认业务空间，也是不同的空间**。单击前往全局管理菜单（[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management) | [新加坡](https://modelstudio.console.aliyun.com/?tab=globalset#/efm/business_management) | [弗吉尼亚](https://modelstudio.console.aliyun.com/us-east-1?tab=globalset#/efm/business_management) | [法兰克福](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management)）。 同时百炼的业务空间是进行**精细化权限管理**的**最小**管理单元，它可管理： @@ -205,7 +205,7 @@ API Key 管理 > 默认业务空间无法设置此限制，所有模型均可调用，且无法限流。 - 在**模型列表**中，通过**模型调用**列的开关控制模型授权状态，并在**当前空间限流**列分别设置请求数限流值与Token限流值及对应的时间单位。 + 在**模型列表**中，通过**模型调用**列的开关控制模型授权状态，并在**当前空间限流**列分别设置请求数限流值与Token限流值及对应的时间单位（支持 1 秒、2 秒、3 秒、5 秒、10 秒、30 秒、60 秒）。未设置限流时，该空间将共享账号级别的总限流配额。 - **限制模型训练**：管理某个模型可否在该业务空间进行调优（通过控制台和API）和调优后部署**。** @@ -219,14 +219,14 @@ API Key 管理 在**模型列表**页签中，**模型授权**区域的**模型部署**列显示各模型的部署授权状态，通过 toggle 开关可将状态切换为**已授权**或**未授权**。 -- **用户（账号）控制台权限管理**：管理某个 RAM 用户是否能使用该业务空间**控制台**的功能以及能使用该业务空间控制台的哪些功能。但无法限制归属该用户的 API Key 的调用。 +- **用户（账号）控制台权限管理**：管理某个 RAM 用户是否能使用该业务空间**控制台**的功能以及能使用该业务空间控制台的哪些功能。控制台页面权限不影响该用户通过 API Key 调用模型（API Key 的权限由所属业务空间的模型授权决定）。 > 阿里云主账号无须设置，可以访问所有业务空间的所有页面。 在**编辑权限**弹窗中，选择**模型**页签，在权限列表中找到并勾选**模型体验-操作**权限项。左侧导航菜单中的**模型体验**与权限列表中的**模型体验-操作**相对应。 -### **API-Key 权限** +### **API Key 权限** 单个 API Key 只能归属一个地域内的一个业务空间和一个用户，且不能转移给其他业务空间或其他用户。API Key 的可调用的功能和模型限流与**归属业务空间**的权限保持一致**，**不受**用户（账号）控制台权限管理**的影响，也无需为不同模型（如文生文、文生图、语音合成）创建不同的API Key。 @@ -268,9 +268,9 @@ API Key 的状态随归属用户（账号）操作的变化： **华北2（北京）**地域的 API Key 支持设置。 -**管理 API-Key**：可以通过百炼控制台**左侧导航栏**中的**权限管理**页签内，为 RAM 用户添加 API-Key 权限。赋予对应 RAM 用户**创建、删除、查看该空间下所有 API-Key** 的权限。 +**管理 API Key**：可以通过百炼控制台**左侧导航栏**中的**权限管理**页签内，为 RAM 用户添加 API Key 权限。赋予对应 RAM 用户**创建、删除、查看该空间下所有 API Key** 的权限。 -在**编辑权限**弹窗中，切换到**其他**页签，勾选**API-Key**即可。 +在**编辑权限**弹窗中，切换到**其他**页签，勾选**API Key**即可。 ### **OpenAPI 接口权限** @@ -280,7 +280,7 @@ RAM 用户默认无权调用百炼**应用**的数据、知识库、Prompt工程 - [AliyunBailianDataFullAccess](https://help.aliyun.com/zh/ram/developer-reference/aliyunbailiandatafullaccess)：可调用百炼应用 [API目录](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-dir/)下的所有API。 -- [AliyunBailianDataReadOnlyAccess](https://help.aliyun.com/zh/ram/developer-reference/aliyunbailiandatareadonlyaccess)：可调用百炼应用 [API目录](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-dir/)下的**只读类**API，例如[DescribeFile - 查询文件状态](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-describefile)、[GetIndexJobStatus - 查询知识库创建任务状态](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexjobstatus)等。 +- [AliyunBailianDataReadOnlyAccess](https://help.aliyun.com/zh/ram/developer-reference/aliyunbailiandatareadonlyaccess)：可调用百炼应用 [API目录](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-dir/)下的**只读类**API，例如[查询文件状态](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-describefile)、[查询知识库创建任务状态](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexjobstatus)等。 ### **应用于生产环境** @@ -305,7 +305,7 @@ RAM 用户默认无权调用百炼**应用**的数据、知识库、Prompt工程 - 将主账号总配额按比例分配给各业务空间，并预留一部分作为缓冲，以应对突发流量。 - **示例：**账号总配额为 1000 QPM，分配方案如下： + **示例：**账号总配额为 1000 QPM，分配方案。 - `project-prod-workspace`: 600 QPM (60%) @@ -673,9 +673,9 @@ AliyunBSSOrderAccess 2. **批量推理-操作** 权限，用于支持 [批量推理](https://help.aliyun.com/zh/model-studio/batch-inference)功能。 - 3. **模型观测-操作** 权限，用于查看模型调用、评测的 Token 消耗量。 + 3. **模型监控-操作** 权限，用于查看模型调用、评测的 Token 消耗量。 -3. **若需要通过百炼的 API 调用，**需要为 RAM 用户在对应业务空间创建或分配 API Key，更多细节请参考本文的：[API-Key 权限](#f2704153a055r)。（需要超级管理员或业务空间管理员操作） +3. **若需要通过百炼的 API 调用，**需要为 RAM 用户在对应业务空间创建或分配 API Key，更多细节请参考本文的：[API Key 权限](#f2704153a055r)。（需要超级管理员或业务空间管理员操作） ### **设置控制台模型调优权限** @@ -696,14 +696,14 @@ AliyunBSSOrderAccess 6. **数据管理-操作** 权限，用于管理调优数据集。 - 7. **模型观测-操作** 权限，用于查看模型调用、评测的 Token 消耗量。 + 7. **模型监控-操作** 权限，用于查看模型调用、评测的 Token 消耗量。 ### **设置 API 模型调优权限** 1. 若不使用**默认业务空间**，需保证业务空间为特定模型开通了[模型调优（训练）](#c180b853793v3)权限。（需要超级管理员操作） -2. 为 RAM 用户在对应业务空间创建或分配 API Key，更多细节请参考本文的：[API-Key 权限](#f2704153a055r)。（需要超级管理员或业务空间管理员操作） +2. 为 RAM 用户在对应业务空间创建或分配 API Key，更多细节请参考本文的：[API Key 权限](#f2704153a055r)。（需要超级管理员或业务空间管理员操作） ## **常见问题** @@ -714,8 +714,27 @@ AliyunBSSOrderAccess ### **2\. 如何使用子业务空间调用模型？** -无需特殊设置，使用子业务空间的 API-Key 即可。 +无需特殊设置，使用子业务空间的 API Key 即可。 ### **3\. 如何使用特定业务空间的应用？** 使用 API 管理、调用特定业务空间的应用需要同时设置 [APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 + +### **4\. 已配置 AliyunBailianFullAccess 但仍无法管理业务空间怎么办？** + +RAM 权限（如 AliyunBailianFullAccess）与百炼平台内部的业务空间级别权限是两套独立的权限体系。RAM 权限仅控制对百炼控制台的访问权限，而业务空间内的管理权限（如编辑空间、管理成员等）需要由主账号或超级管理员在百炼控制台中单独授权。 + +如果子账号已拥有 AliyunBailianFullAccess 权限但仍提示无权限操作业务空间，请按以下步骤处理： + +1. 使用主账号登录百炼控制台。 + +2. 单击右上角设置图标，在左侧导航栏中选择**账号管理**。 + +3. 找到需要授权的子账号，单击**权限管理**。 + +4. 为该子账号添加对应业务空间的管理权限。 + + +### **5\. 如何删除业务空间？** + +访问[阿里云百炼控制台](https://bailian.console.aliyun.com/cn-beijing?spm=5176.29619931.J__Z58Z6CX7MY__Ll8p1ZOR.1.7dd7521cmX1pAh&tab=model#/model-market)，在页面右上角选择目标地域，进入[业务空间管理](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)页面，在操作列选择删除，**删除后不可恢复**。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/after-sales-service-scope.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/after-sales-service-scope.md new file mode 100644 index 00000000..54844525 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/after-sales-service-scope.md @@ -0,0 +1,57 @@ +# 阿里云百炼平台售后服务范围说明 + +## **阿里云百炼平台售后服务范围说明** + +欢迎您使用阿里云百炼。本《**阿里云百炼平台售后服务范围说明》是对您使用阿里云百炼相关产品和服务时适用的售后服务范围的说明。** + +1\. 在您购买的服务期限内，我们将为您提供如下售后基础服务，即通过官网、电话及阿里云APP提供7×24的电话咨询（95187、400电话）、智能在线和标准工单支持。支持范围包括： + +（1）关于阿里云百炼模型服务与产品功能、架构的咨询； + +（2）使用、配置阿里云百炼模型服务的最佳实践； + +（3）阿里云百炼模型服务的使用咨询、技术问题及故障诊断； + +（4）阿里云百炼API及阿里云百炼官方SDK问题的故障诊断； + +（5）与阿里云百炼管理控制台相关的问题； + +（6）与阿里云相关的账号问题咨询支持； + +（7）与阿里云相关的财务、合同及计费问题的咨询支持。 + +2\. 阿里云百炼将以阿里云官网页面公布的[客户服务权益](https://www.aliyun.com/service/customer-service-benefits?spm=5176.support-home.J_3451238410.1.12d1156fPBBxO0)标准向您提供相应的售后服务支持。 + +3\. 阿里云百炼同时提供付费版的售后增值服务（包括支持计划等），该等服务需在阿里云官网订购后生效使用。您还可通过阿里云获得其他付费的售后服务，具体详见阿里云的网站相关页面的收费售后服务内容。如您的项目需要更深度的技术支持（如业务代码编写指导、定制化集成方案等），建议联系阿里云商务经理沟通定制化服务方案。 + +4\. 为了方便您的生产或使用，如您选择将阿里云百炼服务与外部（非阿里云百炼平台上的）第三方工具或产品进行对接，阿里云将尽商业上合理的努力为您提供第三方工具在接入阿里云百炼模型推理服务过程中的方向性建议，但针对非阿里云百炼服务相关的问题，我们无法提供专业意见。 + +（1）我们提供建议的范围包括： + +（i）确认阿里云百炼API接口及服务端的可用状态； + +（ii）阿里云百炼官方API调用示例及SDK使用说明参考； + +（iii）协助核查阿里云百炼服务端调用明细和计费记录； + +（iv）基本的连通性测试建议（如通过curl等标准工具测试阿里云百炼服务地址的可达性）。 + +**（2）我们提供建议的范围不包括：** + +（i）第三方工具（如Cursor、Windsurf、Cline、OpenClaw等）的安装、部署、配置、升级及日常使用指导； + +（ii）第三方工具的产品功能、交互逻辑及内部实现问题的排查； + +（iii）其他云厂商、企业或社区提供的产品或服务的配置与运维； + +（iv）用户业务代码的编写、调试与实现； + +（v）用户本地环境（含内网、代理、VPN、防火墙、操作系统等）导致的连通性或兼容性问题的排查； + +（vi）第三方工具内部显示的Token数量、费用预估值或调用统计与阿里云计费数据之间的差异解释； + +（vii）所有第三方工具的安装、补丁更新、测试、故障诊断、优化等日常运维服务； + +（viii）基于阿里云百炼模型服务原生能力之上的第三方自建业务相关支持。 + +5\. 但请您注意，除我们另有书面说明外，第三方工具不构成我们的代理、受托或联合服务主体，**我们不对外部第三方工具的任何陈述、承诺或行为承担责任。**您知悉并确认，阿里云仅负责阿里云百炼平台自身的运营维护，即百炼服务端的技术架构、API接口、计量计费系统、控制台功能等；**第三方工具的运行维护（如AI编程工具的安装配置、开源代理框架的部署调优等）由您及相应工具提供方负责。**当您的问题出现在百炼模型服务的使用过程中，但其原因、责任范围或依赖关系已超出阿里云百炼平台本身可直接提供支持和保障的范围时——通常涉及您侧系统、外部第三方服务、网络环境、账号权限、业务流程或非标集成，阿里云将协助进行初步排查。若经排查确认问题来源于非阿里云侧，阿里云将给予方向性建议并引导您联系相应的服务主体。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md index fff36d4d..53d67f33 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md @@ -208,6 +208,10 @@ 等待时间取决于您的具体限流值（RPS/RPM）。例如，如果您的限流是120 RPM（每分钟查询数），即每秒2次请求。如果您在0.2秒内连续提交了2次请求，第3次请求就会被限流，您需要等待大约0.8秒后才能再次成功提交。 +17. **qwen-plus-latest 这个模型具体对应哪个系列？是 Qwen3.7 还是 Qwen3.5？** + + qwen-plus-latest 是 qwen-plus 的最新版本，属于 Qwen3 系列，而非 Qwen3.5 或 Qwen3.7 系列。另外 Qwen3.5、Qwen3.7 等是独立的模型系列，与 Qwen3 系列并列，并非 Qwen3 的子版本。 + ## **模型幻觉问题** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-3d-generation/tripo-h3-1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-3d-generation/tripo-h3-1.md new file mode 100644 index 00000000..90ca2526 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-3d-generation/tripo-h3-1.md @@ -0,0 +1,207 @@ +# Tripo/Tripo-H3.1 + +Tripo H3.1 是 Tripo 推出的高精度 3D 生成模型，专为需要极致视觉质量与细节表现的创作者设计。模型通过核心算法升级与模块优化，参数规模达 200 亿级，支持十亿体素级三维分辨率与最高 200 万面多边形生成。在保持高精度几何与真实纹理的同时，Tripo H3.1 对输入参考图的还原度与对齐度进一步提升，在角色形体、面部细节与几何文字等复杂结构上实现更稳定、细致的表达，适用于高质量视觉制作与 3D 打印等高精度资产生产场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**3D-Generation** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +文生3D（标准版+无贴图） + +0.7 + +每次 + +单图生3D（标准版+无贴图） + +1.4 + +每次 + +多图生3D（标准版+无贴图） + +1.4 + +每次 + +文生3D（标准版+带标清贴图） + +1.4 + +每次 + +单图生3D（标准版+带标清贴图） + +2.1 + +每次 + +多图生3D（标准版+带标清贴图） + +2.1 + +每次 + +文生3D（标准版+带高清贴图） + +2.1 + +每次 + +单图生3D（标准版+带高清贴图） + +2.8 + +每次 + +多图生3D（标准版+带高清贴图） + +2.8 + +每次 + +文生3D（超清版+无贴图） + +2.1 + +每次 + +单图生3D（超清版+无贴图） + +2.8 + +每次 + +多图生3D（超清版+无贴图） + +2.8 + +每次 + +文生3D（超清版+带标清贴图） + +2.8 + +每次 + +单图生3D（超清版+带标清贴图） + +3.5 + +每次 + +多图生3D（超清版+带标清贴图） + +3.5 + +每次 + +文生3D（超清版+带高清贴图） + +3.5 + +每次 + +单图生3D（超清版+带高清贴图） + +4.2 + +每次 + +多图生3D（超清版+带高清贴图） + +4.2 + +每次 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +5 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-3d-generation/tripo-p1-0.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-3d-generation/tripo-p1-0.md new file mode 100644 index 00000000..25a22d33 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-3d-generation/tripo-p1-0.md @@ -0,0 +1,153 @@ +# Tripo/Tripo-P1.0 + +Tripo P1.0 是面向实时应用与生产管线的 3D 生成模型，专为需要干净拓扑和引擎可用网格的开发者与创作者设计。模型可在约 2 秒内生成具备专业级拓扑结构的 3D 资产，适用于游戏、Web3D 与各类实时交互场景。针对 UGC 内容生产中对“速度”和“开箱即用”的需求，Tripo P1.0 在保证质量的同时大幅提升生成效率，使资产能够快速接入实时引擎与开发流程。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**3D-Generation** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +文生3D（无贴图） + +2.1 + +每次 + +单图生3D（无贴图） + +2.8 + +每次 + +多图生3D（无贴图） + +2.8 + +每次 + +文生3D（带标清贴图） + +2.8 + +每次 + +单图生3D（带标清贴图） + +3.5 + +每次 + +多图生3D（带标清贴图） + +3.5 + +每次 + +文生3D（带高清贴图） + +3.5 + +每次 + +单图生3D（带高清贴图） + +4.2 + +每次 + +多图生3D（带高清贴图） + +4.2 + +每次 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +5 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/gte-rerank-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/gte-rerank-v2.md new file mode 100644 index 00000000..22abd779 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/gte-rerank-v2.md @@ -0,0 +1,109 @@ +# gte-rerank-v2 + +gte-rerank-v2是通义实验室研发的多语言文本统一排序模型，面向全球多个主流语种，提供高水平的文本排序服务。通常用于语义检索、RAG等场景，可以简单、有效地提升文本检索的效果。给定查询 (Query) 和一系列候选文本 (documents)，模型会根据与查询的语义相关性从高到低对候选文本进行排序。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30000 + +最大输出长度 + +0 + +上下文长度 + +30000 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +文本输入 + +0.8 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +5040 + +TPM（每分钟tokens） + +4,980,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/multimodal-embedding-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/multimodal-embedding-v1.md new file mode 100644 index 00000000..fa0fcc79 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/multimodal-embedding-v1.md @@ -0,0 +1,115 @@ +# multimodal-embedding-v1 + +通义实验室基于预训练多模态大模型构建的多模态向量模型。该模型根据用户的输入生成高维连续向量，这些输入可以是文本、图片或视频。多模态向量在可应用于图片搜索、文搜图、视频搜索、图片分类和视频内容审核等下游任务中。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +512 + +最大输出长度 + +0 + +上下文长度 + +0 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片输入 + +0.9 + +每百万tokens + +文本输入 + +0.7 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen2-5-vl-embedding.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen2-5-vl-embedding.md new file mode 100644 index 00000000..368c9f7f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen2-5-vl-embedding.md @@ -0,0 +1,115 @@ +# qwen2.5-vl-embedding + +基于Qwen2.5-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片输入 + +1.8 + +每百万tokens + +文本输入 + +0.7 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +TPM（每分钟tokens） + +600,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-7-text-embedding.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-7-text-embedding.md new file mode 100644 index 00000000..38595a34 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-7-text-embedding.md @@ -0,0 +1,109 @@ +# qwen3.7-text-embedding + +是通义实验室基于Qwen3.7训练的多语言文本统一向量模型，相较text-embedding-v4版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升20%；支持256~2560维用户自定义向量维度。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +131072 + +最大输出长度 + +— + +上下文长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +文本输入 + +0.5 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +24000 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-rerank.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-rerank.md new file mode 100644 index 00000000..d0f6e41d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-rerank.md @@ -0,0 +1,141 @@ +# qwen3-rerank + +基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐开源Qwen3-Rerank系列模型 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30000 + +最大输出长度 + +0 + +上下文长度 + +30000 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +文本输入 + +0.5 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +文本输入 + +0.749 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +5400 + +TPM（每分钟tokens） + +5,000,000,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +5400 + +TPM（每分钟tokens） + +5,000,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-vl-embedding.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-vl-embedding.md new file mode 100644 index 00000000..dd65f4d8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-vl-embedding.md @@ -0,0 +1,115 @@ +# qwen3-vl-embedding + +基于Qwen3-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**—** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +32000 + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片输入 + +1.8 + +每百万tokens + +文本输入 + +0.7 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +2400 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-vl-rerank.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-vl-rerank.md new file mode 100644 index 00000000..06444037 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/qwen3-vl-rerank.md @@ -0,0 +1,115 @@ +# qwen3-vl-rerank + +Qwen3-VL-Rerank重排模型，它能够深入理解文本、图片、视频的丰富多模态信息。在初步检索获得结果后，Qwen3-VL-Rerank 能够运用其先进的跨模态关联能力，对候选项目进行智能化的二次排序，将最相关的结果置于显要位置。通用用于提升跨模态搜索的准确率、优化图搜和视频检索的精准度、辅助图像聚类的分组质量、以及实现复杂多模态信息的高效检索和精确打标。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +120000 + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片输入 + +1.8 + +每百万tokens + +文本输入 + +0.7 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +TPM（每分钟tokens） + +9,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-async-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-async-v1.md new file mode 100644 index 00000000..91fac5d0 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-async-v1.md @@ -0,0 +1,93 @@ +# text-embedding-async-v1 + +通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +文本输入 + +0.7 + +每百万tokens diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-async-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-async-v2.md new file mode 100644 index 00000000..7a9d3d98 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-async-v2.md @@ -0,0 +1,93 @@ +# text-embedding-async-v2 + +通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +文本输入 + +0.7 + +每百万tokens diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v1.md new file mode 100644 index 00000000..8e422ae0 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v1.md @@ -0,0 +1,115 @@ +# text-embedding-v1 + +通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +向量输入（Batch File） + +0.35 + +每百万tokens + +文本输入 + +0.7 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1800 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v2.md new file mode 100644 index 00000000..fbe32319 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v2.md @@ -0,0 +1,115 @@ +# text-embedding-v2 + +通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +向量输入（Batch File） + +0.35 + +每百万tokens + +文本输入 + +0.7 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1800 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v3.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v3.md new file mode 100644 index 00000000..e6292896 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v3.md @@ -0,0 +1,147 @@ +# text-embedding-v3 + +通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +向量输入（Batch File） + +0.25 + +每百万tokens + +文本输入 + +0.5 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +文本输入 + +0.514 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1800 + +TPM（每分钟tokens） + +1,200,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +6000 + +TPM（每分钟tokens） + +24,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v4.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v4.md new file mode 100644 index 00000000..a405f587 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/text-embedding-v4.md @@ -0,0 +1,147 @@ +# text-embedding-v4 + +通用文本向量V4版本，是通义实验室基于Qwen3训练的多语言文本统一向量模型，相较V3版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升15%~40%；支持64~2048维用户自定义向量维度。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +向量输入（Batch File） + +0.25 + +每百万tokens + +文本输入 + +0.5 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +文本输入 + +0.514 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1800 + +TPM（每分钟tokens） + +1,200,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +1800 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/tongyi-embedding-vision-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/tongyi-embedding-vision-flash.md new file mode 100644 index 00000000..a4f2ae78 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/tongyi-embedding-vision-flash.md @@ -0,0 +1,271 @@ +# tongyi-embedding-vision-flash + +Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，在视觉向量化上具备极高性价比。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片输入 + +0.15 + +每百万tokens + +文本输入 + +0.15 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片输入 + +0.2 + +每百万tokens + +文本输入 + +0.6 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +TPM（每分钟tokens） + +200,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +600 + +TPM（每分钟tokens） + +200,000 + +## 快照版本 + +### tongyi-embedding-vision-flash-2026-03-06 + +Tongyi-Embedding-Vision是基于LLM底座的视觉多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文、以文搜图文等下游多样化任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，具备极高性价比。 2026-03-06版本在保留极致性价比优势的同时，基于Qwen3底座实现了效果与功能全面升级，包括全场景性能提升、多分辨率模式/多向量维度/多语言能力/融合向量等能力的支持。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片输入 + +0.15 + +每百万tokens + +文本输入 + +0.15 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +TPM（每分钟tokens） + +9,600,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/tongyi-embedding-vision-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/tongyi-embedding-vision-plus.md new file mode 100644 index 00000000..620504a1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-embedding-reranking/tongyi-embedding-vision-plus.md @@ -0,0 +1,271 @@ +# tongyi-embedding-vision-plus + +Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片输入 + +0.5 + +每百万tokens + +文本输入 + +0.5 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片输入 + +0.6 + +每百万tokens + +文本输入 + +0.6 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +TPM（每分钟tokens） + +200,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +600 + +TPM（每分钟tokens） + +200,000 + +## 快照版本 + +### tongyi-embedding-vision-plus-2026-03-06 + +Tongyi-Embedding-Vision是基于LLM底座的视觉多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文、以文搜图文等下游多样化任务场景。 2026-03-06版本在保留极致性价比优势的同时，基于Qwen3底座实现了效果与功能全面升级，包括全场景性能提升、多分辨率模式/多向量维度/多语言能力/融合向量等能力的支持。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片输入 + +0.5 + +每百万tokens + +文本输入 + +0.5 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +TPM（每分钟tokens） + +9,600,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-parsing-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-parsing-v1.md new file mode 100644 index 00000000..187a3545 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-parsing-v1.md @@ -0,0 +1,93 @@ +# aitryon-parsing-v1 + +图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片检测 + +0.004 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-plus.md new file mode 100644 index 00000000..3e62e864 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-plus.md @@ -0,0 +1,93 @@ +# aitryon-plus + +aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-refiner.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-refiner.md new file mode 100644 index 00000000..0a12c174 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon-refiner.md @@ -0,0 +1,179 @@ +# aitryon-refiner + +图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +**图片生成数量<=25** + +计费项 + +价格（元） + +单位 + +图片生成 + +0.3 + +每张 + +**25<图片生成数量<=125** + +计费项 + +价格（元） + +单位 + +图片生成 + +0.275 + +每张 + +**125<图片生成数量<=250** + +计费项 + +价格（元） + +单位 + +图片生成 + +0.25 + +每张 + +**250<图片生成数量<=1250** + +计费项 + +价格（元） + +单位 + +图片生成 + +0.225 + +每张 + +**1250<图片生成数量<=2500** + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +**2500<图片生成数量<=25000** + +计费项 + +价格（元） + +单位 + +图片生成 + +0.175 + +每张 + +**25000<图片生成数量** + +计费项 + +价格（元） + +单位 + +图片生成 + +0.15 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon.md new file mode 100644 index 00000000..d2f7596c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/aitryon.md @@ -0,0 +1,93 @@ +# aitryon + +aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/facechain-facedetect.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/facechain-facedetect.md new file mode 100644 index 00000000..05289818 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/facechain-facedetect.md @@ -0,0 +1,79 @@ +# facechain-facedetect + +对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**—** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/image-erase-completion.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/image-erase-completion.md new file mode 100644 index 00000000..9292ad58 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/image-erase-completion.md @@ -0,0 +1,79 @@ +# image-erase-completion + +图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/image-out-painting.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/image-out-painting.md new file mode 100644 index 00000000..27667566 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/image-out-painting.md @@ -0,0 +1,79 @@ +# image-out-painting + +图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/kling-v3-image-generation.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/kling-v3-image-generation.md new file mode 100644 index 00000000..e65e80f1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/kling-v3-image-generation.md @@ -0,0 +1,111 @@ +# kling/kling-v3-image-generation + +支持最多10张参考图，可锁定主体、元素和色调，保证风格一致。融合风格转绘、人像/角色参考、多图融合及局部重绘，操作灵活。人像细节真实，整体画面细腻丰富，色彩氛围兼具影视感。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Text** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成（1K） + +0.2 + +每张 + +图片生成（2K） + +0.2 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/kling-v3-omni-image-generation.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/kling-v3-omni-image-generation.md new file mode 100644 index 00000000..eaa4867e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/kling-v3-omni-image-generation.md @@ -0,0 +1,117 @@ +# kling/kling-v3-omni-image-generation + +解锁影视级叙事画面，新增系列组图及2K/4K直出。深度解析提示词视听元素，精确响应创作指令。支持自由多参考图及全面效果升级，适合分镜、剧情概念图及场景设定。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Text** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成（1K） + +0.2 + +每张 + +图片生成（2K） + +0.2 + +每张 + +图片生成（4K） + +0.4 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-facechain-generation.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-facechain-generation.md new file mode 100644 index 00000000..9965c8a3 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-facechain-generation.md @@ -0,0 +1,93 @@ +# facechain-generation + +基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.18 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-image-instance-segmentation.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-image-instance-segmentation.md new file mode 100644 index 00000000..b6357bac --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-image-instance-segmentation.md @@ -0,0 +1,79 @@ +# image-instance-segmentation + +人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-qwen-image-edit.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-qwen-image-edit.md new file mode 100644 index 00000000..23603d8d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/model-qwen-image-edit.md @@ -0,0 +1,133 @@ +# qwen-image-edit + +千问系列首个图像编辑模型，成功将Qwen-Image的文本渲染能力拓展到编辑任务上。支持精准的中英双语文字编辑、视觉外观与语义双重编辑、具备强大的跨基准性能表现。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.3 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.330266 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-2-0-pro.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-2-0-pro.md new file mode 100644 index 00000000..02dbc8a2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-2-0-pro.md @@ -0,0 +1,537 @@ +# qwen-image-2.0-pro + +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。该模型版本功能等同于快照模型 qwen-image-2.0-pro-2026-04-22。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.550443 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +2 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +2 + +## 快照版本 + +### qwen-image-2.0-pro-2026-06-22 + +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.550443 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +2 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +2 + +### qwen-image-2.0-pro-2026-04-22 + +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合。相较于3月3日快照，本模型在画面质感，尤其是纹理细节、光影、材质上有明显跃升；支持多语言的图内文字生成；艺术风格表现更加均衡。该版本为2026年4月22日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.550443 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +2 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +2 + +### qwen-image-2.0-pro-2026-03-03 + +Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。该版本为2026年3月3日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.550443 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +2 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +2 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-2-0.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-2-0.md new file mode 100644 index 00000000..e9323f34 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-2-0.md @@ -0,0 +1,269 @@ +# qwen-image-2.0 + +Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。该模型版本功能等同于快照模型 qwen-image-2.0-2026-03-03。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.256873 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 快照版本 + +### qwen-image-2.0-2026-03-03 + +Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。该版本为2026年3月3日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.256873 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-3-0-pro.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-3-0-pro.md new file mode 100644 index 00000000..9ba4fec6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-3-0-pro.md @@ -0,0 +1,103 @@ +# qwen-image-3.0-pro + +内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。 细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。 知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。 Qwen-Image-3.0-Pro 不只是在追求"好看"，更在追求“好用”——让图像生成真正成为可落地的生产力工具。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +1 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-edit-max.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-edit-max.md new file mode 100644 index 00000000..1105208a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-edit-max.md @@ -0,0 +1,269 @@ +# qwen-image-edit-max + +千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。该模型版本功能等同于快照模型 qwen-image-edit-max-2026-01-16。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.550443 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +2 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +2 + +## 快照版本 + +### qwen-image-edit-max-2026-01-16 + +千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.550443 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +2 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +2 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-edit-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-edit-plus.md new file mode 100644 index 00000000..d24783b6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-edit-plus.md @@ -0,0 +1,403 @@ +# qwen-image-edit-plus + +千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 快照版本 + +### qwen-image-edit-plus-2025-12-15 + +千问系列图像编辑Plus模型，相较10月30日快照提升角色一致性、工业设计能力、几何推理能力；同时集成例如打光等Lora能力、减轻偏移问题。此版本为2025年12月15日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 + +### qwen-image-edit-plus-2025-10-30 + +千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。此版本为2025年10月30日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-max.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-max.md new file mode 100644 index 00000000..ec275344 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-max.md @@ -0,0 +1,253 @@ +# qwen-image-max + +千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。该模型版本功能等同于快照模型 qwen-image-max-2025-12-30。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.550443 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +2 + +TPM（每分钟tokens） + +1,000,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +2 + +TPM（每分钟tokens） + +1,000,000 + +## 快照版本 + +### qwen-image-max-2025-12-30 + +千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。此版本为2025年12月30日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.550443 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-plus.md new file mode 100644 index 00000000..01173503 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image-plus.md @@ -0,0 +1,273 @@ +# qwen-image-plus + +千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 快照版本 + +### qwen-image-plus-2026-01-09 + +千问系列图像生成模型，具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色。此版本为2026年1月9日快照，为Qwen-Image-Max的蒸馏加速版，可以更快速地生成高质量图片。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image.md new file mode 100644 index 00000000..3d2249d9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-image.md @@ -0,0 +1,137 @@ +# qwen-image + +千问系列首个图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.25 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.256873 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +TPM（每分钟tokens） + +1,000,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-mt-image.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-mt-image.md new file mode 100644 index 00000000..90fdd451 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/qwen-mt-image.md @@ -0,0 +1,105 @@ +# qwen-mt-image + +专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.003 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/shoemodel-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/shoemodel-v1.md new file mode 100644 index 00000000..19782b35 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/shoemodel-v1.md @@ -0,0 +1,79 @@ +# shoemodel-v1 + +鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/vidu-image-reference2image.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/vidu-image-reference2image.md new file mode 100644 index 00000000..ac70168e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/vidu-image-reference2image.md @@ -0,0 +1,117 @@ +# vidu/vidu-image_reference2image + +输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，对中英文字的精准渲染、UI/图表等设计细节的像素级还原，适合制作海报、信息图等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成（1K） + +0.625 + +每张 + +图片生成（2K） + +1 + +每张 + +图片生成（4K） + +1.46875 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq2-fast-reference2image.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq2-fast-reference2image.md new file mode 100644 index 00000000..d78c99b7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq2-fast-reference2image.md @@ -0,0 +1,105 @@ +# vidu/viduq2-fast_reference2image + +输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，语义理解能力大幅提升，支持更多风格。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成（1K） + +0.28125 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq2-pro-reference2image.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq2-pro-reference2image.md new file mode 100644 index 00000000..6114ecc1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq2-pro-reference2image.md @@ -0,0 +1,117 @@ +# vidu/viduq2-pro_reference2image + +输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，擅长处理复杂逻辑，具备超强上下文一致性和工业级稳定性。适合专业设计、漫剧制作等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成（1K） + +0.9375 + +每张 + +图片生成（2K） + +0.9375 + +每张 + +图片生成（4K） + +1.71875 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq3-fast-reference2image.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq3-fast-reference2image.md new file mode 100644 index 00000000..ba54bf3e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/viduq3-fast-reference2image.md @@ -0,0 +1,117 @@ +# vidu/viduq3-fast_reference2image + +输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，主打高速高质与低成本，成本比Pro降低约50%。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成（1K） + +0.46875 + +每张 + +图片生成（2K） + +0.78125 + +每张 + +图片生成（4K） + +1.09375 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/virtualmodel-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/virtualmodel-v2.md new file mode 100644 index 00000000..3137808f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/virtualmodel-v2.md @@ -0,0 +1,79 @@ +# virtualmodel-v2 + +虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-1-t2i-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-1-t2i-plus.md new file mode 100644 index 00000000..db20527f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-1-t2i-plus.md @@ -0,0 +1,109 @@ +# wan2.1-t2i-plus + +Wan2.1 Text-to-Image Plus version, Generate more image details. Upgraded in image beauty, realism, and artistry. Stronger semantic understanding ability, rich style generalization ability, supports up to 2 million pixel generation, supports smart prompt rewriting. + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.366962 + +每张 + +## 限流 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-1-t2i-turbo.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-1-t2i-turbo.md new file mode 100644 index 00000000..0ce22ceb --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-1-t2i-turbo.md @@ -0,0 +1,109 @@ +# wan2.1-t2i-turbo + +Wan2.1 Text-to-Image Turbo version, faster generation speed. Upgraded in image beauty, realism, and artistry. Stronger semantic understanding ability, rich style generalization ability, supports up to 2 million pixel generation, supports smart prompt rewriting. + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.183481 + +每张 + +## 限流 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-2-t2i-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-2-t2i-flash.md new file mode 100644 index 00000000..47a662c8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-2-t2i-flash.md @@ -0,0 +1,133 @@ +# wan2.2-t2i-flash + +全新升级的万相2.2文生图，更快的生成速度。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.14 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.183481 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-2-t2i-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-2-t2i-plus.md new file mode 100644 index 00000000..1e6851e5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-2-t2i-plus.md @@ -0,0 +1,133 @@ +# wan2.2-t2i-plus + +全新升级的万相2.2文生图，更丰富的画面细节。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.366962 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-5-i2i.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-5-i2i.md new file mode 100644 index 00000000..a427eb0b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-5-i2i.md @@ -0,0 +1,269 @@ +# wan2.5-i2i-preview + +万相2.5-图像编辑-Preview，全新升级模型架构。支持指令控制实现丰富的图像编辑能力，指令遵循能力进一步提升，支持高一致性保持的多图参考生成，文字生成表现优异。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 快照版本 + +### wan2.5-i2i-preview + +万相2.5-图像编辑-Preview，全新升级模型架构。支持指令控制实现丰富的图像编辑能力，指令遵循能力进一步提升，支持高一致性保持的多图参考生成，文字生成表现优异。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-5-t2i.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-5-t2i.md new file mode 100644 index 00000000..11aa6e7d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-5-t2i.md @@ -0,0 +1,269 @@ +# wan2.5-t2i-preview + +万相2.5-文生图-Preview，全新升级模型架构。画面美学、设计感、真实质感显著提升，精准指令遵循，擅长中英文和小语种文字生成，支持复杂结构化长文本和图表、架构图等内容生成。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 快照版本 + +### wan2.5-t2i-preview + +万相2.5-文生图-Preview，全新升级模型架构。画面美学、设计感、真实质感显著提升，精准指令遵循，擅长中英文和小语种文字生成，支持复杂结构化长文本和图表、架构图等内容生成。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-6-image.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-6-image.md new file mode 100644 index 00000000..607786ff --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-6-image.md @@ -0,0 +1,189 @@ +# wan2.6-image + +万相2.6-图像生成，全能图像生成模型，支持图文一体化推理生成，具备多图创意融合、商用级一致性、美学要素迁移与镜头光影精确控制，全面提升图像生成的一致性、可控性和表现力。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image**　**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +## 德国（法兰克福） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 美国（弗吉尼亚） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 德国（法兰克福） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 美国（弗吉尼亚） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-6-t2i.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-6-t2i.md new file mode 100644 index 00000000..78736bec --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-6-t2i.md @@ -0,0 +1,189 @@ +# wan2.6-t2i + +万相2.6-文生图，画面质感、美学表现、指令遵循升级，在艺术风格精准控制、真实感人像、长文本生图及广泛历史文化IP覆盖上均表现出卓越能力，可生成高质量且富有表现力的视觉内容。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.220177 + +每张 + +## 德国（法兰克福） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 美国（弗吉尼亚） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 德国（法兰克福） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +## 美国（弗吉尼亚） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +60 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-7-image-pro.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-7-image-pro.md new file mode 100644 index 00000000..bdbe604e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-7-image-pro.md @@ -0,0 +1,133 @@ +# wan2.7-image-pro + +万相2.7-图像生成与编辑旗舰版模型，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.5 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.562065 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-7-image.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-7-image.md new file mode 100644 index 00000000..c105e3ed --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wan2-7-image.md @@ -0,0 +1,161 @@ +# wan2.7-image + +万相2.7-图像生成与编辑，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.224826 + +每张 + +## 日本（东京） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +300 + +## 日本（东京） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +300 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-background-generation-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-background-generation-v2.md new file mode 100644 index 00000000..f7c4c8a0 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-background-generation-v2.md @@ -0,0 +1,93 @@ +# wanx-background-generation-v2 + +图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.08 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-poster-generation-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-poster-generation-v1.md new file mode 100644 index 00000000..418ae55c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-poster-generation-v1.md @@ -0,0 +1,79 @@ +# wanx-poster-generation-v1 + +创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-sketch-to-image-lite.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-sketch-to-image-lite.md new file mode 100644 index 00000000..b5621c86 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-sketch-to-image-lite.md @@ -0,0 +1,93 @@ +# wanx-sketch-to-image-lite + +万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.06 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-style-repaint-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-style-repaint-v1.md new file mode 100644 index 00000000..e4b64c62 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-style-repaint-v1.md @@ -0,0 +1,93 @@ +# wanx-style-repaint-v1 + +人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.12 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-v1.md new file mode 100644 index 00000000..505665a9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-v1.md @@ -0,0 +1,93 @@ +# wanx-v1 + +万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.16 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-virtualmodel.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-virtualmodel.md new file mode 100644 index 00000000..76587b56 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-virtualmodel.md @@ -0,0 +1,79 @@ +# wanx-virtualmodel + +虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-x-painting.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-x-painting.md new file mode 100644 index 00000000..0a5ebe7e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx-x-painting.md @@ -0,0 +1,79 @@ +# wanx-x-painting + +万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-0-t2i-turbo.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-0-t2i-turbo.md new file mode 100644 index 00000000..e70449c1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-0-t2i-turbo.md @@ -0,0 +1,105 @@ +# wanx2.0-t2i-turbo + +Wan2.0-T2I-Turbo，更擅长质感人像和创意设计画作生成，在图像美观度、真实感、艺术性上全面升级，支持最大200万像素生成，支持智能提示词改写等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.04 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-imageedit.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-imageedit.md new file mode 100644 index 00000000..6a1133ee --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-imageedit.md @@ -0,0 +1,93 @@ +# wanx2.1-imageedit + +万相-通义图像编辑，支持预设编辑任务与指令式编辑，包含多种局部/全图编辑能力，如图像风格化、线稿生图、局部重绘、参考图生成、图像外扩、图像超分等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.14 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-t2i-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-t2i-plus.md new file mode 100644 index 00000000..0405cd39 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-t2i-plus.md @@ -0,0 +1,105 @@ +# wanx2.1-t2i-plus + +万相2.1-文生图-Plus，更丰富的画面细节，在图像美观度、真实感、艺术性上全面升级，更强的语义理解能力、丰富的风格泛化性、支持最大200万像素生成，支持智能提示词改写等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.2 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-t2i-turbo.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-t2i-turbo.md new file mode 100644 index 00000000..a73233ff --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wanx2-1-t2i-turbo.md @@ -0,0 +1,105 @@ +# wanx2.1-t2i-turbo + +万相2.1-文生图-Turbo，更快的生成速度，在图像美观度、真实感、艺术性上全面升级，更强的语义理解能力、丰富的风格泛化性、支持最大200万像素生成，支持智能提示词改写等。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.14 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wordart-semantic.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wordart-semantic.md new file mode 100644 index 00000000..d499104c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wordart-semantic.md @@ -0,0 +1,93 @@ +# wordart-semantic + +WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.24 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wordart-texture.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wordart-texture.md new file mode 100644 index 00000000..67bfaace --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/wordart-texture.md @@ -0,0 +1,93 @@ +# wordart-texture + +WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Image** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成 + +0.08 + +每张 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/z-image-turbo.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/z-image-turbo.md new file mode 100644 index 00000000..2c4d6679 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-image-generation/z-image-turbo.md @@ -0,0 +1,145 @@ +# z-image-turbo + +Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text**　**Image** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +图片生成（标准） + +0.1 + +每张 + +图片生成（思考） + +0.2 + +每张 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +图片生成（标准） + +0.220177 + +每张 + +图片生成（思考） + +0.110089 + +每张 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +120 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen-omni-turbo-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen-omni-turbo-realtime.md new file mode 100644 index 00000000..d172c6a7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen-omni-turbo-realtime.md @@ -0,0 +1,609 @@ +# qwen-omni-turbo-realtime + +千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。该模型版本功能等同于快照模型 qwen-omni-turbo-realtime-2025-05-08。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30720 + +最大输出长度 + +2048 + +上下文长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.6 + +每百万tokens + +输入：音频 + +25 + +每百万tokens + +输入：图片/视频 + +6 + +每百万tokens + +输出：文本（输入仅包含文本时） + +6.4 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +18 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +50 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.982 + +每百万tokens + +输入：音频 + +32.586 + +每百万tokens + +输入：图片/视频 + +6.165 + +每百万tokens + +输出：文本（输入仅包含文本时） + +7.853 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +18.495 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +65.246 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +10,000 + +## 动态更新 + +### qwen-omni-turbo-realtime-latest + +千问全新多模态理解生成大模型实时版，此版本为动态更新版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30720 + +最大输出长度 + +2048 + +上下文长度 + +32768 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.6 + +每百万tokens + +输入：音频 + +25 + +每百万tokens + +输入：图片/视频 + +6 + +每百万tokens + +输出：文本（输入仅包含文本时） + +6.4 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +18 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +50 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.982 + +每百万tokens + +输入：音频 + +32.586 + +每百万tokens + +输入：图片/视频 + +6.165 + +每百万tokens + +输出：文本（输入仅包含文本时） + +7.853 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +18.495 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +65.246 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +10,000 + +## 快照版本 + +### qwen-omni-turbo-realtime-2025-05-08 + +千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。此版本为2025年5月8日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30720 + +最大输出长度 + +2048 + +上下文长度 + +32768 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.6 + +每百万tokens + +输入：音频 + +25 + +每百万tokens + +输入：图片/视频 + +6 + +每百万tokens + +输出：文本（输入仅包含文本时） + +6.4 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +18 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +50 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.982 + +每百万tokens + +输入：音频 + +32.586 + +每百万tokens + +输入：图片/视频 + +6.165 + +每百万tokens + +输出：文本（输入仅包含文本时） + +7.853 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +18.495 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +65.246 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +10,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen-omni-turbo.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen-omni-turbo.md new file mode 100644 index 00000000..e5af0ff6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen-omni-turbo.md @@ -0,0 +1,821 @@ +# qwen-omni-turbo + +千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。该模型版本功能等同于快照模型 qwen-omni-turbo-2025-03-26。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30720 + +最大输出长度 + +2048 + +上下文长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.4 + +每百万tokens + +输入：音频 + +25 + +每百万tokens + +输入：图片/视频 + +1.5 + +每百万tokens + +输出：文本（输入仅包含文本时） + +1.6 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +4.5 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +50 + +每百万tokens + +输入：图片/视频（缓存命中） + +0.3 + +每百万tokens + +输入：文本（缓存命中） + +0.08 + +每百万tokens + +输入：音频（缓存命中） + +5 + +每百万tokens + +输入：文本（Batch File） + +0.2 + +每百万tokens + +输入：音频（Batch File） + +12.5 + +每百万tokens + +输入：图片/视频（Batch File） + +0.75 + +每百万tokens + +输出：文本+音频（Batch File） + +25 + +每百万tokens + +输出：文本（Batch File，输入包含图片/音频/视频时） + +2.25 + +每百万tokens + +输出：文本（Batch File，输入仅包含文本时） + +0.8 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.514 + +每百万tokens + +输入：音频 + +32.586 + +每百万tokens + +输入：图片/视频 + +1.541 + +每百万tokens + +输出：文本（输入仅包含文本时） + +1.982 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +4.624 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +65.246 + +每百万tokens + +输入：图片/视频（缓存命中） + +0.294 + +每百万tokens + +输入：文本（缓存命中） + +0.11 + +每百万tokens + +输入：音频（缓存命中） + +6.532 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 动态更新 + +### qwen-omni-turbo-latest + +千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色，此版本为动态更新版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30720 + +最大输出长度 + +2048 + +上下文长度 + +32768 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.4 + +每百万tokens + +输入：音频 + +25 + +每百万tokens + +输入：图片/视频 + +1.5 + +每百万tokens + +输出：文本（输入仅包含文本时） + +1.6 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +4.5 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +50 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.514 + +每百万tokens + +输入：音频 + +32.586 + +每百万tokens + +输入：图片/视频 + +1.541 + +每百万tokens + +输出：文本（输入仅包含文本时） + +1.982 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +4.624 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +65.246 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen-omni-turbo-2025-03-26 + +千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色，此版本为2025年3月26日的快照版本，相比1月19日的快照版本在视觉能力上有大幅提升。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30720 + +最大输出长度 + +2048 + +上下文长度 + +32768 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.4 + +每百万tokens + +输入：音频 + +25 + +每百万tokens + +输入：图片/视频 + +1.5 + +每百万tokens + +输出：文本（输入仅包含文本时） + +1.6 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +4.5 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +50 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.514 + +每百万tokens + +输入：音频 + +32.586 + +每百万tokens + +输入：图片/视频 + +1.541 + +每百万tokens + +输出：文本（输入仅包含文本时） + +1.982 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +4.624 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +65.246 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +### qwen-omni-turbo-2025-01-19 + +千问全模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色，此版本为2025年1月19日的快照版本，预计维护至下一个快照发布前的一个月左右。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30720 + +最大输出长度 + +2048 + +上下文长度 + +32768 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.4 + +每百万tokens + +输入：音频 + +25 + +每百万tokens + +输入：图片/视频 + +1.5 + +每百万tokens + +输出：文本（输入仅包含文本时） + +1.6 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +4.5 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +50 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen2-5-omni-7b.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen2-5-omni-7b.md new file mode 100644 index 00000000..06bd72da --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen2-5-omni-7b.md @@ -0,0 +1,185 @@ +# qwen2.5-omni-7b + +基于Qwen2.5训练的全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +30720 + +最大输出长度 + +2048 + +上下文长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.6 + +每百万tokens + +输入：音频 + +38 + +每百万tokens + +输入：图片/视频 + +2 + +每百万tokens + +输出：文本（输入仅包含文本时） + +2.4 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +6 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +76 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +0.734 + +每百万tokens + +输入：音频 + +49.613 + +每百万tokens + +输入：图片/视频 + +2.055 + +每百万tokens + +输出：文本（输入仅包含文本时） + +2.936 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +6.165 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +99.153 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-flash-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-flash-realtime.md new file mode 100644 index 00000000..dc5fdd36 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-flash-realtime.md @@ -0,0 +1,357 @@ +# qwen3.5-omni-flash-realtime + +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。该模型版本功能等同于快照模型 qwen3.5-omni-flash-realtime-2026-03-15。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +27 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +107 + +每百万tokens + +输入：文本/图片/视频 + +3.3 + +每百万tokens + +输出：文本 + +20 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +33.72 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +132.65 + +每百万tokens + +输入：文本/图片/视频 + +4.12 + +每百万tokens + +输出：文本 + +24.73 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3.5-omni-flash-realtime-2026-03-15 + +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。该版本为2026年3月15日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +27 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +107 + +每百万tokens + +输入：文本/图片/视频 + +3.3 + +每百万tokens + +输出：文本 + +20 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +33.72 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +132.65 + +每百万tokens + +输入：文本/图片/视频 + +4.12 + +每百万tokens + +输出：文本 + +24.73 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-flash.md new file mode 100644 index 00000000..1c47fd1f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-flash.md @@ -0,0 +1,357 @@ +# qwen3.5-omni-flash + +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。该模型版本功能等同于快照模型 qwen3.5-omni-flash-2026-03-15。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +18 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +72 + +每百万tokens + +输入：文本/图片/视频 + +2.2 + +每百万tokens + +输出：文本 + +13.3 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +22.48 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +89.18 + +每百万tokens + +输入：文本/图片/视频 + +3 + +每百万tokens + +输出：文本 + +16.49 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3.5-omni-flash-2026-03-15 + +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。该版本为2026年3月15日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +18 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +72 + +每百万tokens + +输入：文本/图片/视频 + +2.2 + +每百万tokens + +输出：文本 + +13.3 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +22.48 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +89.18 + +每百万tokens + +输入：文本/图片/视频 + +3 + +每百万tokens + +输出：文本 + +16.49 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-plus-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-plus-realtime.md new file mode 100644 index 00000000..a56c64e9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-plus-realtime.md @@ -0,0 +1,357 @@ +# qwen3.5-omni-plus-realtime + +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。该模型版本功能等同于快照模型 qwen3.5-omni-plus-realtime-2026-03-15。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +80 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +300 + +每百万tokens + +输入：文本/图片/视频 + +10 + +每百万tokens + +输出：文本 + +60 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +123.65 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +464.64 + +每百万tokens + +输入：文本/图片/视频 + +15.74 + +每百万tokens + +输出：文本 + +92.93 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3.5-omni-plus-realtime-2026-03-15 + +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。该版本为2026年3月15日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +80 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +300 + +每百万tokens + +输入：文本/图片/视频 + +10 + +每百万tokens + +输出：文本 + +60 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +123.65 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +464.64 + +每百万tokens + +输入：文本/图片/视频 + +15.74 + +每百万tokens + +输出：文本 + +92.93 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-plus.md new file mode 100644 index 00000000..5d947b50 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-5-omni-plus.md @@ -0,0 +1,393 @@ +# qwen3.5-omni-plus + +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。该模型版本功能等同于快照模型 qwen3.5-omni-plus-2026-03-15。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +53 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +213 + +每百万tokens + +输入：文本/图片/视频 + +7 + +每百万tokens + +输出：文本 + +40 + +每百万tokens + +输入：音频（Batch File） + +26.5 + +每百万tokens + +输入：文本/图片/视频（Batch File） + +3.5 + +每百万tokens + +输出：文本（Batch File） + +20 + +每百万tokens + +输入：音频（Batch Chat） + +53 + +每百万tokens + +输入：文本/图片/视频（Batch Chat） + +7 + +每百万tokens + +输出：文本（Batch Chat） + +40 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +82.44 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +329.74 + +每百万tokens + +输入：文本/图片/视频 + +10.49 + +每百万tokens + +输出：文本 + +62.2 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3.5-omni-plus-2026-03-15 + +Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。该版本为2026年3月15日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +53 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +213 + +每百万tokens + +输入：文本/图片/视频 + +7 + +每百万tokens + +输出：文本 + +40 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +82.44 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +329.74 + +每百万tokens + +输入：文本/图片/视频 + +10.49 + +每百万tokens + +输出：文本 + +62.2 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-omni-flash-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-omni-flash-realtime.md new file mode 100644 index 00000000..64516c1d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-omni-flash-realtime.md @@ -0,0 +1,561 @@ +# qwen3-omni-flash-realtime + +千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。该模型版本功能等同于快照模型 qwen3-omni-flash-realtime-2025-12-01。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +16384 + +上下文长度 + +65536 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +2.2 + +每百万tokens + +输入：音频 + +18.9 + +每百万tokens + +输入：图片/视频 + +3.9 + +每百万tokens + +输出：文本（输入仅包含文本时） + +8.3 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +15.2 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +75.1 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +3.816 + +每百万tokens + +输入：音频 + +33.54 + +每百万tokens + +输入：图片/视频 + +6.899 + +每百万tokens + +输出：文本（输入仅包含文本时） + +14.605 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +26.935 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +133.06 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3-omni-flash-realtime-2025-12-01 + +千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，支持49种语音音色，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。此版本为2025年12月01日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Audio**　**Video** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +16384 + +上下文长度 + +65536 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +2.2 + +每百万tokens + +输入：音频 + +18.9 + +每百万tokens + +输入：图片/视频 + +3.9 + +每百万tokens + +输出：文本（输入仅包含文本时） + +8.3 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +15.2 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +75.1 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +### qwen3-omni-flash-realtime-2025-09-15 + +千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。此版本为2025年9月15日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +16384 + +上下文长度 + +65536 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +2.2 + +每百万tokens + +输入：音频 + +18.9 + +每百万tokens + +输入：图片/视频 + +3.9 + +每百万tokens + +输出：文本（输入仅包含文本时） + +8.3 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +15.2 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +75.1 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +3.816 + +每百万tokens + +输入：音频 + +33.54 + +每百万tokens + +输入：图片/视频 + +6.899 + +每百万tokens + +输出：文本（输入仅包含文本时） + +14.605 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +26.935 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +133.06 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-omni-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-omni-flash.md new file mode 100644 index 00000000..8ff3093a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-omni/qwen3-omni-flash.md @@ -0,0 +1,747 @@ +# qwen3-omni-flash + +千问3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。该模型版本功能等同于快照模型 qwen3-omni-flash-2025-12-01。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +16384 + +上下文长度 + +65536 + +最大输入长度（思考模式下） + +16384 + +最大输出长度（思考模式下） + +16384 + +最大思维链长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.8 + +每百万tokens + +输入：音频 + +15.8 + +每百万tokens + +输入：图片/视频 + +3.3 + +每百万tokens + +输出：文本（输入仅包含文本时） + +6.9 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +12.7 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +62.6 + +每百万tokens + +输入：文本（思考） + +1.8 + +每百万tokens + +输入：音频（思考） + +15.8 + +每百万tokens + +输入：图片/视频（思考） + +3.3 + +每百万tokens + +输出：文本（思考模式下，输入仅包含文本时） + +6.9 + +每百万tokens + +输出：文本（思考模式下，输入包含图片/音频/视频时） + +12.7 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +3.156 + +每百万tokens + +输入：音频 + +27.962 + +每百万tokens + +输入：图片/视频 + +5.725 + +每百万tokens + +输出：文本（输入仅包含文本时） + +12.183 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +22.458 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +110.896 + +每百万tokens + +输入：文本（思考） + +3.156 + +每百万tokens + +输入：音频（思考） + +27.962 + +每百万tokens + +输入：图片/视频（思考） + +5.725 + +每百万tokens + +输出：文本（思考模式下，输入仅包含文本时） + +12.183 + +每百万tokens + +输出：文本（思考模式下，输入包含图片/音频/视频时） + +22.458 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3-omni-flash-2025-12-01 + +千问3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，支持49种语音音色，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。此版本为2025年12月01日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Audio**　**Video** + +输出模态 + +**Text**　**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +16384 + +上下文长度 + +65536 + +最大输入长度（思考模式下） + +16384 + +最大输出长度（思考模式下） + +16384 + +最大思维链长度 + +32768 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.8 + +每百万tokens + +输入：音频 + +15.8 + +每百万tokens + +输入：图片/视频 + +3.3 + +每百万tokens + +输出：文本（输入仅包含文本时） + +6.9 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +12.7 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +62.6 + +每百万tokens + +输入：文本（思考） + +1.8 + +每百万tokens + +输入：音频（思考） + +15.8 + +每百万tokens + +输入：图片/视频（思考） + +3.3 + +每百万tokens + +输出：文本（思考模式下，输入仅包含文本时） + +6.9 + +每百万tokens + +输出：文本（思考模式下，输入包含图片/音频/视频时） + +12.7 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +### qwen3-omni-flash-2025-09-15 + +Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。此版本为2025年9月15日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +16384 + +上下文长度 + +65536 + +最大输入长度（思考模式下） + +16384 + +最大输出长度（思考模式下） + +16384 + +最大思维链长度 + +32768 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.8 + +每百万tokens + +输入：音频 + +15.8 + +每百万tokens + +输入：图片/视频 + +3.3 + +每百万tokens + +输出：文本（输入仅包含文本时） + +6.9 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +12.7 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +62.6 + +每百万tokens + +输入：文本（思考） + +1.8 + +每百万tokens + +输入：音频（思考） + +15.8 + +每百万tokens + +输入：图片/视频（思考） + +3.3 + +每百万tokens + +输出：文本（思考模式下，输入仅包含文本时） + +6.9 + +每百万tokens + +输出：文本（思考模式下，输入包含图片/音频/视频时） + +12.7 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：文本 + +3.156 + +每百万tokens + +输入：音频 + +27.962 + +每百万tokens + +输入：图片/视频 + +5.725 + +每百万tokens + +输出：文本（输入仅包含文本时） + +12.183 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +22.458 + +每百万tokens + +输出：文本+音频（输出的文本不计费） + +110.896 + +每百万tokens + +输入：文本（思考） + +3.156 + +每百万tokens + +输入：音频（思考） + +27.962 + +每百万tokens + +输入：图片/视频（思考） + +5.725 + +每百万tokens + +输出：文本（思考模式下，输入仅包含文本时） + +12.183 + +每百万tokens + +输出：文本（思考模式下，输入包含图片/音频/视频时） + +22.458 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-flash-8k-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-flash-8k-realtime.md new file mode 100644 index 00000000..e7965377 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-flash-8k-realtime.md @@ -0,0 +1,213 @@ +# fun-asr-flash-8k-realtime + +通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。该模型版本功能等同于快照模型 fun-asr-flash-8k-realtime-2026-01-28。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +## 快照版本 + +### fun-asr-flash-8k-realtime-2026-01-28 + +通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。此版本为2026年1月28日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-flash.md new file mode 100644 index 00000000..f90e43ae --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-flash.md @@ -0,0 +1,269 @@ +# fun-asr-flash-2026-06-15 + +百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 快照版本 + +### fun-asr-flash-2026-06-15 + +百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +600 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-mtl.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-mtl.md new file mode 100644 index 00000000..b91e8935 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-mtl.md @@ -0,0 +1,237 @@ +# fun-asr-mtl + +百聆多语言语音识别大模型，支持超过31种语言，支持语种自由切换，出海用户首推，尤其东南亚出海。fun-asr为该模型的升级版本，建议切换使用fun-asr。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 + +## 快照版本 + +### fun-asr-mtl-2025-08-25 + +百聆多语言语音识别大模型，支持超过31种语言，支持语种自由切换，出海用户首推，尤其东南亚出海。fun-asr为该模型的升级版本，建议切换使用fun-asr。此版本为2025年8月25日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-realtime.md new file mode 100644 index 00000000..b7eece61 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr-realtime.md @@ -0,0 +1,449 @@ +# fun-asr-realtime + +通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00033 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +## 快照版本 + +### fun-asr-realtime-2026-02-28 + +通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。此版本为2026年2月28日的快照版本，在近场VAD方面进行了优化。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00033 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +### fun-asr-realtime-2025-11-07 + +通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。此版本为2025年11月7日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00033 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +### fun-asr-realtime-2025-09-15 + +通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。此版本为2025年9月15日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00033 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr.md new file mode 100644 index 00000000..3d2164ac --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/fun-asr.md @@ -0,0 +1,403 @@ +# fun-asr + +百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本等同于2025年11月7日的快照版本。该模型版本功能等同于快照模型 fun-asr-2025-11-07。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 快照版本 + +### fun-asr-2025-11-07 + +百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本为2025年11月7日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +600 + +### fun-asr-2025-08-25 + +百聆新一代语音识别大模型，主打中文、英文、日文语音识别，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境，国内用户首推。此版本为2025年8月25日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +600 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/gummy-chat-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/gummy-chat-v1.md new file mode 100644 index 00000000..57a29e8b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/gummy-chat-v1.md @@ -0,0 +1,105 @@ +# gummy-chat-v1 + +多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00015 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/gummy-realtime-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/gummy-realtime-v1.md new file mode 100644 index 00000000..45bcde9d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/gummy-realtime-v1.md @@ -0,0 +1,105 @@ +# gummy-realtime-v1 + +多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00015 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/model-qwen3-livetranslate-flash-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/model-qwen3-livetranslate-flash-realtime.md new file mode 100644 index 00000000..5b36abc9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/model-qwen3-livetranslate-flash-realtime.md @@ -0,0 +1,357 @@ +# qwen3-livetranslate-flash-realtime + +Qwen3-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。该模型版本功能等同于快照模型 qwen3-livetranslate-flash-realtime-2025-09-22。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +4096 + +上下文长度 + +53248 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +64 + +每百万tokens + +输入：图片 + +8 + +每百万tokens + +输出：文本 + +64 + +每百万tokens + +输出：音频 + +240 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +73.392 + +每百万tokens + +输入：图片 + +9.541 + +每百万tokens + +输出：文本 + +73.392 + +每百万tokens + +输出：音频 + +278.891 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3-livetranslate-flash-realtime-2025-09-22 + +千问3-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，千问3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。此版本为2025年9月22日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Audio** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +4096 + +上下文长度 + +53248 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +64 + +每百万tokens + +输入：图片 + +8 + +每百万tokens + +输出：文本 + +64 + +每百万tokens + +输出：音频 + +240 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +73.392 + +每百万tokens + +输入：图片 + +9.541 + +每百万tokens + +输出：文本 + +73.392 + +每百万tokens + +输出：音频 + +278.891 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/model-qwen3-livetranslate-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/model-qwen3-livetranslate-flash.md new file mode 100644 index 00000000..fe34a355 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/model-qwen3-livetranslate-flash.md @@ -0,0 +1,289 @@ +# qwen3-livetranslate-flash + +Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio**　**Video** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +4096 + +上下文长度 + +53248 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +10 + +每百万tokens + +输入：图片 + +4 + +每百万tokens + +输出：文本 + +10 + +每百万tokens + +输出：音频 + +40 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +100 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3-livetranslate-flash-2025-12-01 + +Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。此版本为2025年12月01日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio**　**Video** + +输出模态 + +**Text**　**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +4096 + +上下文长度 + +53248 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +10 + +每百万tokens + +输入：图片 + +4 + +每百万tokens + +输出：文本 + +10 + +每百万tokens + +输出：音频 + +40 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +100 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-8k-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-8k-v1.md new file mode 100644 index 00000000..e74e4ad4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-8k-v1.md @@ -0,0 +1,93 @@ +# paraformer-8k-v1 + +Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00008 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-8k-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-8k-v2.md new file mode 100644 index 00000000..e00dfd56 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-8k-v2.md @@ -0,0 +1,93 @@ +# paraformer-8k-v2 + +Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00008 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-mtl-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-mtl-v1.md new file mode 100644 index 00000000..3f51d7ef --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-mtl-v1.md @@ -0,0 +1,93 @@ +# paraformer-mtl-v1 + +Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。 支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话、宁夏话、山西话、陕西话、山东话、四川话、天津话）、英语、日语、韩语、西班牙语、印尼语、法语、德语、意大利语、马来语。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00008 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-8k-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-8k-v1.md new file mode 100644 index 00000000..4122c259 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-8k-v1.md @@ -0,0 +1,93 @@ +# paraformer-realtime-8k-v1 + +Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00024 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-8k-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-8k-v2.md new file mode 100644 index 00000000..2e0ee24b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-8k-v2.md @@ -0,0 +1,93 @@ +# paraformer-realtime-8k-v2 + +推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language\_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服等场景下的实时语音识别。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00024 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-v1.md new file mode 100644 index 00000000..7af6d727 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-v1.md @@ -0,0 +1,93 @@ +# paraformer-realtime-v1 + +Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00024 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-v2.md new file mode 100644 index 00000000..28610811 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-realtime-v2.md @@ -0,0 +1,93 @@ +# paraformer-realtime-v2 + +推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language\_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 可支持热词。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00024 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-v1.md new file mode 100644 index 00000000..299effd8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-v1.md @@ -0,0 +1,93 @@ +# paraformer-v1 + +Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00008 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-v2.md new file mode 100644 index 00000000..92200cd3 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/paraformer-v2.md @@ -0,0 +1,93 @@ +# paraformer-v2 + +推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language\_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。可支持热词。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00008 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-6.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-6.md new file mode 100644 index 00000000..73db5c7b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-6.md @@ -0,0 +1,357 @@ +# qwen3.5-livetranslate-flash-realtime + +Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio**　**Image** + +输出模态 + +**Audio**　**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +4096 + +上下文长度 + +53248 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +40 + +每百万tokens + +输入：图片 + +3.3 + +每百万tokens + +输出：文本 + +100 + +每百万tokens + +输出：音频 + +160 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +56.207 + +每百万tokens + +输入：图片 + +4.122 + +每百万tokens + +输出：文本 + +149.884 + +每百万tokens + +输出：音频 + +224.826 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen3.5-livetranslate-flash-realtime-2026-05-19 + +Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio**　**Image** + +输出模态 + +**Audio**　**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +49152 + +最大输出长度 + +4096 + +上下文长度 + +53248 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +40 + +每百万tokens + +输入：图片 + +3.3 + +每百万tokens + +输出：文本 + +100 + +每百万tokens + +输出：音频 + +160 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +56.207 + +每百万tokens + +输入：图片 + +4.122 + +每百万tokens + +输出：文本 + +149.884 + +每百万tokens + +输出：音频 + +224.826 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-filetrans.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-filetrans.md new file mode 100644 index 00000000..f9bd9cfc --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-filetrans.md @@ -0,0 +1,269 @@ +# qwen3-asr-flash-filetrans + +千问3-ASR-Flash的大文件转录版本，千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +100 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 + +## 快照版本 + +### qwen3-asr-flash-filetrans-2025-11-17 + +千问3-ASR-Flash的大文件转录版本，千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2025年11月17日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +100 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-realtime.md new file mode 100644 index 00000000..283cac5e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-realtime.md @@ -0,0 +1,403 @@ +# qwen3-asr-flash-realtime + +千问3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。该模型版本功能等同于快照模型 qwen3-asr-flash-realtime-2025-10-27。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00033 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00066 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +## 快照版本 + +### qwen3-asr-flash-realtime-2026-02-10 + +Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2026年2月10日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00033 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00066 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +### qwen3-asr-flash-realtime-2025-10-27 + +千问3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2025年10月27日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00033 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00066 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +1200 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +1200 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-us.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-us.md new file mode 100644 index 00000000..efa07e23 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash-us.md @@ -0,0 +1,193 @@ +# qwen3-asr-flash-us + +Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 11 个语种的语音，在复杂的音频环境下能够保证精确转录。该模型版本功能等同于快照模型 qwen3-asr-flash-2025-09-08-us。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 美国（弗吉尼亚） + +部署范围：美国 + +计费项 + +价格（元） + +单位 + +语音识别 + +0.00026 + +每秒 + +## 快照版本 + +### qwen3-asr-flash-2025-09-08-us + +Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 11 个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2025年9月8日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 美国（弗吉尼亚） + +部署范围：美国 + +计费项 + +价格（元） + +单位 + +语音识别 + +0.00026 + +每秒 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash.md new file mode 100644 index 00000000..003cea20 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-asr-flash.md @@ -0,0 +1,403 @@ +# qwen3-asr-flash + +千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。该模型版本功能等同于快照模型 qwen3-asr-flash-2025-09-08。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +100 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 + +## 快照版本 + +### qwen3-asr-flash-2026-02-10 + +通义千问3-ASR-Flash，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别 多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2026年2月10日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +100 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 + +### qwen3-asr-flash-2025-09-08 + +千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。此版本为2025年9月8日的快照版本。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00022 + +每秒 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +音频时长 + +0.00026 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +100 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +100 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-omni-30b-a3b-captioner.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-omni-30b-a3b-captioner.md new file mode 100644 index 00000000..e84168ac --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/qwen3-omni-30b-a3b-captioner.md @@ -0,0 +1,153 @@ +# qwen3-omni-30b-a3b-captioner + +千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +32768 + +最大输出长度 + +32768 + +上下文长度 + +65536 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：音频 + +15.8 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +12.7 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入：音频 + +27.962 + +每百万tokens + +输出：文本（输入包含图片/音频/视频时） + +22.458 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/speech-biasing.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/speech-biasing.md new file mode 100644 index 00000000..86e2161a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-recognition/speech-biasing.md @@ -0,0 +1,91 @@ +# speech-biasing + +热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-clone-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-clone-v1.md new file mode 100644 index 00000000..a50ef4ed --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-clone-v1.md @@ -0,0 +1,93 @@ +# cosyvoice-clone-v1 + +声音复刻Cosyvoice大模型，依托先进的大模型技术进行特征提取，从而完成声音的复刻，且无需训练过程。仅需提供时长较短的音频，即可迅速生成高度相似且听感自然的定制声音。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +2 + +每万字符 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v1.md new file mode 100644 index 00000000..2b35a405 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v1.md @@ -0,0 +1,93 @@ +# cosyvoice-v1 + +CosyVoice 是通义实验室依托大规模预训练语言模型，深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +2 + +每万字符 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v2.md new file mode 100644 index 00000000..4682b2f9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v2.md @@ -0,0 +1,93 @@ +# cosyvoice-v2 + +cosyvoice-V2是通义实验室依托大规模预训练语言模型，在深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +2 + +每万字符 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-5-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-5-flash.md new file mode 100644 index 00000000..e8d68e1f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-5-flash.md @@ -0,0 +1,105 @@ +# cosyvoice-v3.5-flash + +CosyVoice-v3.5-Flash是通义实验室CosyVoice系列的高性能语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-5-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-5-plus.md new file mode 100644 index 00000000..3cc13936 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-5-plus.md @@ -0,0 +1,105 @@ +# cosyvoice-v3.5-plus + +CosyVoice-v3.5-Plus是通义实验室CosyVoice系列的超高表现力语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1.5 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-flash.md new file mode 100644 index 00000000..730a99d0 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-flash.md @@ -0,0 +1,133 @@ +# cosyvoice-v3-flash + +合成能力：CosyVoice-v3-Flash是通义实验室CosyVoice系列最新版高性能的语音合成大模型，较之前版本在自然度、音质、韵律、情感表现力上有更好的表现。该模型支持文本至语音的实时流式合成。克隆能力：CosyVoice-v3-Flash也是通义实验室CosyVoice系列最新版的语音克隆大模型，较之前版本提升了发音准确性、音色相似度，并且增加了更多小语种支持（德、西、法、意、俄）。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.9541 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-plus.md new file mode 100644 index 00000000..d5c4473a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/cosyvoice-v3-plus.md @@ -0,0 +1,133 @@ +# cosyvoice-v3-plus + +克隆能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音克隆大模型，具有更好的音质和复刻相似度，适用于更专业的场景。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。合成能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音合成大模型，具有更好的音质和表现力，适用于更专业的场景。该模型支持文本至语音的实时流式合成。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +2 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +1.9082 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/fun-music-v1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/fun-music-v1.md new file mode 100644 index 00000000..a2089a77 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/fun-music-v1.md @@ -0,0 +1,105 @@ +# fun-music-v1 + +百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.002 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/model-qwen-tts-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/model-qwen-tts-realtime.md new file mode 100644 index 00000000..a5a1c8fe --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/model-qwen-tts-realtime.md @@ -0,0 +1,351 @@ +# qwen-tts-realtime + +Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成模型。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +512 + +最大输出长度 + +7680 + +上下文长度 + +8192 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2.4 + +每百万tokens + +输出 + +12 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 动态更新 + +### qwen-tts-realtime-latest + +Qwen-TTS实时模型是通义实验室千问模型中语音合成利器，始终与最新快照版能力相同。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。本模型是动态更新版本，模型更新不会提前通知。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +512 + +最大输出长度 + +7680 + +上下文长度 + +8192 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2.4 + +每百万tokens + +输出 + +12 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen-tts-realtime-2025-07-15 + +Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成利器。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。此版本为2025年7月15日快照模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +512 + +最大输出长度 + +7680 + +上下文长度 + +8192 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2.4 + +每百万tokens + +输出 + +12 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/model-qwen-tts.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/model-qwen-tts.md new file mode 100644 index 00000000..ff43d0bd --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/model-qwen-tts.md @@ -0,0 +1,467 @@ +# qwen-tts + +千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持流式输出。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +512 + +最大输出长度 + +7680 + +上下文长度 + +8192 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.6 + +每百万tokens + +输出：音频 + +10 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 动态更新 + +### qwen-tts-latest + +模型是动态更新版本，等同于最新版本快照模型，模型更新时不会提前通知。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +512 + +最大输出长度 + +7680 + +上下文长度 + +8192 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.6 + +每百万tokens + +输出：音频 + +10 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +## 快照版本 + +### qwen-tts-2025-05-22 + +千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持流式输出。本版本是2025年5月22日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +512 + +最大输出长度 + +7680 + +上下文长度 + +8192 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.6 + +每百万tokens + +输出：音频 + +10 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 + +### qwen-tts-2025-04-10 + +千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持流式输出。本版本是2025年4月10日快照。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +512 + +最大输出长度 + +7680 + +上下文长度 + +8192 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入：文本 + +1.6 + +每百万tokens + +输出：音频 + +10 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/music-generation-preview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/music-generation-preview.md new file mode 100644 index 00000000..d9aa9421 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/music-generation-preview.md @@ -0,0 +1,213 @@ +# fun-music-preview + +百聆音乐生成preview版大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。本次版本为预览快照版 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.005 + +每秒 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### fun-music-preview + +百聆音乐生成preview版大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。本次版本为预览快照版 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +音频时长 + +0.005 + +每秒 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen-voice-design.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen-voice-design.md new file mode 100644 index 00000000..1be256d4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen-voice-design.md @@ -0,0 +1,133 @@ +# qwen-voice-design + +千问voice-design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +声音复刻及声音设计 + +0.2 + +每次 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +声音复刻及声音设计 + +0.2 + +每次 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen-voice-enrollment.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen-voice-enrollment.md new file mode 100644 index 00000000..b66cbea6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen-voice-enrollment.md @@ -0,0 +1,133 @@ +# qwen-voice-enrollment + +千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +声音复刻及声音设计 + +0.01 + +次 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.01 + +每次 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-flash-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-flash-realtime.md new file mode 100644 index 00000000..d6d40d64 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-flash-realtime.md @@ -0,0 +1,403 @@ +# qwen3-tts-flash-realtime + +千问3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型版本功能等同于快照模型 qwen3-tts-flash-realtime-2025-11-27。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### qwen3-tts-flash-realtime-2025-11-27 + +千问3-TTS-Flash-Realtime模型是通义最新的实时语音合成大模型，不仅拥有51种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年11月27日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +### qwen3-tts-flash-realtime-2025-09-18 + +Qwen3-TTS-Flash-Realtime-2025-09-18模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年9月18日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-flash.md new file mode 100644 index 00000000..73c4bc3c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-flash.md @@ -0,0 +1,403 @@ +# qwen3-tts-flash + +Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型版本功能等同于快照模型 qwen3-tts-flash-2025-11-27。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.733924 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### qwen3-tts-flash-2025-11-27 + +千问3-TTS-Flash模型是通义最新推出的离线语音合成大模型，不仅拥有51种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年11月27日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.733924 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +### qwen3-tts-flash-2025-09-18 + +千问3-TTS-Flash-2025-09-18模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年9月18日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.733924 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +10 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-instruct-flash-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-instruct-flash-realtime.md new file mode 100644 index 00000000..e89c98a9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-instruct-flash-realtime.md @@ -0,0 +1,269 @@ +# qwen3-tts-instruct-flash-realtime + +千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。该模型版本功能等同于快照模型 qwen3-tts-instruct-flash-realtime-2026-01-22。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### qwen3-tts-instruct-flash-realtime-2026-01-22 + +千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型为2026年01月22日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-instruct-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-instruct-flash.md new file mode 100644 index 00000000..60ac8112 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-instruct-flash.md @@ -0,0 +1,269 @@ +# qwen3-tts-instruct-flash + +Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型版本功能等同于快照模型 qwen3-tts-instruct-flash-2026-01-26。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### qwen3-tts-instruct-flash-2026-01-26 + +Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型为2026年01月26日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vc-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vc-realtime.md new file mode 100644 index 00000000..71f9fbbc --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vc-realtime.md @@ -0,0 +1,403 @@ +# qwen3-tts-vc-realtime-2026-01-15 + +千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### qwen3-tts-vc-realtime-2026-01-15 + +千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +### qwen3-tts-vc-realtime-2025-11-27 + +千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年11月27日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vc.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vc.md new file mode 100644 index 00000000..a6261d79 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vc.md @@ -0,0 +1,269 @@ +# qwen3-tts-vc-2026-01-22 + +Qwen3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月22日快照版本模型。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### qwen3-tts-vc-2026-01-22 + +Qwen3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月22日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vd-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vd-realtime.md new file mode 100644 index 00000000..ab514758 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vd-realtime.md @@ -0,0 +1,403 @@ +# qwen3-tts-vd-realtime-2026-01-15 + +千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### qwen3-tts-vd-realtime-2026-01-15 + +千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +### qwen3-tts-vd-realtime-2025-12-16 + +千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年12月16日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +1 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.954101 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vd.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vd.md new file mode 100644 index 00000000..87d3a0da --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/qwen3-tts-vd.md @@ -0,0 +1,269 @@ +# qwen3-tts-vd-2026-01-26 + +Qwen3-TTS-VD模型是通义最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月26日快照版本模型。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 快照版本 + +### qwen3-tts-vd-2026-01-26 + +Qwen3-TTS-VD模型是通义最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月26日快照版本模型。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +语音合成 + +0.8 + +每万字符 + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +180 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +180 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/sambert.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/sambert.md new file mode 100644 index 00000000..6431df81 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/sambert.md @@ -0,0 +1,267 @@ +# Sambert语音合成 + +提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +Sambert 语音合成 + +1 + +每万字符 + +## 音色列表 + +模型ID + +音色名称 + +sambert-betty-v1 + +Sambert语音合成-Betty + +sambert-brian-v1 + +Sambert语音合成-Brian + +sambert-cally-v1 + +Sambert语音合成-Cally + +sambert-camila-v1 + +Sambert语音合成-Camila + +sambert-cindy-v1 + +Sambert语音合成-Cindy + +sambert-clara-v1 + +Sambert语音合成-Clara + +sambert-donna-v1 + +Sambert语音合成-Donna + +sambert-eva-v1 + +Sambert语音合成-Eva + +sambert-hanna-v1 + +Sambert语音合成-Hanna + +sambert-indah-v1 + +Sambert语音合成-Indah + +sambert-perla-v1 + +Sambert语音合成-Perla + +sambert-waan-v1 + +Sambert语音合成-Waan + +sambert-zhichu-v1 + +Sambert语音合成-知厨 + +sambert-zhida-v1 + +Sambert语音合成-知达 + +sambert-zhide-v1 + +Sambert语音合成-知德 + +sambert-zhifei-v1 + +Sambert语音合成-知飞 + +sambert-zhigui-v1 + +Sambert语音合成-知柜 + +sambert-zhihao-v1 + +Sambert语音合成-知浩 + +sambert-zhijia-v1 + +Sambert语音合成-知佳 + +sambert-zhijing-v1 + +Sambert语音合成-知婧 + +sambert-zhilun-v1 + +Sambert语音合成-知伦 + +sambert-zhimao-v1 + +Sambert语音合成-知猫 + +sambert-zhimiao-emo-v1 + +Sambert语音合成-知妙（多情感） + +sambert-zhiming-v1 + +Sambert语音合成-知茗 + +sambert-zhimo-v1 + +Sambert语音合成-知墨 + +sambert-zhina-v1 + +Sambert语音合成-知娜 + +sambert-zhinan-v1 + +Sambert语音合成-知楠 + +sambert-zhiqi-v1 + +Sambert语音合成-知琪 + +sambert-zhiqian-v1 + +Sambert语音合成-知倩 + +sambert-zhiru-v1 + +Sambert语音合成-知茹 + +sambert-zhishu-v1 + +Sambert语音合成-知树 + +sambert-zhishuo-v1 + +Sambert语音合成-知硕 + +sambert-zhistella-v1 + +Sambert语音合成-知莎 + +sambert-zhiting-v1 + +Sambert语音合成-知婷 + +sambert-zhiwei-v1 + +Sambert语音合成-知薇 + +sambert-zhixiang-v1 + +Sambert语音合成-知祥 + +sambert-zhixiao-v1 + +Sambert语音合成-知笑 + +sambert-zhiya-v1 + +Sambert语音合成-知雅 + +sambert-zhiye-v1 + +Sambert语音合成-知晔 + +sambert-zhiying-v1 + +Sambert语音合成-知颖 + +sambert-zhiyuan-v1 + +Sambert语音合成-知媛 + +sambert-zhiyue-v1 + +Sambert语音合成-知悦 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-02-hd.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-02-hd.md new file mode 100644 index 00000000..78dfdd06 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-02-hd.md @@ -0,0 +1,115 @@ +# MiniMax/speech-02-hd + +MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +10000 + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +声音复刻及声音设计 + +9.9 + +每次 + +语音合成 + +3.5 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +20 + +TPM（每分钟tokens） + +20,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-02-turbo.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-02-turbo.md new file mode 100644 index 00000000..495a8092 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-02-turbo.md @@ -0,0 +1,115 @@ +# MiniMax/speech-02-turbo + +MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +10000 + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +声音复刻及声音设计 + +9.9 + +每次 + +语音合成 + +2 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +20 + +TPM（每分钟tokens） + +20,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-2-8-hd.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-2-8-hd.md new file mode 100644 index 00000000..dab06c52 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-2-8-hd.md @@ -0,0 +1,115 @@ +# MiniMax/speech-2.8-hd + +MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +10000 + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +声音复刻及声音设计 + +9.9 + +每次 + +语音合成 + +3.5 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +20 + +TPM（每分钟tokens） + +20,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-2-8-turbo.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-2-8-turbo.md new file mode 100644 index 00000000..36d321db --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/speech-2-8-turbo.md @@ -0,0 +1,115 @@ +# MiniMax/speech-2.8-turbo + +MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +10000 + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +声音复刻及声音设计 + +9.9 + +每次 + +语音合成 + +2 + +每万字符 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +20 + +TPM（每分钟tokens） + +20,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/voice-enrollment.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/voice-enrollment.md new file mode 100644 index 00000000..2b2ce542 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-speech-synthesis/voice-enrollment.md @@ -0,0 +1,103 @@ +# voice-enrollment + +大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。 大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Audio** + +输出模态 + +**Audio** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +— + +最大输出长度 + +— + +上下文长度 + +— + +## 模型价格 + +暂无公开定价信息。 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +600 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +600 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-ocr.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-ocr.md new file mode 100644 index 00000000..a23ee63f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-ocr.md @@ -0,0 +1,115 @@ +# vanchin/deepseek-ocr + +DeepSeek-OCR以 “探索视觉 - 文本压缩边界” 为核心目标，从大语言模型（LLM）视角重新定义视觉编码器的功能定位，为文档识别、图像转文本等高频场景提供了兼顾精度与效率的全新解决方案。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +8192 + +最大输出长度 + +8192 + +上下文长度 + +8192 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +0.216 + +每百万tokens + +输出 + +0.216 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-1-5b.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-1-5b.md new file mode 100644 index 00000000..a1e8129c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-1-5b.md @@ -0,0 +1,95 @@ +# deepseek-r1-distill-qwen-1.5b + +DeepSeek-R1-Distill-Qwen-1.5B是一个基于Qwen2.5-Math-1.5B的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +32768 + +最大输出长度 + +16384 + +上下文长度 + +32768 + +## 模型价格 + +暂无公开定价信息。 + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-14b.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-14b.md new file mode 100644 index 00000000..2bc285dc --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-14b.md @@ -0,0 +1,115 @@ +# deepseek-r1-distill-qwen-14b + +DeepSeek-R1-Distill-Qwen-14B是一个基于Qwen2.5-14B的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +32768 + +最大输出长度 + +16384 + +上下文长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +1 + +每百万tokens + +输出 + +3 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-32b.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-32b.md new file mode 100644 index 00000000..1f337274 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-32b.md @@ -0,0 +1,115 @@ +# deepseek-r1-distill-qwen-32b + +DeepSeek-R1-Distill-Qwen-32B是一个基于Qwen2.5-32B的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +32768 + +最大输出长度 + +16384 + +上下文长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2 + +每百万tokens + +输出 + +6 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-7b.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-7b.md new file mode 100644 index 00000000..82c2f6c5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1-distill-qwen-7b.md @@ -0,0 +1,115 @@ +# deepseek-r1-distill-qwen-7b + +DeepSeek-R1-Distill-Qwen-7B是一个基于Qwen2.5-Math-7B的蒸馏大型语言模型，使用了 DeepSeek R1 的输出。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +32768 + +最大输出长度 + +16384 + +上下文长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +0.5 + +每百万tokens + +输出 + +1 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1.md new file mode 100644 index 00000000..63cf7b03 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-r1.md @@ -0,0 +1,509 @@ +# deepseek-r1 + +DeepSeek-R1 在后训练阶段大规模使用了强化学习技术，在仅有极少标注数据的情况下，极大提升了模型推理能力。在数学、代码、自然语言推理等任务上，性能较高，能力较强。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +16384 + +上下文长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +输入（Batch File） + +2 + +每百万tokens + +输出（Batch File） + +8 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## siliconflow/deepseek-r1-0528 + +DeepSeek-R1-0528 是一款强化学习（RL）驱动的推理模型，解决了模型中的重复性和可读性问题。在 RL 之前，DeepSeek-R1 引入了冷启动数据，进一步优化了推理性能。它在数学、代码和推理任务中与 OpenAI-o1 表现相当，并且通过精心设计的训练方法，提升了整体效果。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +163840 + +最大输出长度 + +32768 + +上下文长度 + +163840 + +最大输入长度（思考模式下） + +163840 + +最大输出长度（思考模式下） + +32768 + +最大思维链长度 + +65536 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +500,000 + +## vanchin/deepseek-r1 + +DeepSeek-R1 是深度求索于 2025 年 1 月开源的 6710 亿参数混合专家（MoE）推理模型，推理时仅激活 370 亿参数。作为首个通过纯强化学习（无监督微调）训练的千亿级模型，实现了链式思维（CoT）的自然涌现。模型在 RL 前加入冷启动数据解决了 R1-Zero 的重复和混语问题，在数学、代码、推理任务上达到 OpenAI o1 水平。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +32768 + +上下文长度 + +131072 + +最大输入长度（思考模式下） + +98304 + +最大输出长度（思考模式下） + +32768 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +输入（缓存命中） + +1.6 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 快照版本 + +### deepseek-r1-0528 + +0528为R1模型的小版本升级，相较于旧版 R1，新版在复杂推理任务中的表现有了显著提升。在数学、编程与通用逻辑等多个基准测评中取得了优异成绩。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +16384 + +上下文长度 + +131072 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +100,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-1-terminus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-1-terminus.md new file mode 100644 index 00000000..ba3b48fe --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-1-terminus.md @@ -0,0 +1,257 @@ +# vanchin/deepseek-v3.1-terminus + +DeepSeek-V3.1-Terminus 是由深度求索（DeepSeek）发布的 V3.1 模型的更新版本，定位为混合智能体大语言模型。此次更新在保持模型原有能力的基础上，专注于修复用户反馈的问题并提升稳定性。它显著改善了语言一致性，减少了中英文混用和异常字符的出现。模型集成了“思考模式”（Thinking Mode）和“非思考模式”（Non-thinking Mode），用户可通过聊天模板灵活切换以适应不同任务。作为一个重要的优化，V3.1-Terminus 增强了代码智能体（Code Agent）和搜索智能体（Search Agent）的性能，使其在工具调用和执行多步复杂任务方面更加可靠 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +163840 + +最大输出长度 + +65536 + +上下文长度 + +163840 + +最大输入长度（思考模式下） + +163840 + +最大输出长度（思考模式下） + +65536 + +最大思维链长度 + +65536 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +12 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +500,000 + +## vanchin/deepseek-v3.1-terminus + +DeepSeek-V3.1-Terminus 是 DeepSeek-V3.1 的更新版本，旨在保持模型原有核心能力的同时，针对用户反馈的问题进行了修复和优化。该版本的模型结构与 DeepSeek-V3 保持一致，并在特定领域进行了显著增强。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +32768 + +最大输出长度 + +65536 + +上下文长度 + +131072 + +最大输入长度（思考模式下） + +32768 + +最大输出长度（思考模式下） + +65536 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +12 + +每百万tokens + +输入（缓存命中） + +1.6 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-1.md new file mode 100644 index 00000000..7b62c749 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-1.md @@ -0,0 +1,121 @@ +# deepseek-v3.1 + +DeepSeek-V3.1为混合推理架构模型，同时支持思考模式与非思考模式，具备更高的推理效率和更强的Agent能力。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +65536 + +上下文长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +12 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-2-exp.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-2-exp.md new file mode 100644 index 00000000..1dcad555 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-2-exp.md @@ -0,0 +1,115 @@ +# deepseek-v3.2-exp + +引入了DeepSeek Sparse Attention（一种稀疏注意力机制）的实验性质版本，针对长文本的训练和推理效率进行了探索性的优化和验证。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +65536 + +上下文长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2 + +每百万tokens + +输出 + +3 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-2-think.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-2-think.md new file mode 100644 index 00000000..61b10aa8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-2-think.md @@ -0,0 +1,129 @@ +# vanchin/deepseek-v3.2-think + +DeepSeek-V3.2 是一款实现了高计算效率与卓越推理及代理（Agent）性能完美协调的模型。该模型建立在 DeepSeek-V3 的基础之上，通过引入 DeepSeek 稀疏注意力（DSA）、可扩展的强化学习框架以及大规模代理任务合成流水线等关键技术突破，推动了开源大语言模型的前沿发展。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +65536 + +上下文长度 + +131072 + +最大输入长度（思考模式下） + +98304 + +最大输出长度（思考模式下） + +65536 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2 + +每百万tokens + +输出 + +3 + +每百万tokens + +输入（缓存命中） + +0.2 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +30 + +TPM（每分钟tokens） + +600,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-3.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-3.md new file mode 100644 index 00000000..20ddf8c9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3-3.md @@ -0,0 +1,341 @@ +# deepseek-v3.2 + +DeepSeek-V3.2是引入DeepSeek Sparse Attention（一种稀疏注意力机制）的正式版模型，也是DeepSeek推出的首个将思考融入工具使用的模型，同时支持思考模式与非思考模式的工具调用。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +65536 + +上下文长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2 + +每百万tokens + +输出 + +3 + +每百万tokens + +输入（缓存命中） + +0.4 + +每百万tokens + +输入（Batch File） + +1 + +每百万tokens + +输出（Batch File） + +1.5 + +每百万tokens + +显式缓存创建 + +2.5 + +每百万tokens + +显式缓存命中 + +0.2 + +每百万tokens + +输入（Batch Chat) + +2 + +每百万tokens + +输出（Batch Chat) + +3 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入 + +4.272 + +每百万tokens + +输出 + +12.815 + +每百万tokens + +输入（缓存命中） + +0.854 + +每百万tokens + +显式缓存创建 + +5.343 + +每百万tokens + +显式缓存命中 + +0.427 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +200,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10000 + +TPM（每分钟tokens） + +1,200,000 + +## siliconflow/deepseek-v3.2 + +DeepSeek-V3.2 是一款兼具高计算效率与卓越推理和 Agent 性能的模型。其方法建立在三大关键技术突破之上：DeepSeek 稀疏注意力（DSA），一种高效的注意力机制，在保持模型性能的同时显著降低了计算复杂性，并特别针对长上下文场景进行了优化；可扩展的强化学习框架，通过该框架，模型性能可与 GPT-5 相媲美，其高算力版本在推理能力上可与 Gemini-3.0-Pro 匹敌；以及大规模 Agent 任务合成管线，旨在将推理能力整合到工具使用场景中，从而提高在复杂交互环境中的指令遵循和泛化能力。该模型在 2025 年国际数学奥林匹克（IMO）和国际信息学奥林匹克（IOI）中取得了金牌表现 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +163840 + +最大输出长度 + +65536 + +上下文长度 + +163840 + +最大输入长度（思考模式下） + +163840 + +最大输出长度（思考模式下） + +65536 + +最大思维链长度 + +65536 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2 + +每百万tokens + +输出 + +3 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +500,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3.md new file mode 100644 index 00000000..0a7f35ca --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v3.md @@ -0,0 +1,371 @@ +# deepseek-v3 + +DeepSeek-V3 为自研 MoE 模型，671B 参数，激活 37B，在 14.8T token 上进行了预训练，在长文本、代码、数学、百科、中文 能力上表现优秀。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +57344 + +最大输出长度 + +8192 + +上下文长度 + +65536 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2 + +每百万tokens + +输出 + +8 + +每百万tokens + +输入（缓存命中） + +0.4 + +每百万tokens + +输入（Batch File） + +1 + +每百万tokens + +输出（Batch File） + +4 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## siliconflow/deepseek-v3-0324 + +新版 DeepSeek-V3 （DeepSeek-V3-0324）与之前的 DeepSeek-V3-1226 使用同样的 base 模型，仅改进了后训练方法。新版 V3 模型借鉴 DeepSeek-R1 模型训练过程中所使用的强化学习技术，大幅提高了在推理类任务上的表现水平，在数学、代码类相关评测集上取得了超过 GPT-4.5 的得分成绩。此外该模型在工具调用、角色扮演、问答闲聊等方面也得到了一定幅度的能力提升。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +163840 + +最大输出长度 + +163840 + +上下文长度 + +163840 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2 + +每百万tokens + +输出 + +8 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +500,000 + +## vanchin/deepseek-v3 + +DeepSeek-V3 由深度求索（DeepSeek）于 2024 年 12 月发布，是目前开源社区领先的混合专家（MoE）语言模型：总参数 671B，每个 token 仅激活 37B 参数。模型在 14.8 万亿高质量 tokens 上完成预训练，原生支持 128k 上下文。通过创新的无辅助损失负载均衡策略、多头潜在注意力（MLA）架构和 FP8 混合精度训练。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +131072 + +最大输出长度 + +16384 + +上下文长度 + +131072 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2 + +每百万tokens + +输出 + +8 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-flash-us.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-flash-us.md new file mode 100644 index 00000000..9da1cfa5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-flash-us.md @@ -0,0 +1,107 @@ +# deepseek-v4-flash-us + +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1000000 + +最大输出长度 + +393216 + +上下文长度 + +1000000 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 美国（弗吉尼亚） + +部署范围：美国 + +计费项 + +价格（元） + +单位 + +输入 + +1.499 + +每百万tokens + +输出 + +2.998 + +每百万tokens + +输入（缓存命中） + +0.3 + +每百万tokens diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-flash.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-flash.md new file mode 100644 index 00000000..f3c4bf43 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-flash.md @@ -0,0 +1,297 @@ +# deepseek-v4-flash + +高效轻量化MoE模型，总参284B，激活13B，原生支持百万超长上下文能力。推理速度快、延迟低、调用成本低廉，综合能力均衡，主打高并发、轻量化任务，适合日常对话、内容创作、基础 RAG、批量文案处理等普惠刚需场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1000000 + +最大输出长度 + +393216 + +上下文长度 + +1000000 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +1 + +每百万tokens + +输出 + +2 + +每百万tokens + +输入（缓存命中） + +0.2 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入 + +1.499 + +每百万tokens + +输出 + +2.998 + +每百万tokens + +输入（缓存命中） + +0.3 + +每百万tokens + +## 德国（法兰克福） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +1 + +每百万tokens + +输出 + +2 + +每百万tokens + +输入（缓存命中） + +0.2 + +每百万tokens + +## 美国（弗吉尼亚） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +1 + +每百万tokens + +输出 + +2 + +每百万tokens + +输入（缓存命中） + +0.2 + +每百万tokens + +## 日本（东京） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +1 + +每百万tokens + +输出 + +2 + +每百万tokens + +输入（缓存命中） + +0.2 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10000 + +TPM（每分钟tokens） + +1,200,000 + +## 德国（法兰克福） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## 美国（弗吉尼亚） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## 日本（东京） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-pro-us.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-pro-us.md new file mode 100644 index 00000000..b7a67afe --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-pro-us.md @@ -0,0 +1,107 @@ +# deepseek-v4-pro-us + +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1000000 + +最大输出长度 + +393216 + +上下文长度 + +1000000 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 美国（弗吉尼亚） + +部署范围：美国 + +计费项 + +价格（元） + +单位 + +输入 + +17.986 + +每百万tokens + +输出 + +35.972 + +每百万tokens + +输入（缓存命中） + +1.499 + +每百万tokens diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-pro.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-pro.md new file mode 100644 index 00000000..3ba6d7a1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/deepseek-v4-pro.md @@ -0,0 +1,427 @@ +# deepseek-v4-pro + +旗舰级 MoE 大模型，总参1.6T、激活 49B，原生支持百万级超长上下文。依托海量高质量训练数据，具备顶尖数学逻辑、复杂推理、专业代码与长文本深度解析能力，适配高阶科研、复杂办公、深度智能代理等高难度场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1000000 + +最大输出长度 + +393216 + +上下文长度 + +1000000 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +12 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入 + +17.986 + +每百万tokens + +输出 + +35.972 + +每百万tokens + +输入（缓存命中） + +1.499 + +每百万tokens + +## 德国（法兰克福） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +12 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1 + +每百万tokens + +## 美国（弗吉尼亚） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +12 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1 + +每百万tokens + +## 日本（东京） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +12 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +10000 + +TPM（每分钟tokens） + +1,200,000 + +## 德国（法兰克福） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## 美国（弗吉尼亚） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## 日本（东京） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +15000 + +TPM（每分钟tokens） + +1,200,000 + +## vanchin/deepseek-v4-pro + +DeepSeek-V4系列是强大的混合专家（MoE）语言模型，包含DeepSeek-V4-Pro（1.6T总参数，49B激活参数）。支持高达100万（1M）token的上下文长度，是在超过32T高质量多样化token上预训练的开源模型。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +393216 + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +最大输出长度（思考模式下） + +393216 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +12 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +300,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/farui-plus.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/farui-plus.md new file mode 100644 index 00000000..549f4f76 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/farui-plus.md @@ -0,0 +1,115 @@ +# farui-plus + +通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分析、生成法律文书、检索法律知识、审查合同条款等功能 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +12000 + +最大输出长度 + +2000 + +上下文长度 + +12000 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +20 + +每百万tokens + +输出 + +20 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +240 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-5-air.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-5-air.md new file mode 100644 index 00000000..0ccabaec --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-5-air.md @@ -0,0 +1,137 @@ +# glm-4.5-air + +GLM-4.5-Air采用混合专家（MoE）架构，总参数量为1060亿，激活参数120亿，相较GLM-4.5更紧凑、轻量，适用于对模型规模和资源消耗有一定限制的场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +16384 + +上下文长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +0.8 + +每百万tokens + +输出 + +6 + +每百万tokens + +**32k<输入<=128k** + +计费项 + +价格（元） + +单位 + +输入 + +1.2 + +每百万tokens + +输出 + +8 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-6.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-6.md new file mode 100644 index 00000000..fa746387 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-6.md @@ -0,0 +1,161 @@ +# glm-4.6 + +GLM新一代旗舰模型，核心能力较4.5全面提升。总参数量为3550 亿，激活参数320亿，上下文窗口扩展至200K。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +169984 + +最大输出长度 + +16384 + +上下文长度 + +202752 + +最大输入长度（思考模式下） + +169984 + +最大输出长度（思考模式下） + +16384 + +最大思维链长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +3 + +每百万tokens + +输出 + +14 + +每百万tokens + +输入（缓存命中） + +0.6 + +每百万tokens + +**32k<输入<=200k** + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-7.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-7.md new file mode 100644 index 00000000..cbca8e9b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4-7.md @@ -0,0 +1,149 @@ +# glm-4.7 + +智谱最新旗舰，具备更强的编程能力与更稳定的多步骤推理/执行能力。总参数355B，支持长程任务规划、编码、工具协同，问答自然、写作沉浸、创意角色扮演能力强。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +169984 + +最大输出长度 + +16384 + +上下文长度 + +202752 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +3 + +每百万tokens + +输出 + +14 + +每百万tokens + +输入（缓存命中） + +0.6 + +每百万tokens + +**32k<输入<=200k** + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4.md new file mode 100644 index 00000000..0ac548d6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-4.md @@ -0,0 +1,137 @@ +# glm-4.5 + +GLM-4.5采用混合专家（MoE）架构，总参数量为3550 亿，激活参数320亿，在复杂推理、代码生成及智能体交互等通用能力上实现了能力融合与技术突破。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +98304 + +最大输出长度 + +16384 + +上下文长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +3 + +每百万tokens + +输出 + +14 + +每百万tokens + +**32k<输入<=128k** + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +60 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-1.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-1.md new file mode 100644 index 00000000..e068ce80 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-1.md @@ -0,0 +1,647 @@ +# glm-5.1 + +GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +202745 + +最大输出长度 + +131072 + +上下文长度 + +202745 + +最大输入长度（思考模式下） + +169984 + +最大输出长度（思考模式下） + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +6 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1.2 + +每百万tokens + +显式缓存创建 + +7.5 + +每百万tokens + +显式缓存命中 + +0.6 + +每百万tokens + +**32k<输入<=200k** + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +1.6 + +每百万tokens + +显式缓存创建 + +10 + +每百万tokens + +显式缓存命中 + +0.8 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入 + +10.492 + +每百万tokens + +输出 + +32.974 + +每百万tokens + +输入（缓存命中） + +1.948 + +每百万tokens + +## 德国（法兰克福） + +部署范围：全球 + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +6 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1.2 + +每百万tokens + +显式缓存创建 + +7.5 + +每百万tokens + +显式缓存命中 + +0.6 + +每百万tokens + +**32k<输入<=200k** + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +1.6 + +每百万tokens + +显式缓存创建 + +10 + +每百万tokens + +显式缓存命中 + +0.8 + +每百万tokens + +## 美国（弗吉尼亚） + +部署范围：全球 + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +6 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1.2 + +每百万tokens + +显式缓存创建 + +7.5 + +每百万tokens + +显式缓存命中 + +0.6 + +每百万tokens + +**32k<输入<=200k** + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +1.6 + +每百万tokens + +显式缓存创建 + +10 + +每百万tokens + +显式缓存命中 + +0.8 + +每百万tokens + +## 日本（东京） + +部署范围：全球 + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +6 + +每百万tokens + +输出 + +24 + +每百万tokens + +输入（缓存命中） + +1.2 + +每百万tokens + +显式缓存创建 + +7.5 + +每百万tokens + +显式缓存命中 + +0.6 + +每百万tokens + +**32k<输入<=200k** + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +1.6 + +每百万tokens + +显式缓存创建 + +10 + +每百万tokens + +显式缓存命中 + +0.8 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 德国（法兰克福） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 美国（弗吉尼亚） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 日本（东京） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## ZHIPU/GLM-5.1 + +GLM-5.1 是智谱最新旗舰模型，代码能力大大增强，长程任务显著提升，能够在单次任务中持续、自主地工作长达 8 小时，完成从规划、执行到迭代优化的完整闭环，交付工程级成果。 在综合能力与 Coding 能力上，GLM-5.1 整体表现对齐 Claude Opus 4.6，并在长程自主执行、复杂工程优化与真实开发场景中展现出更强的持续工作能力，是构建 Autonomous Agent 与长程 Coding Agent 的理想基座。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +204800 + +最大输出长度 + +131072 + +上下文长度 + +204800 + +最大输入长度（思考模式下） + +204800 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +2 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +200 + +TPM（每分钟tokens） + +3,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2-fast.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2-fast.md new file mode 100644 index 00000000..f3dc2f54 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2-fast.md @@ -0,0 +1,349 @@ +# glm-5.2-fast-preview + +GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +131072 + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +最大输出长度（思考模式下） + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +16 + +每百万tokens + +输出 + +56 + +每百万tokens + +输入（缓存命中） + +4 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入 + +20.984 + +每百万tokens + +输出 + +65.948 + +每百万tokens + +输入（缓存命中） + +4.196 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 快照版本 + +### glm-5.2-fast-preview + +GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +131072 + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +最大输出长度（思考模式下） + +131072 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +16 + +每百万tokens + +输出 + +56 + +每百万tokens + +输入（缓存命中） + +4 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入 + +20.984 + +每百万tokens + +输出 + +65.948 + +每百万tokens + +输入（缓存命中） + +4.196 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2-us.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2-us.md new file mode 100644 index 00000000..7b563335 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2-us.md @@ -0,0 +1,119 @@ +# glm-5.2-us + +GLM-5.2-US是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +131072 + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 美国（弗吉尼亚） + +部署范围：美国 + +计费项 + +价格（元） + +单位 + +输入 + +10.492 + +每百万tokens + +输出 + +32.974 + +每百万tokens + +输入（缓存命中） + +2.098 + +每百万tokens diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2.md new file mode 100644 index 00000000..b764ae15 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-5-2.md @@ -0,0 +1,399 @@ +# glm-5.2 + +GLM-5.2是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +131072 + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +2 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入 + +10.492 + +每百万tokens + +输出 + +32.974 + +每百万tokens + +输入（缓存命中） + +2.098 + +每百万tokens + +## 德国（法兰克福） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +2 + +每百万tokens + +## 美国（弗吉尼亚） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +2 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +2,000,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 德国（法兰克福） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +2,000,000 + +## 美国（弗吉尼亚） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +2,000,000 + +## ZHIPU/GLM-5.2 + +智谱原厂直供，最新旗舰模型。GLM-5.2 是智谱迄今能力最强的开源模型，支持真正可用的 1M 上下文，并在长程任务中继续保持领先。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +131072 + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +8 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +2 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +200 + +TPM（每分钟tokens） + +3,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-9.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-9.md new file mode 100644 index 00000000..42af4273 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/glm-9.md @@ -0,0 +1,295 @@ +# glm-5 + +GLM-5是面向Coding与Agent场景的新一代大模型，在复杂系统工程与长程任务中达到开源 SOTA，真实编程体验逼近 Claude Opus 级别；基于 744B 新基座、异步强化学习与稀疏注意力，实现从“写代码”到“写工程”的全面升级。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +169984 + +最大输出长度 + +16384 + +上下文长度 + +202752 + +最大输入长度（思考模式下） + +169984 + +最大输出长度（思考模式下） + +16384 + +最大思维链长度 + +32768 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +18 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +**32k<输入<=200k** + +计费项 + +价格（元） + +单位 + +输入 + +6 + +每百万tokens + +输出 + +22 + +每百万tokens + +输入（缓存命中） + +1.2 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## ZHIPU/GLM-5 + +智谱新一代旗舰基座,面向AgenticEngineering,实现从代码到工程的范式跃迁,擅长复杂系统工程与长程Agent任务。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +204800 + +最大输出长度 + +131072 + +上下文长度 + +204800 + +最大输入长度（思考模式下） + +204800 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +6 + +每百万tokens + +输出 + +22 + +每百万tokens + +输入（缓存命中） + +1.5 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +200 + +TPM（每分钟tokens） + +3,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-5.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-5.md new file mode 100644 index 00000000..e340883d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-5.md @@ -0,0 +1,379 @@ +# kimi-k2.5 + +kimi-k2.5是月之暗面迄今发布最全能的模型，原生多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与Agent任务。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +229376 + +最大输出长度 + +16384 + +上下文长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +21 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +显式缓存创建 + +5 + +每百万tokens + +显式缓存命中 + +0.4 + +每百万tokens + +## 德国（法兰克福） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +21 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +显式缓存创建 + +5 + +每百万tokens + +显式缓存命中 + +0.4 + +每百万tokens + +## 美国（弗吉尼亚） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +21 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +显式缓存创建 + +5 + +每百万tokens + +显式缓存命中 + +0.4 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 德国（法兰克福） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 美国（弗吉尼亚） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## kimi/kimi-k2.5 + +Kimi K2.5 是 Kimi 在2026年最新推出的智能模型，在 Agent、代码、视觉理解及一系列通用智能任务上取得开源 SoTA 表现。同时 Kimi K2.5 也是 Kimi 迄今最全能的模型，原生的多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与 Agent 任务。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +262144 + +最大输出长度 + +262144 + +上下文长度 + +262144 + +最大输入长度（思考模式下） + +262144 + +最大输出长度（思考模式下） + +262144 + +最大思维链长度 + +262144 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +21 + +每百万tokens + +输入（缓存命中） + +0.7 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +3,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-6.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-6.md new file mode 100644 index 00000000..36afdaef --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-6.md @@ -0,0 +1,267 @@ +# kimi-k2.6 + +kimi-k2.6是Kimi最新最智能的模型，具备更强更稳的长程代码编写能力，指令遵循和自我纠错能力显著提升，同时支持文本、图片与视频输入，思考与非思考模式，对话与Agent任务。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Text**　**Video** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +229376 + +最大输出长度 + +16384 + +上下文长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +6.5 + +每百万tokens + +输出 + +27 + +每百万tokens + +输入（缓存命中） + +1.3 + +每百万tokens + +显式缓存创建 + +8.125 + +每百万tokens + +显式缓存命中 + +0.65 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## kimi/kimi-k2.6 + +Kimi K2.6 是 Kimi 最新最智能的模型，Kimi K2.6 的通用 Agent、代码、视觉理解等综合能力得到全面提升，其中在博士级难度的完整版人类最后的考试（Humanity’s Last Exam）、在考察模型真实软件工程能力的 SWE-Bench Pro、评估 Agent 深度检索能力的 DeepSearchQA 等基准测试中均取得行业领先的成绩，同时支持文本、图片与视频输入，思考与非思考模式，对话与 Agent 任务。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +262144 + +最大输出长度 + +262144 + +上下文长度 + +262144 + +最大输入长度（思考模式下） + +262144 + +最大输出长度（思考模式下） + +262144 + +最大思维链长度 + +262144 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +6.5 + +每百万tokens + +输出 + +27 + +每百万tokens + +输入（缓存命中） + +1.1 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +3,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-7-code-highspeed.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-7-code-highspeed.md new file mode 100644 index 00000000..9bf20482 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-7-code-highspeed.md @@ -0,0 +1,133 @@ +# kimi/kimi-k2.7-code-highspeed + +K2.7 Code高速版与普通版是同一个模型，但输出速度约为普通版的 5-6 倍，常规编程场景下（取输入长度中位数）输出速度约 180 Token/s，短上下文场景可达 260 Token/s ，带来更极致的编程体验。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +262144 + +最大输出长度 + +262144 + +上下文长度 + +262144 + +最大输入长度（思考模式下） + +262144 + +最大输出长度（思考模式下） + +262144 + +最大思维链长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +13 + +每百万tokens + +输出 + +54 + +每百万tokens + +输入（缓存命中） + +2.6 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +3,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-7-code.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-7-code.md new file mode 100644 index 00000000..878213b1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-7-code.md @@ -0,0 +1,423 @@ +# kimi-k2.7-code + +kimi-k2.7-code是 kimi 迄今最智能的coding模型，在长上下文中更可靠地遵循指令，能以更高的成功率完成编程任务，同时支持文本、图片与视频输入，思考模式，对话与 Agent 任务。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +229376 + +最大输出长度 + +16384 + +上下文长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +6.5 + +每百万tokens + +输出 + +27 + +每百万tokens + +输入（缓存命中） + +1.3 + +每百万tokens + +显式缓存创建 + +8.125 + +每百万tokens + +显式缓存命中 + +0.65 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +计费项 + +价格（元） + +单位 + +输入 + +7.119 + +每百万tokens + +输出 + +29.977 + +每百万tokens + +输入（缓存命中） + +1.424 + +每百万tokens + +## 德国（法兰克福） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +6.5 + +每百万tokens + +输出 + +27 + +每百万tokens + +输入（缓存命中） + +1.3 + +每百万tokens + +显式缓存创建 + +8.125 + +每百万tokens + +显式缓存命中 + +0.65 + +每百万tokens + +## 美国（弗吉尼亚） + +部署范围：全球 + +计费项 + +价格（元） + +单位 + +输入 + +6.5 + +每百万tokens + +输出 + +27 + +每百万tokens + +输入（缓存命中） + +1.3 + +每百万tokens + +显式缓存创建 + +8.125 + +每百万tokens + +显式缓存命中 + +0.65 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 新加坡 + +部署范围：国际 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 德国（法兰克福） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## 美国（弗吉尼亚） + +部署范围：全球 + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## kimi/kimi-k2.7-code + +Kimi K2.7 Code 是月之暗面 Kimi发布并开源的新一代编程专用模型，定位为 Kimi 迄今最智能的 Coding 模型。Kimi K2.7 Code 是一个以编码为中心的智能体模型（coding-focused agentic model），专为长程软件工程任务优化。它擅长跨多文件重构、功能实现、长会话调试等需要可靠指令遵循和端到端完成率的复杂工作流。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +262144 + +最大输出长度 + +262144 + +上下文长度 + +262144 + +最大输入长度（思考模式下） + +262144 + +最大输出长度（思考模式下） + +262144 + +最大思维链长度 + +262144 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +6.5 + +每百万tokens + +输出 + +27 + +每百万tokens + +输入（缓存命中） + +1.3 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +3,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-thinking.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-thinking.md new file mode 100644 index 00000000..356d32d7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k2-thinking.md @@ -0,0 +1,121 @@ +# kimi-k2-thinking + +kimi-k2-thinking模型是月之暗面提供的具有通用 Agentic能力和推理能力的思考模型，它擅长深度推理，并可通过多步工具调用，帮助解决各类难题。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +229376 + +最大输出长度 + +16384 + +上下文长度 + +262144 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k3.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k3.md new file mode 100644 index 00000000..5843285d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/kimi-k3.md @@ -0,0 +1,133 @@ +# kimi/kimi-k3 + +Kimi K3 是 Kimi 迄今能力最强的旗舰模型，拥有 2.8 万亿参数，基于 KDA 混合线性注意力机制（Kimi Delta Attention）和注意力残差（Attention Residuals）技术构建，原生支持视觉理解，并拥有 100 万 token 上下文窗口。它是全球首个开源的 3 万亿级别模型，面向长程编程、知识工作和推理等前沿智能场景而设计。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text**　**Image**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +1048576 + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +最大输出长度（思考模式下） + +1048576 + +最大思维链长度 + +1048576 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +20 + +每百万tokens + +输出 + +100 + +每百万tokens + +输入（缓存命中） + +2 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +3,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/mimo-v2-5-pro.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/mimo-v2-5-pro.md new file mode 100644 index 00000000..6ba047ba --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/mimo-v2-5-pro.md @@ -0,0 +1,133 @@ +# xiaomi/mimo-v2.5-pro + +MiMo-V2.5-Pro 是小米发布的最新旗舰模型。与前代模型相比，它在通用智能体能力、复杂软件工程以及长程任务等方面都有显著提升，在 ClawEval、GDPVal 和 SWE-bench Pro 等基准测试中均位列前茅。它能够独立且完全自主地完成需要人类专家耗时数天甚至数周的专业任务，涉及上千次工具调用。其高达 100 万 token 的上下文长度，非常适合集成到各种智能体框架中使用。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +131072 + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +7 + +每百万tokens + +输出 + +21 + +每百万tokens + +输入（缓存命中） + +1.4 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +100 + +TPM（每分钟tokens） + +10,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2-5.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2-5.md new file mode 100644 index 00000000..dfe2f3c2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2-5.md @@ -0,0 +1,255 @@ +# MiniMax-M2.5 + +MiniMax-M2.5是MiniMax推出的旗舰级开源大模型，经过数十万个真实复杂环境中的大规模强化学习训练，M2.5 在编程、工具调用和搜索、办公等生产力场景都达到或者刷新了行业的 SOTA。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +196608 + +最大输出长度 + +131072 + +上下文长度 + +204800 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2.1 + +每百万tokens + +输出 + +8.4 + +每百万tokens + +输入（缓存命中） + +0.42 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +1,000,000 + +## MiniMax/MiniMax-M2.5 + +智能体世界的SOTA，专为智能体2.0设计，将编码扩展到现实世界包括工作空间、娱乐和个人助理。模型亮点：全球SOTA开源编码与智能体模型；SWE-bench Pro和SWE-bench Verified得分高于Opus 4.6；在Excel、搜索与研究以及文档摘要方面的全球SOTA；未来工作空间的完美主力模型；闪电般快速：优化思维效率，100+ TPS，实现比 Opus 快 3 倍的速度；极致性价比，以支持始终在线的智能体。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +204800 + +最大输出长度 + +131072 + +上下文长度 + +204800 + +最大输入长度（思考模式下） + +204800 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2.1 + +每百万tokens + +输出 + +8.4 + +每百万tokens + +输入（缓存命中） + +0.21 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +20,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2-7.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2-7.md new file mode 100644 index 00000000..4b89345b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2-7.md @@ -0,0 +1,133 @@ +# MiniMax/MiniMax-M2.7 + +M2.7 能够自行构建复杂 Agent Harness，并基于 Agent Teams、复杂 Skills、Tool Search tool 等能力，完成高度复杂的生产力任务。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +204800 + +最大输出长度 + +131072 + +上下文长度 + +204800 + +最大输入长度（思考模式下） + +204800 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2.1 + +每百万tokens + +输出 + +8.4 + +每百万tokens + +输入（缓存命中） + +0.42 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +20,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2.md new file mode 100644 index 00000000..15867b9f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m2.md @@ -0,0 +1,239 @@ +# MiniMax-M2.1 + +MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，以多语言编程与长链 Agent 能力为核心优势。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +172032 + +最大输出长度 + +32768 + +上下文长度 + +204800 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2.1 + +每百万tokens + +输出 + +8.4 + +每百万tokens + +输入（缓存命中） + +0.42 + +每百万tokens + +## MiniMax/MiniMax-M2.1 + +M2.1 的设计初衷在于打破“最顶级的 Agent 能力仅存在于闭源模型”的壁垒。我们在模型层面进行了针对性优化，显著提升了模型在代码生成、工具调用、复杂指令遵循及长程规划任务中的性能。从自动化进行多语言的软件开发，到执行多步骤的复杂办公工作流，MiniMax-M2.1 均表现出卓越的稳定性。我们致力于为开发者提供一个完全透明、可控且高可用的基础模型，以构建下一代自主智能体应用。 + +### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +204800 + +最大输出长度 + +131072 + +上下文长度 + +204800 + +最大输入长度（思考模式下） + +204800 + +最大输出长度（思考模式下） + +131072 + +最大思维链长度 + +131072 + +### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +2.1 + +每百万tokens + +输出 + +8.4 + +每百万tokens + +输入（缓存命中） + +0.21 + +每百万tokens + +### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +20,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m3.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m3.md new file mode 100644 index 00000000..763b226c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/minimax-m3.md @@ -0,0 +1,125 @@ +# MiniMax/MiniMax-M3 + +MiniMax M3 凭借业界领先的 Coding 与 Agentic 能力、1M 超长上下文窗口以及原生多模态特性，可出色胜任企业级长文档理解、高质量内容生成、代码编写、Bug 修复及原生应用构建等任务；强大的 Agentic 能力端到端贯通工作流，原生多模态更带来流畅自然的图文混合交互体验。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Image**　**Text**　**Video** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +1048576 + +最大输出长度 + +— + +上下文长度 + +1048576 + +最大输入长度（思考模式下） + +1048576 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +4.2 + +每百万tokens + +输出 + +16.8 + +每百万tokens + +输入（缓存命中） + +0.84 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +500 + +TPM（每分钟tokens） + +20,000,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/model-qwen-deep-research.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/model-qwen-deep-research.md new file mode 100644 index 00000000..a7970b08 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/model-qwen-deep-research.md @@ -0,0 +1,233 @@ +# qwen-deep-research + +千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +997952 + +最大输出长度 + +32768 + +上下文长度 + +1000000 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +54 + +每百万tokens + +输出 + +163 + +每百万tokens + +## 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +TPM（每分钟tokens） + +1,200,000 + +## 快照版本 + +### qwen-deep-research-2025-12-15 + +千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。 + +#### 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +不支持 + +Function Calling + +不支持 + +结构化输出 + +不支持 + +联网搜索 + +不支持 + +前缀续写 + +不支持 + +上下文缓存 + +不支持 + +批量推理 + +不支持 + +模型调优 + +不支持 + +#### 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +997952 + +最大输出长度 + +32768 + +上下文长度 + +1000000 + +#### 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +计费项 + +价格（元） + +单位 + +输入 + +79 + +每百万tokens + +输出 + +236 + +每百万tokens + +#### 限流 + +## 华北2（北京） + +参数 + +值 + +RPM（每分钟请求数） + +120 + +TPM（每分钟tokens） + +1,200,000 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/model-qwen3-max.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/model-qwen3-max.md new file mode 100644 index 00000000..c0d0f101 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/model-studio-model-list/model-list-text-generation/model-qwen3-max.md @@ -0,0 +1,2039 @@ +# qwen3-max + +千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。该模型版本功能等同于快照模型 qwen3-max-2026-01-23。 + +## 模型能力 + +能力项 + +支持情况 + +能力项 + +支持情况 + +输入模态 + +**Text** + +输出模态 + +**Text** + +模型体验 + +支持 + +Function Calling + +支持 + +结构化输出 + +支持 + +联网搜索 + +支持 + +前缀续写 + +支持 + +上下文缓存 + +支持 + +批量推理 + +支持 + +模型调优 + +不支持 + +## 上下文限制 + +参数 + +值 + +参数 + +值 + +最大输入长度 + +258048 + +最大输出长度 + +65536 + +上下文长度 + +262144 + +最大输入长度（思考模式下） + +258048 + +最大输出长度（思考模式下） + +32768 + +最大思维链长度 + +81920 + +## 模型价格 + +本文仅展示模型调用原价，不包含限时优惠等活动信息，请前往[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)查看活动优惠。 + +## 华北2（北京） + +**输入<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +2.5 + +每百万tokens + +输出 + +10 + +每百万tokens + +输入（缓存命中） + +0.5 + +每百万tokens + +输入（Batch File） + +1.25 + +每百万tokens + +输出（Batch File） + +5 + +每百万tokens + +显式缓存创建 + +3.125 + +每百万tokens + +显式缓存命中 + +0.25 + +每百万tokens + +输入（Batch Chat) + +2.5 + +每百万tokens + +输出（Batch Chat) + +10 + +每百万tokens + +**32k<输入<=128k** + +计费项 + +价格（元） + +单位 + +输入 + +4 + +每百万tokens + +输出 + +16 + +每百万tokens + +输入（缓存命中） + +0.8 + +每百万tokens + +输入（Batch File） + +2 + +每百万tokens + +输出（Batch File） + +8 + +每百万tokens + +显式缓存创建 + +5 + +每百万tokens + +显式缓存命中 + +0.4 + +每百万tokens + +输入（Batch Chat) + +4 + +每百万tokens + +输出（Batch Chat) + +16 + +每百万tokens + +**128k<输入<=256k** + +计费项 + +价格（元） + +单位 + +输入 + +7 + +每百万tokens + +输出 + +28 + +每百万tokens + +输入（缓存命中） + +1.4 + +每百万tokens + +输入（Batch File） + +3.5 + +每百万tokens + +输出（Batch File） + +14 + +每百万tokens + +显式缓存创建 + +8.75 + +每百万tokens + +显式缓存命中 + +0.7 + +每百万tokens + +输入（Batch Chat) + +7 + +每百万tokens + +输出（Batch Chat) + +28 + +每百万tokens + +## 新加坡 + +部署范围：国际 + +**Input<=32k** + +计费项 + +价格（元） + +单位 + +输入 + +8.807 + +每百万tokens + +输出 + +44.035 + +每百万tokens + +输入（缓存命中） + +1.761 + +每百万tokens + +显式缓存创建 + +11.009 + +每百万tokens + +显式缓存命中 + +0.881 + +每百万tokens + +**32k [**费用概览**](https://bailian.console.aliyun.com/?tab=model#/costing-balance/overview)，选择**账期月份**： +登录[百炼控制台](https://bailian.console.aliyun.com/?tab=model)，单击顶部**模型**标签页，选择[**费用概览**](https://bailian.console.aliyun.com/?tab=model#/costing-balance)，选择**账期月份**： > 该页面仅展示**大模型推理**相关费用。**模型训练**和**知识库**等费用请通过[账单详情](#29f8b9b9a4lmc)查看。 @@ -31,14 +31,18 @@ 2. 选择**产品名称**为**大模型服务平台百炼**，单击**搜索**。 -3. 单击账单列表右上角的导出图标，将账单下载到本地。 +3. 单击页面顶部的**导出明细**，将账单下载到本地。 4. 打开文件，找到 实例 ID（出账粒度）列，根据下文规则进行核对。 -#### **2\. 解读关键字段** +#### **2\. 解读关键字段（模型推理账单）** -**“实例 ID（出账粒度）”字段**以英文分号 `;` 分隔，完整格式为`ApiKeyID;业务空间 ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`。 +**说明** + +以下字段格式说明适用于**模型推理**账单。**模型训练与调优**账单的实例ID字段格式不同（使用英文感叹号 `!` 分隔），请参见下方[训练与调优账单字段](#wi82451543-train-billing-h)。 + +**“实例 ID（出账粒度）”字段**以英文分号 `;` 分隔，完整格式为`ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`。 - 格式 A：标准调用（包含ApiKeyID） @@ -64,11 +68,21 @@ - 多个标签之间用英文分号 ; 隔开。 -#### **3\. 数据溯源与术语说明** +#### **3\. 训练与调优账单字段** + +**模型训练**与**模型调优**账单的**“实例 ID（出账粒度）”字段**以英文感叹号 `!` 分隔，常见格式为`业务空间ID!地域!训练任务标识`，与模型推理账单的分号 `;` 分隔格式不同。 + +**示例：**`ws-xxxxxxxxx!cn-beijing!qwen3-14b-ft-2026051510-1c3b` + +**说明** + +账单的**商品名称**字段可直接区分计费类型：训练与调优费用对应**商品名称**为`百炼大模型训练`。若无法确定对应的训练任务，可结合**账单月份**与**模型调优**页面的任务提交时间进行匹配。 + +#### **4\. 数据溯源与术语说明** - 查询 API Key：复制账单中的 `ApiKeyID`，前往[百炼API Key管理](https://bailian.console.aliyun.com/?tab=model#/api-key)页面查找对应的 Key 名称。 -- 查询业务空间：复制账单中的 `业务空间ID`，前往[百炼控制台](https://bailian.console.aliyun.com/?tab=model#/api-key)，点击左侧菜单底部的**默认业务空间**，点击**业务空间详情**，确认具体空间ID。您也可以切换到其他业务空间。 +- 查询业务空间：复制账单中的 `业务空间ID`，前往[业务空间管理](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)页面，在列表的**业务空间ID**列确认具体空间。当前所处的业务空间显示在控制台顶部右上角的**默认业务空间**切换器中，悬停可查看其地域、业务空间ID与创建时间等详情。 - 调用渠道说明： @@ -83,13 +97,13 @@ 给**业务空间**绑定**标签**，可按部门或项目归集费用。 -1. **获取业务空间信息**：在[**业务空间管理**](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)确定标签绑定的业务空间**Workspace ID**（示例：llm-xxx），并在[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)确定业务空间的**地域**信息。 +1. **获取业务空间信息**：在[**业务空间管理**](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)确定标签绑定的**业务空间ID**（示例：llm-xxx），并在[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)确定业务空间的**地域**信息。 2. **绑定标签**： 1. 在[**标签管理**](https://resourcemanager.console.aliyun.com/tags#/)页面选择**资源绑定标签。** - 2. 资源选择方式选择“**输入多个资源ID**”，在产品选项卡搜索并选择“**大模型服务平台百炼:业务空间**”并选择业务空间对应地域，资源ID输入框中填写**Workspace ID**，完成后点击绑定标签按钮执行操作。 + 2. 资源选择方式选择“**输入多个资源ID**”，在产品选项卡搜索并选择“**大模型服务平台百炼:业务空间**”并选择业务空间对应地域，资源ID输入框中填写**业务空间ID**，完成后点击绑定标签按钮执行操作。 3. 在绑定标签页面中，创建标签键值或使用已创建的预置标签与业务空间绑定。当完成键值输入或选择好预置标签后，点击**确定**完成业务空间标签的绑定。 @@ -102,16 +116,14 @@ **账户可用额度 < 0** 视为欠费，可能导致模型调用等服务暂停。在[费用与成本首页](https://billing-cost.console.aliyun.com/home)悬停**账户可用额度**区域可查看，公式为：可用额度 =（现金余额 + 信控额度）-（当月未结清 + 历史未结清）。 -- **欠费影响**：按账单**商品名称**维度判定。 +- **欠费影响**：账户欠费将导致**按量付费（后付费）**模型调用等服务暂停，能否继续使用取决于计费方式。 - - 仍有**免费额度**：可继续使用，用完后停用。 - - - 仍有**节省计划**或**资源包**额度：可继续使用。 + - **免费额度、节省计划、资源包**：三者均用于抵扣按量付费费用，**欠费期间即使仍有剩余额度，也无法调用模型**，需结清欠费后恢复。 - 已购 **Coding Plan** 或 **Token Plan**：套餐额度独立于账户余额，欠费期间可继续使用，但会导致自动续费失败，到期后无法续用。 - - 以上额度均无：该商品下服务将**暂停**，需结清欠费后恢复。 - +- **代金券与余额说明**：代金券（含学生权益优惠券）不计入账户可用额度中的现金余额，后付费场景下系统会冻结当月消费金额，账户实际余额须大于当月冻结金额方可正常调用模型，详情参见[代金券说明文档](https://help.aliyun.com/zh/user-center/voucher-management)。 + - **结清欠费**：在[费用与成本](https://usercenter2.aliyun.com/home)页面单击**充值汇款**，输入金额并完成支付。 - **预防欠费**：在[高额消费预警](https://usercenter2.aliyun.com/home/alarm-threshold)页面设置消费阈值，达阈值即提醒。 @@ -121,21 +133,21 @@ 不再使用百炼时，按以下方式关停对应服务即可停止计费。 -- **停止模型推理**：停止代码中的 API 调用、关闭控制台的模型体验，即不再产生费用。为防止意外调用，可在[**API-KEY**](https://bailian.console.aliyun.com/?apiKey=1&tab=globalset#/efm/api_key)页面删除已创建的 Key。 +- **停止模型推理**：停止代码中的 API 调用、关闭控制台的模型体验，即不再产生费用。为防止意外调用，可在[**API-KEY**](https://bailian.console.aliyun.com/?tab=model#/api-key)页面删除已创建的 Key。 - **停止模型训练**：没有正在进行的训练任务时即不产生费用。 -- **取消 Coding Plan 订阅**：Coding Plan 为包月订阅产品，到期自动停止，中途不支持取消和退款。如已开启自动续费，请在[Coding Plan](https://bailian.console.aliyun.com/?tab=model#/efm/coding_plan) 页面关闭自动续费。 +- **停止 Coding Plan 计费**：Coding Plan 为包月订阅产品，到期自动停止，中途不支持取消和退款。如已开启自动续费，请在[Coding Plan](https://bailian.console.aliyun.com/?tab=model#/efm/subscription/coding-plan) 页面关闭自动续费。 - **退订 Token Plan 团队版**：在[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)的**我的订阅**页面按席位退订，未产生用量消耗的席位可退订，退款原路退回支付账户。如不再续费，请关闭自动续费。 - **停止模型部署**：根据部署时的计费方式操作不同： - - 按模型调用量计费：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型，或删除[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)防止意外调用。 + - **按 Token 调用计费（后付费）**：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型，或删除[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)防止意外调用。 - - 按算力使用时长计费：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型。 + - **按算力单元或模型单元计费（后付费）**：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型。 - - 包月预付费：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型，然后在[退订管理](https://usercenter2.aliyun.com/refund/refund)页面退订实例。退订时按已消费金额扣减，退回剩余金额（详见[退订说明](https://help.aliyun.com/zh/user-center/user-guide/refund-management/)）。 + - **按预置吞吐单元计费（预付费）**：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型，然后在[退订管理](https://usercenter2.aliyun.com/refund/refund)页面退订实例。退订时已使用部分按 1.5 倍系数结算（详见退订说明），退回剩余金额。 ## **常见问题** @@ -161,11 +173,11 @@ **原因：**账单的“计费项”统一显示为“大模型文本消耗量”，未直接展示模型名称。 -**解决方案：**查看[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)页的**实例 ID（出账粒度）**列。字段以英文分号分隔，紧跟业务空间 ID（如 llm-xxx）之后的字段即为模型名称。例：`12xxx;llm-xxx;**qwen3.6-plus**;context_0-128k_input_token;bmp;0`表示 qwen3.6-plus 模型。 +**解决方案：**查看[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)页的**实例 ID（出账粒度）**列。字段以英文分号分隔，紧跟业务空间ID（如 llm-xxx）之后的字段即为模型名称。例：`12xxx;llm-xxx;**qwen3.6-plus**;context_0-128k_input_token;bmp;0`表示 qwen3.6-plus 模型。 **在哪里查看模型调用次数和统计？** -进入[阿里云百炼控制台](https://bailian.console.aliyun.com/?tab=model)，右上角选择目标地域，单击顶部**模型**标签页，在左侧菜单选择**用量 & 费用** > [模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)。 +进入[阿里云百炼控制台](https://bailian.console.aliyun.com/?tab=model)，右上角选择目标地域，单击顶部**模型**标签页，选择[模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)。 **按量付费是实时扣款吗？** @@ -190,3 +202,27 @@ 2. 检查应用代码或百炼应用配置中是否开启了 `enable_search`，如不再需要联网搜索，将该参数设为 `false` 或移除。 3. 如已停止所有调用但仍有扣费，检查是否有其他 API Key 或应用仍在运行，可在[API Key 管理](https://bailian.console.aliyun.com/?tab=model#/api-key)页面逐一排查或删除不再使用的 Key。 + + +**为什么没有主动调用 API 也会产生费用？** + +**原因：**百炼的模型部署按使用时长计费，模型完成部署即状态为**运行中**时开始收费，不依赖 API 调用。即使未主动通过 API 调用该模型，只要部署状态为**运行中**就会持续产生费用。 + +**解决方案：** + +- 前往[**模型部署**](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)页面，下线不再使用的已部署模型，停止按时长计费。 + +- 如需防止意外调用产生推理费用，可在[**API-KEY**](https://bailian.console.aliyun.com/?tab=model#/api-key)页面删除不再使用的 Key（注意：删除后无法恢复，请谨慎操作）。 + + +**如何判断账户是否被盗用？** + +如果怀疑账户被他人盗用产生非预期费用，按以下步骤排查： + +1. 在[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)中筛选**大模型服务平台百炼**，查看**实例 ID（出账粒度）**列中的 `ApiKeyID`，确认产生费用的 API Key。 + +2. 前往[API Key 管理页面](https://bailian.console.aliyun.com/?tab=model#/api-key)，核对每个 Key 的**创建时间**，确认是否为本人创建。API Key 管理页面仅显示创建时间，不显示调用时间。 + +3. 查看调用时段分布，判断是否存在非本人操作的异常调用模式：进入[模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)页面，按**模型**或 **API Key ID** 筛选，切换至**列表**视图查看调用时间分布。 + +4. 如发现未授权调用，立即在[API Key 管理页面](https://bailian.console.aliyun.com/?tab=model#/api-key)删除对应 API Key 并重新生成。更新所有合法调用方使用新 Key。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md index 7137579d..7536077e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md @@ -28,14 +28,12 @@ **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **模式** **单次请求的输入Token数** @@ -48,7 +46,7 @@ **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3.7-max @@ -58,8 +56,6 @@ qwen3.7-max > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 当前能力等同于qwen3.7-max-2026-05-17 -中国内地 - 仅思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 0 当前能力等同于qwen3.6-plus-2026-04-02 -中国内地 - 0 当前能力等同于qwen3.5-plus-2026-02-15 -中国内地 - 0 [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0 [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0 当前能力等同于qwen3.7-flash-2026-07-15 + +> [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 + +> [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 + +非思考和思考模式 + +0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 当前能力等同于qwen3.5-flash-2026-02-23 @@ -2866,6 +2860,58 @@ qwen-flash-2025-07-28-us > **思维链+回答** +qwen3.7-flash + +> 当前能力等同于qwen3.7-flash-2026-07-15 + +> [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 + +国际 + +非思考和思考模式 + +0 当前能力等同于qwen3.6-flash-2026-04-16 @@ -3070,6 +3116,8 @@ qwen3.5-flash 非思考和思考模式 +0 [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0.5元 2元 @@ -3386,8 +3424,6 @@ qwen-long qwen-long-latest -中国内地 - 0.5元 2元 @@ -3396,8 +3432,6 @@ qwen-long-latest qwen-long-2025-01-25 -中国内地 - 0.5元 2元 @@ -3410,21 +3444,19 @@ qwen-long-2025-01-25 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **文本/图片/视频** @@ -3442,8 +3474,6 @@ qwen3.5-omni-plus > 当前能力等同于qwen3.5-omni-plus-2026-03-15 -中国内地 - 7元 53元 @@ -3456,8 +3486,6 @@ qwen3.5-omni-plus qwen3.5-omni-plus-2026-03-15 -中国内地 - 7元 53元 @@ -3472,8 +3500,6 @@ qwen3.5-omni-flash > 当前能力等同于qwen3.5-omni-flash-2026-03-15 -中国内地 - 2.2元 18元 @@ -3486,8 +3512,6 @@ qwen3.5-omni-flash qwen3.5-omni-flash-2026-03-15 -中国内地 - 2.2元 18元 @@ -3502,8 +3526,6 @@ qwen3.5-omni-flash-2026-03-15 **模型 ID（Model ID）** -**服务部署范围** - **模式** **输入单价（每百万Token）** @@ -3512,7 +3534,7 @@ qwen3.5-omni-flash-2026-03-15 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **文本** @@ -3536,8 +3558,6 @@ qwen3-omni-flash > 当前能力等同于qwen3-omni-flash-2025-12-01 -中国内地 - 非思考和思考模式 1.8元 @@ -3556,8 +3576,6 @@ qwen3-omni-flash qwen3-omni-flash-2025-12-01 -中国内地 - 非思考和思考模式 1.8元 @@ -3576,8 +3594,6 @@ qwen3-omni-flash-2025-12-01 qwen3-omni-flash-2025-09-15 -中国内地 - 非思考和思考模式 1.8元 @@ -3598,8 +3614,6 @@ qwen-omni-turbo > 当前能力等同于qwen-omni-turbo-2025-03-26 -中国内地 - 非思考模式 0.4元 @@ -3618,8 +3632,6 @@ qwen-omni-turbo qwen-omni-turbo-latest -中国内地 - 非思考模式 0.4元 @@ -3638,8 +3650,6 @@ qwen-omni-turbo-latest qwen-omni-turbo-2025-03-26 -中国内地 - 非思考模式 0.4元 @@ -3658,8 +3668,6 @@ qwen-omni-turbo-2025-03-26 qwen-omni-turbo-2025-01-19 -中国内地 - 非思考模式 0.4元 @@ -3898,21 +3906,19 @@ qwen-omni-turbo-2025-03-26 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **文本/图片** @@ -3930,8 +3936,6 @@ qwen3.5-omni-plus-realtime > 当前能力等同于qwen3.5-omni-plus-realtime-2026-03-15 -中国内地 - 10元 80元 @@ -3944,8 +3948,6 @@ qwen3.5-omni-plus-realtime qwen3.5-omni-plus-realtime-2026-03-15 -中国内地 - 10元 80元 @@ -3960,8 +3962,6 @@ qwen3.5-omni-flash-realtime > 当前能力等同于qwen3.5-omni-flash-realtime-2026-03-15 -中国内地 - 3.3元 27元 @@ -3974,8 +3974,6 @@ qwen3.5-omni-flash-realtime qwen3.5-omni-flash-realtime-2026-03-15 -中国内地 - 3.3元 27元 @@ -3990,15 +3988,13 @@ qwen3.5-omni-flash-realtime-2026-03-15 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **文本** @@ -4022,8 +4018,6 @@ qwen3-omni-flash-realtime > 当前能力等同于qwen3-omni-flash-realtime-2025-12-01 -中国内地 - 2.2元 18.9元 @@ -4040,8 +4034,6 @@ qwen3-omni-flash-realtime qwen3-omni-flash-realtime-2025-12-01 -中国内地 - 2.2元 18.9元 @@ -4058,8 +4050,6 @@ qwen3-omni-flash-realtime-2025-12-01 qwen3-omni-flash-realtime-2025-09-15 -中国内地 - 2.2元 18.9元 @@ -4078,8 +4068,6 @@ qwen-omni-turbo-realtime > 当前能力等同于qwen-omni-turbo-realtime-2025-05-08 -中国内地 - 1.6元 25元 @@ -4096,8 +4084,6 @@ qwen-omni-turbo-realtime qwen-omni-turbo-realtime-latest -中国内地 - 1.6元 25元 @@ -4114,8 +4100,6 @@ qwen-omni-turbo-realtime-latest qwen-omni-turbo-realtime-2025-05-08 -中国内地 - 1.6元 25元 @@ -4338,26 +4322,22 @@ qwen-omni-turbo-realtime-2025-05-08 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qvq-max -中国内地 - 8元 32元 @@ -4366,8 +4346,6 @@ qvq-max qvq-plus -中国内地 - 2元 5元 @@ -4400,14 +4378,12 @@ qvq-max **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **模式** **单次请求的输入Token数** @@ -4420,7 +4396,7 @@ qvq-max **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-vl-plus @@ -4430,8 +4406,6 @@ qwen3-vl-plus > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 无阶梯计价 1.6元 @@ -4624,8 +4584,6 @@ qwen-vl-plus > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 无阶梯计价 0.8元 @@ -5280,26 +5238,22 @@ qwen3-vl-plus-2025-09-23 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3.5-ocr -中国内地 - 0.5元 2元 @@ -5312,8 +5266,6 @@ qwen-vl-ocr > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0.3元 0.5元 @@ -5324,8 +5276,6 @@ qwen-vl-ocr-latest > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0.3元 0.5元 @@ -5334,8 +5284,6 @@ qwen-vl-ocr-latest qwen-vl-ocr-2025-11-20 -中国内地 - 0.3元 0.5元 @@ -5344,8 +5292,6 @@ qwen-vl-ocr-2025-11-20 qwen-vl-ocr-2025-08-28 -中国内地 - 5元 5元 @@ -5354,8 +5300,6 @@ qwen-vl-ocr-2025-08-28 qwen-vl-ocr-2025-04-13 -中国内地 - 5元 5元 @@ -5364,8 +5308,6 @@ qwen-vl-ocr-2025-04-13 qwen-vl-ocr-2024-10-28 -中国内地 - 5元 5元 @@ -5466,20 +5408,16 @@ qwen-vl-ocr-2025-11-20 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-audio-turbo -中国内地 - 目前仅供免费体验。 > 免费额度用完后不可调用，推荐使用[全模态（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni)作为替代模型 @@ -5488,8 +5426,6 @@ qwen-audio-turbo qwen-audio-turbo-latest -中国内地 - ### 千问数学模型 计费规则：按输入Token和输出Token计费。 @@ -5498,20 +5434,16 @@ qwen-audio-turbo-latest **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-math-plus -中国内地 - 4元 12元 @@ -5520,8 +5452,6 @@ qwen-math-plus qwen-math-turbo -中国内地 - 2元 6元 @@ -5534,14 +5464,12 @@ qwen-math-turbo **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **单次请求的输入Token数** **输入单价（每百万Token）** @@ -5550,7 +5478,7 @@ qwen-math-turbo **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-coder-plus @@ -5558,8 +5486,6 @@ qwen3-coder-plus > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 0 当前能力等同于qwen3-coder-flash-2025-07-28 -中国内地 - 0 免费额度用完后不可调用，推荐使用[全模态（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni)作为替代模型。 @@ -8260,22 +8076,18 @@ qwen2-audio-instruct qwen-audio-chat -中国内地 - ### **Qwen-Coder** 计费规则：按输入Token和输出Token计费。 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **单次请求的输入Token数** **输入单价（每百万Token）** @@ -8284,12 +8096,10 @@ qwen-audio-chat **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-coder-next -中国内地 - 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 12元 24元 @@ -8614,8 +8416,6 @@ deepseek-v4-flash > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 1元 2元 @@ -8626,8 +8426,6 @@ deepseek-v3.2 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 2元 3元 @@ -8636,8 +8434,6 @@ deepseek-v3.2 deepseek-v3.2-exp -中国内地 - 2元 3元 @@ -8646,8 +8442,6 @@ deepseek-v3.2-exp deepseek-v3.1 -中国内地 - 4元 12元 @@ -8658,8 +8452,6 @@ deepseek-r1 > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 4元 16元 @@ -8668,8 +8460,6 @@ deepseek-r1 deepseek-r1-0528 -中国内地 - 4元 16元 @@ -8680,8 +8470,6 @@ deepseek-v3 > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 2元 8元 @@ -8690,14 +8478,10 @@ deepseek-v3 deepseek-r1-distill-qwen-1.5b -中国内地 - 限时免费 deepseek-r1-distill-qwen-7b -中国内地 - 0.5元 1元 @@ -8706,8 +8490,6 @@ deepseek-r1-distill-qwen-7b deepseek-r1-distill-qwen-14b -中国内地 - 1元 3元 @@ -8716,8 +8498,6 @@ deepseek-r1-distill-qwen-14b deepseek-r1-distill-qwen-32b -中国内地 - 2元 6元 @@ -8726,16 +8506,12 @@ deepseek-r1-distill-qwen-32b deepseek-r1-distill-llama-8b -中国内地 - 已下线 > 该模型已下线，推荐使用[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)、[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api)、[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api)作为替代模型。 deepseek-r1-distill-llama-70b -中国内地 - 目前仅供免费体验 > 免费额度用完后不可调用，推荐使用[深度思考](https://help.aliyun.com/zh/model-studio/deep-thinking)、[DeepSeek-阿里云](https://help.aliyun.com/zh/model-studio/deepseek-api)、[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api)作为替代模型 @@ -8926,8 +8702,6 @@ deepseek-v4-flash **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** @@ -8938,8 +8712,6 @@ deepseek-v4-flash siliconflow/deepseek-v3.2 -中国内地 - 2元 3元 @@ -8948,24 +8720,18 @@ siliconflow/deepseek-v3.2 siliconflow/deepseek-v3.1-terminus -中国内地 - 4元 12元 siliconflow/deepseek-r1-0528 -中国内地 - 4元 16元 siliconflow/deepseek-v3-0324 -中国内地 - 2元 8元 @@ -8976,8 +8742,6 @@ siliconflow/deepseek-v3-0324 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** @@ -8990,8 +8754,6 @@ vanchin/deepseek-v3.2-think > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 2元 3元 @@ -9002,8 +8764,6 @@ vanchin/deepseek-v3.1-terminus > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 4元 12元 @@ -9012,8 +8772,6 @@ vanchin/deepseek-r1 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 4元 16元 @@ -9022,16 +8780,12 @@ vanchin/deepseek-v3 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 2元 8元 vanchin/deepseek-ocr -中国内地 - 0.216元 0.216元 @@ -9042,14 +8796,12 @@ vanchin/deepseek-ocr **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **模式** **输入单价（每百万Token）** @@ -9058,12 +8810,10 @@ vanchin/deepseek-ocr **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） kimi-k2.7-code -中国内地 - 仅思考模式 6.5元 @@ -9074,8 +8824,6 @@ kimi-k2.7-code kimi-k2.6 -中国内地 - 非思考和思考模式 6.5元 @@ -9086,8 +8834,6 @@ kimi-k2.6 kimi-k2.5 -中国内地 - 非思考和思考模式 4元 @@ -9098,8 +8844,6 @@ kimi-k2.5 kimi-k2-thinking -中国内地 - 仅思考模式 4元 @@ -9110,8 +8854,6 @@ kimi-k2-thinking Moonshot-Kimi-K2-Instruct -中国内地 - 非思考模式 4元 @@ -9236,8 +8978,6 @@ kimi-k2.7-code **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** @@ -9246,24 +8986,28 @@ kimi-k2.7-code **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -kimi/kimi-k2.7-code-highspeed +kimi/kimi-k3 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 +20元 + +100元 + +无 + +kimi/kimi-k2.7-code-highspeed + +> [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 13元 54元 -无 - kimi/kimi-k2.7-code > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 6.5元 27元 @@ -9272,8 +9016,6 @@ kimi/kimi-k2.6 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 6.5元 27元 @@ -9282,8 +9024,6 @@ kimi/kimi-k2.5 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 4元 21元 @@ -9294,14 +9034,12 @@ kimi/kimi-k2.5 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **模式** **单次请求的输入Token数** @@ -9314,12 +9052,10 @@ kimi/kimi-k2.5 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） glm-5.2 -中国内地 - 非思考和思考模式 不区分阶梯 @@ -9332,8 +9068,6 @@ glm-5.2 glm-5.2-fast-preview -中国内地 - 非思考和思考模式 不区分阶梯 @@ -9346,8 +9080,6 @@ glm-5.2-fast-preview glm-5.1 -中国内地 - 非思考和思考模式 0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 非思考和思考模式 4.2元 @@ -9804,8 +9508,6 @@ MiniMax/MiniMax-M2.7 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 仅思考模式 2.1元 @@ -9816,8 +9518,6 @@ MiniMax/MiniMax-M2.5 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 仅思考模式 2.1元 @@ -9828,8 +9528,6 @@ MiniMax/MiniMax-M2.1 > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 -中国内地 - 仅思考模式 2.1元 @@ -9844,8 +9542,6 @@ MiniMax/MiniMax-M2.1 **模型 ID（Model ID）** -**服务部署范围** - **输入Token数量** **输入单价（每百万Token）** @@ -9858,8 +9554,6 @@ MiniMax/MiniMax-M2.1 xiaomi/mimo-v2.5-pro -中国内地 - 0 当前能力等同于qwen-image-2.0-pro-2026-04-22 -中国内地 - 0.5元/张 100张 qwen-image-2.0-pro-2026-06-22 -中国内地 - 0.5元/张 100张 qwen-image-2.0-pro-2026-04-22 -中国内地 - 0.5元/张 100张 qwen-image-2.0-pro-2026-03-03 -中国内地 - 0.5元/张 100张 @@ -9980,16 +9664,12 @@ qwen-image-2.0 > 当前能力等同于qwen-image-2.0-2026-03-03 -中国内地 - 0.2元/张 100张 qwen-image-2.0-2026-03-03 -中国内地 - 0.2元/张 100张 @@ -9998,16 +9678,12 @@ qwen-image-max > 当前能力等同于qwen-image-max-2025-12-30 -中国内地 - 0.5元/张 100张 qwen-image-max-2025-12-30 -中国内地 - 0.5元/张 100张 @@ -10016,24 +9692,18 @@ qwen-image-plus > 当前能力等同于qwen-image -中国内地 - 0.2元/张 100张 qwen-image-plus-2026-01-09 -中国内地 - 0.2元/张 100张 qwen-image -中国内地 - 0.25元/张 100张 @@ -10046,6 +9716,12 @@ qwen-image **输出单价** +qwen-image-3.0-pro + +国际 + +限时免费 + qwen-image-2.0-pro > 当前能力等同于qwen-image-2.0-pro-2026-04-22 @@ -10126,50 +9802,44 @@ qwen-image **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） + +qwen-image-3.0-pro + +限时免费 qwen-image-2.0-pro > 当前能力等同于qwen-image-2.0-pro-2026-04-22 -中国内地 - 0.5元/张 100张 qwen-image-2.0-pro-2026-06-22 -中国内地 - 0.5元/张 100张 qwen-image-2.0-pro-2026-04-22 -中国内地 - 0.5元/张 100张 qwen-image-2.0-pro-2026-03-03 -中国内地 - 0.5元/张 100张 @@ -10178,16 +9848,12 @@ qwen-image-2.0 > 当前能力等同于qwen-image-2.0-2026-03-03 -中国内地 - 0.2元/张 100张 qwen-image-2.0-2026-03-03 -中国内地 - 0.2元/张 100张 @@ -10196,16 +9862,12 @@ qwen-image-edit-max > 当前能力等同于qwen-image-edit-max-2026-01-16 -中国内地 - 0.5元/张 100张 qwen-image-edit-max-2026-01-16 -中国内地 - 0.5元/张 100张 @@ -10214,32 +9876,24 @@ qwen-image-edit-plus > 当前能力等同于qwen-image-edit-plus-2025-10-30 -中国内地 - 0.2元/张 100张 qwen-image-edit-plus-2025-12-15 -中国内地 - 0.2元/张 100张 qwen-image-edit-plus-2025-10-30 -中国内地 - 0.2元/张 100张 qwen-image-edit -中国内地 - 0.3元/张 100张 @@ -10252,6 +9906,12 @@ qwen-image-edit **输出单价** +qwen-image-3.0-pro + +国际 + +限时免费 + qwen-image-2.0-pro > 当前能力等同于qwen-image-2.0-pro-2026-04-22 @@ -10340,18 +10000,14 @@ qwen-image-edit **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-mt-image -中国内地 - 0.003元/张 100张 @@ -10362,24 +10018,20 @@ qwen-mt-image **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） z-image-turbo -中国内地 - 关闭提示词改写（`prompt_extend=false`）：0.1元/张 开启提示词改写（`prompt_extend=true`）：0.2元/张 @@ -10408,80 +10060,62 @@ z-image-turbo **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.6-t2i -中国内地 - 0.20元/张 50张 wan2.5-t2i-preview -中国内地 - 0.20元/张 50张 wan2.2-t2i-plus -中国内地 - 0.20元/张 100张 wan2.2-t2i-flash -中国内地 - 0.14元/张 100张 wanx2.1-t2i-plus -中国内地 - 0.20元/张 500张 wanx2.1-t2i-turbo -中国内地 - 0.14元/张 500张 wanx2.0-t2i-turbo -中国内地 - 0.04元/张 500张 wanx-v1 -中国内地 - 0.16元/张 500张 @@ -10564,40 +10198,32 @@ wan2.6-t2i **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-image-pro -中国内地 - 0.50元/张 50张 wan2.7-image -中国内地 - 0.20元/张 50张 wan2.6-image -中国内地 - 0.20元/张 50张 @@ -10662,32 +10288,26 @@ wan2.6-image **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.5-i2i-preview -中国内地 - 0.20元/张 50张 wanx2.1-imageedit -中国内地 - 0.14元/张 500张 @@ -10714,18 +10334,14 @@ wan2.5-i2i-preview **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-sketch-to-image-lite -中国内地 - 0.06元/张 500张 @@ -10738,18 +10354,14 @@ wanx-sketch-to-image-lite **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-x-painting -中国内地 - 目前仅供免费体验。 > 免费额度用完后不可调用，推荐参考[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide)或[图像编辑-万相2.1](https://help.aliyun.com/zh/model-studio/wanx-image-edit)获取替代方案。 @@ -10764,18 +10376,14 @@ wanx-x-painting **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-style-repaint-v1 -中国内地 - 0.12元/张 500张 @@ -10788,18 +10396,14 @@ wanx-style-repaint-v1 **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-background-generation-v2 -中国内地 - 0.08元/张 500张 @@ -10812,18 +10416,14 @@ wanx-background-generation-v2 **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） image-out-painting -中国内地 - 0.18元/张 500张 @@ -10836,18 +10436,14 @@ image-out-painting **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） image-instance-segmentation -中国内地 - 目前仅供免费体验。 > 免费额度用完后不可调用。 @@ -10862,18 +10458,14 @@ image-instance-segmentation **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） image-erase-completion -中国内地 - 目前仅供免费体验。 > 免费额度用完后不可调用，推荐参考[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide)或[图像编辑-万相2.1](https://help.aliyun.com/zh/model-studio/wanx-image-edit)获取替代方案。 @@ -10888,18 +10480,14 @@ image-erase-completion **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-virtualmodel -中国内地 - 目前仅供免费体验。 > 免费额度用完后不可调用，推荐参考[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide)或[图像编辑-万相2.1](https://help.aliyun.com/zh/model-studio/wanx-image-edit)获取替代方案。 @@ -10908,8 +10496,6 @@ wanx-virtualmodel virtualmodel-v2 -中国内地 - ### **鞋靴模特** > 仅输出计费，计费规则请参见[图像生成](#26310bc5cf4do)。 @@ -10918,18 +10504,14 @@ virtualmodel-v2 **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） shoemodel-v1 -中国内地 - 目前仅供免费体验。 > 免费额度用完后不可调用。 @@ -10944,18 +10526,14 @@ shoemodel-v1 **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-poster-generation-v1 -中国内地 - 目前仅供免费体验。 > 免费额度用完后不可调用，推荐参考[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide)或[图像编辑-万相2.1](https://help.aliyun.com/zh/model-studio/wanx-image-edit)获取替代方案。 @@ -10975,39 +10553,31 @@ wanx-poster-generation-v1 **模型 ID（Model ID）** -**服务部署范围** - **单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) facechain-facedetect -中国内地 - 限时免费 限时免费 facechain-finetune -中国内地 - 2.5元/次 50次 -有效期：申请通过后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） facechain-generation -中国内地 - 0.18元/张 500张 -有效期：申请通过后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） ### **创意文字生成-WordArt锦书** @@ -11017,26 +10587,20 @@ facechain-generation **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wordart-texture -中国内地 - 0.08元/张 500张 wordart-semantic -中国内地 - 0.24元/张 ### **AI试衣-OutfitAnyone** @@ -11054,42 +10618,30 @@ wordart-semantic **模型 ID（Model ID）** -**服务部署范围** - **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） aitryon -中国内地 - 400张 aitryon-plus -中国内地 - 400张 aitryon-parsing-v1 -中国内地 - 400张 aitryon-refiner -中国内地 - 100张 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **单价** **折扣** @@ -11098,8 +10650,6 @@ aitryon-refiner aitryon -中国内地 - 0.20元/张 无 @@ -11108,8 +10658,6 @@ aitryon aitryon-plus -中国内地 - 0.50元/张 无 @@ -11118,8 +10666,6 @@ aitryon-plus aitryon-parsing-v1 -中国内地 - 0.004元/张 无 @@ -11128,8 +10674,6 @@ aitryon-parsing-v1 aitryon-refiner -中国内地 - 0.30元/张 无 @@ -11182,8 +10726,6 @@ aitryon-refiner **模型 ID（Model ID）** -**服务部署范围** - **输出图像分辨率** **输出单价** @@ -11192,8 +10734,6 @@ aitryon-refiner kling/kling-v3-image-generation -中国内地 - 1K 0.2元/张 @@ -11206,8 +10746,6 @@ kling/kling-v3-image-generation kling/kling-v3-omni-image-generation -中国内地 - 1K 0.2元/张 @@ -11228,8 +10766,6 @@ kling/kling-v3-omni-image-generation **模型 ID（Model ID）** -**服务部署范围** - **输出图像分辨率** **输出单价** @@ -11238,8 +10774,6 @@ kling/kling-v3-omni-image-generation vidu/vidu-image\_reference2image -中国内地 - 1K 0.625元/张 @@ -11256,8 +10790,6 @@ vidu/vidu-image\_reference2image vidu/viduq3-fast\_reference2image -中国内地 - 1K 0.46875元/张 @@ -11272,8 +10804,6 @@ vidu/viduq3-fast\_reference2image vidu/viduq2-pro\_reference2image -中国内地 - 1K 0.9375元/张 @@ -11288,8 +10818,6 @@ vidu/viduq2-pro\_reference2image vidu/viduq2-fast\_reference2image -中国内地 - 1K 0.28125元/张 @@ -11302,26 +10830,20 @@ vidu/viduq2-fast\_reference2image **模型 ID（Model ID）** -**服务部署范围** - **输出单价（每秒）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） fun-music-preview -中国内地 - 0.005元 1,000秒 fun-music-v1 -中国内地 - 0.002元 ## **语音合成（文本转语音）** @@ -11332,32 +10854,26 @@ fun-music-v1 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-audio-3.0-tts-plus -中国内地 - 1.4元 1万字符 qwen-audio-3.0-tts-flash -中国内地 - 1元 1万字符 @@ -11386,7 +10902,7 @@ qwen-audio-3.0-tts-flash **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） @@ -11396,22 +10912,18 @@ qwen-audio-3.0-tts-flash **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-instruct-flash > 当前能力等同于qwen3-tts-instruct-flash-2026-01-26 -中国内地 - 0.8元 不计费 @@ -11420,8 +10932,6 @@ qwen3-tts-instruct-flash qwen3-tts-instruct-flash-2026-01-26 -中国内地 - 0.8元 不计费 @@ -11434,20 +10944,16 @@ qwen3-tts-instruct-flash-2026-01-26 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-vd-2026-01-26 -中国内地 - 0.8元 不计费 @@ -11460,20 +10966,16 @@ qwen3-tts-vd-2026-01-26 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-vc-2026-01-22 -中国内地 - 0.8元 不计费 @@ -11486,22 +10988,18 @@ qwen3-tts-vc-2026-01-22 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-flash > 当前能力等同于qwen3-tts-flash-2025-11-27 -中国内地 - 0.8元 不计费 @@ -11510,8 +11008,6 @@ qwen3-tts-flash qwen3-tts-flash-2025-11-27 -中国内地 - 0.8元 不计费 @@ -11520,8 +11016,6 @@ qwen3-tts-flash-2025-11-27 qwen3-tts-flash-2025-09-18 -中国内地 - 0.8元 不计费 @@ -11534,20 +11028,16 @@ qwen3-tts-flash-2025-09-18 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-tts-flash -中国内地 - 1.6元 10元 @@ -11556,8 +11046,6 @@ qwen-tts-flash qwen-tts-latest -中国内地 - 1.6元 10元 @@ -11566,8 +11054,6 @@ qwen-tts-latest qwen-tts-2025-05-22 -中国内地 - 1.6元 10元 @@ -11576,8 +11062,6 @@ qwen-tts-2025-05-22 qwen-tts-2025-04-10 -中国内地 - 1.6元 10元 @@ -11676,7 +11160,7 @@ qwen3-tts-flash-2025-09-18 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） @@ -11686,22 +11170,18 @@ qwen3-tts-flash-2025-09-18 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-instruct-flash-realtime > 当前能力等同于qwen3-tts-instruct-flash-realtime-2026-01-22 -中国内地 - 1元 不计费 @@ -11710,8 +11190,6 @@ qwen3-tts-instruct-flash-realtime qwen3-tts-instruct-flash-realtime-2026-01-22 -中国内地 - 1元 不计费 @@ -11724,20 +11202,16 @@ qwen3-tts-instruct-flash-realtime-2026-01-22 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-vd-realtime-2026-01-15 -中国内地 - 1元 不计费 @@ -11746,8 +11220,6 @@ qwen3-tts-vd-realtime-2026-01-15 qwen3-tts-vd-realtime-2025-12-16 -中国内地 - 1元 不计费 @@ -11760,20 +11232,16 @@ qwen3-tts-vd-realtime-2025-12-16 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-vc-realtime-2026-01-15 -中国内地 - 1元 不计费 @@ -11782,8 +11250,6 @@ qwen3-tts-vc-realtime-2026-01-15 qwen3-tts-vc-realtime-2025-11-27 -中国内地 - 1万字符 #### 千问3-TTS-Flash-Realtime @@ -11792,20 +11258,16 @@ qwen3-tts-vc-realtime-2025-11-27 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-flash-realtime -中国内地 - 1元 不计费 @@ -11814,8 +11276,6 @@ qwen3-tts-flash-realtime qwen3-tts-flash-realtime-2025-11-27 -中国内地 - 1元 不计费 @@ -11824,8 +11284,6 @@ qwen3-tts-flash-realtime-2025-11-27 qwen3-tts-flash-realtime-2025-09-18 -中国内地 - 1元 不计费 @@ -11838,20 +11296,16 @@ qwen3-tts-flash-realtime-2025-09-18 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-tts-realtime -中国内地 - 2.4元 12元 @@ -11860,8 +11314,6 @@ qwen-tts-realtime qwen-tts-realtime-latest -中国内地 - 2.4元 12元 @@ -11870,8 +11322,6 @@ qwen-tts-realtime-latest qwen-tts-realtime-2025-07-15 -中国内地 - 2.4元 12元 @@ -11982,24 +11432,20 @@ qwen3-tts-flash-realtime-2025-09-18 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **单价（每个音色）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-voice-enrollment -中国内地 - 0.01元 1000个音色/账号 @@ -12024,24 +11470,20 @@ qwen-voice-enrollment **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **单价（每个音色）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-voice-design -中国内地 - 0.2元 10个音色/账号 @@ -12064,7 +11506,7 @@ qwen-voice-design **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） @@ -12072,58 +11514,44 @@ qwen-voice-design **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） cosyvoice-v3.5-plus -中国内地 - 1.5元 1万字符 cosyvoice-v3.5-flash -中国内地 - 0.8元 1万字符 cosyvoice-v3-plus -中国内地 - 2元 1万字符 cosyvoice-v3-flash -中国内地 - 1元 1万字符 cosyvoice-v2 -中国内地 - 2元 1万字符 cosyvoice-v1 -中国内地 - 2元 1万字符 @@ -12158,16 +11586,12 @@ cosyvoice-v3-flash **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每万字符）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) 参见[模型列表](https://help.aliyun.com/zh/model-studio/sambert-java-sdk#57d33631f7doi) -中国内地 - 1元 每主账号每模型每月3万字符。 @@ -12180,8 +11604,6 @@ cosyvoice-v3-flash **模型名称** -**服务部署范围** - **语音合成单价（每万字符）** [复刻一个音色](https://help.aliyun.com/zh/model-studio/mini-clone-api) @@ -12190,8 +11612,6 @@ cosyvoice-v3-flash MiniMax/speech-2.8-hd -中国内地 - 3.5元 9.9元 @@ -12202,20 +11622,14 @@ MiniMax/speech-2.8-hd MiniMax/speech-02-hd -中国内地 - 3.5元 MiniMax/speech-2.8-turbo -中国内地 - 2元 MiniMax/speech-02-turbo -中国内地 - 2元 ## **语音识别（语音转文本）与翻译（语音转成指定语种的文本）** @@ -12226,21 +11640,19 @@ MiniMax/speech-02-turbo **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **输入：音频** @@ -12252,8 +11664,6 @@ MiniMax/speech-02-turbo qwen3.5-livetranslate-flash-realtime -中国内地 - 40元 3.3元 @@ -12266,8 +11676,6 @@ qwen3.5-livetranslate-flash-realtime qwen3.5-livetranslate-flash-realtime-2026-05-19 -中国内地 - 40元 3.3元 @@ -12282,8 +11690,6 @@ qwen3-livetranslate-flash-realtime > 当前能力等同于qwen3-livetranslate-flash-realtime-2025-09-22 -中国内地 - 64元 8元 @@ -12296,8 +11702,6 @@ qwen3-livetranslate-flash-realtime qwen3-livetranslate-flash-realtime-2025-09-22 -中国内地 - 64元 8元 @@ -12382,21 +11786,19 @@ qwen3-livetranslate-flash-realtime-2025-09-22 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **输入：音频** @@ -12408,8 +11810,6 @@ qwen3-livetranslate-flash-realtime-2025-09-22 qwen3-livetranslate-flash -中国内地 - 10元 4元 @@ -12422,8 +11822,6 @@ qwen3-livetranslate-flash qwen3-livetranslate-flash-2025-12-01 -中国内地 - 10元 4元 @@ -12480,7 +11878,7 @@ qwen3-livetranslate-flash-2025-12-01 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） @@ -12488,20 +11886,16 @@ qwen3-livetranslate-flash-2025-12-01 **模型 ID（Model ID）** -**服务部署范围** - **输入单价** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-asr-flash-filetrans -中国内地 - 0.00022元/秒 不计费 @@ -12510,28 +11904,20 @@ qwen3-asr-flash-filetrans qwen3-asr-flash-filetrans-2025-11-17 -中国内地 - 36,000秒（10小时） qwen3-asr-flash > 当前能力等同于qwen3-asr-flash-2025-09-08 -中国内地 - 36,000秒（10小时） qwen3-asr-flash-2026-02-10 -中国内地 - 36,000秒（10小时） qwen3-asr-flash-2025-09-08 -中国内地 - 36,000秒（10小时） #### 美国（弗吉尼亚） @@ -12612,40 +11998,32 @@ qwen3-asr-flash-2025-09-08 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-asr-flash-realtime > 当前能力等同于qwen3-asr-flash-realtime-2025-10-27 -中国内地 - 0.00033元/秒 36,000秒（10小时） qwen3-asr-flash-realtime-2026-02-10 -中国内地 - 36,000秒（10小时） qwen3-asr-flash-realtime-2025-10-27 -中国内地 - 36,000秒（10小时） #### 新加坡 @@ -12680,58 +12058,44 @@ qwen3-asr-flash-realtime-2025-10-27 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） fun-asr > 当前能力等同于fun-asr-2025-11-07 -中国内地 - 0.00022元/秒 36,000秒（10小时） fun-asr-2025-11-07 -中国内地 - 36,000秒（10小时） fun-asr-2025-08-25 -中国内地 - 36,000秒（10小时） fun-asr-mtl -中国内地 - 36,000秒（10小时） fun-asr-mtl-2025-08-25 -中国内地 - 36,000秒（10小时） fun-asr-flash-2026-06-15 -中国内地 - 0.00022元/秒 36,000秒（10小时） @@ -12782,66 +12146,48 @@ fun-asr-flash-2026-06-15 **模型 ID（Model ID）** -**服务部署范围** - **输入单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） fun-asr-realtime -中国内地 - 0.00033元/秒 36,000秒（10小时） fun-asr-realtime-2026-02-28 -中国内地 - 36,000秒（10小时） fun-asr-realtime-2025-11-07 -中国内地 - 36,000秒（10小时） fun-asr-realtime-2025-09-15 -中国内地 - 36,000秒（10小时） fun-asr-mtl-realtime -中国内地 - 36,000秒（10小时） fun-asr-mtl-realtime-2025-12-10 -中国内地 - 36,000秒（10小时） fun-asr-flash-8k-realtime > 当前能力等同于fun-asr-flash-8k-realtime-2026-01-28 -中国内地 - 0.00022元/秒 36,000秒（10小时） fun-asr-flash-8k-realtime-2026-01-28 -中国内地 - 36,000秒（10小时） #### 新加坡 @@ -12872,16 +12218,12 @@ fun-asr-realtime-2025-11-07 **模型 ID（Model ID）** -**服务部署范围** - **输入单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) paraformer-v2 -中国内地 - 0.00008元/秒 36,000秒（10小时） @@ -12892,20 +12234,12 @@ paraformer-v2 paraformer-8k-v2 -中国内地 - paraformer-v1 -中国内地 - paraformer-8k-v1 -中国内地 - paraformer-mtl-v1 -中国内地 - #### **实时语音识别** 计费规则：按输入音频的秒数计费，输出不计费。 @@ -12914,16 +12248,12 @@ paraformer-mtl-v1 **模型 ID（Model ID）** -**服务部署范围** - **输入单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) paraformer-realtime-v2 -中国内地 - 0.00024元/秒 36,000秒（10小时） @@ -12934,16 +12264,10 @@ paraformer-realtime-v2 paraformer-realtime-v1 -中国内地 - paraformer-realtime-8k-v2 -中国内地 - paraformer-realtime-8k-v1 -中国内地 - ## **语音对话** ### **实时语音对话** @@ -12965,21 +12289,19 @@ paraformer-realtime-8k-v1 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **文本** @@ -12991,8 +12313,6 @@ paraformer-realtime-8k-v1 qwen-audio-3.0-realtime-plus -中国内地 - 5元 40元 @@ -13005,8 +12325,6 @@ qwen-audio-3.0-realtime-plus qwen-audio-3.0-realtime-flash -中国内地 - 3元 30元 @@ -13042,26 +12360,22 @@ qwen-audio-3.0-realtime-flash **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） happyhorse-1.1-t2v -中国内地 - 720P 原价0.9元/秒 限时6折 @@ -13074,8 +12388,6 @@ happyhorse-1.1-t2v happyhorse-1.0-t2v -中国内地 - 720P 原价0.9元/秒 限时8折 @@ -13188,32 +12500,50 @@ happyhorse-1.0-t2v 原价1.6元/秒 限时8折 +#### 日本（东京） + +**模型 ID（Model ID）** + +**服务部署范围** + +**输出视频分辨率** + +**输出单价** + +happyhorse-1.1-t2v + +全球 + +720P + +原价0.9元/秒 限时6折 + +1080P + +原价1.2元/秒 限时6折 + ### **HappyHorse-图生视频-基于首帧** > 仅输出计费，计费规则请参见[视频生成](#d809366847gza)。 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） happyhorse-1.1-i2v -中国内地 - 720P 原价0.9元/秒 限时6折 @@ -13226,8 +12556,6 @@ happyhorse-1.1-i2v happyhorse-1.0-i2v -中国内地 - 720P 原价0.9元/秒 限时8折 @@ -13340,32 +12668,50 @@ happyhorse-1.0-i2v 原价1.6元/秒 限时8折 +#### 日本（东京） + +**模型 ID（Model ID）** + +**服务部署范围** + +**输出视频分辨率** + +**输出单价** + +happyhorse-1.1-i2v + +全球 + +720P + +原价0.9元/秒 限时6折 + +1080P + +原价1.2元/秒 限时6折 + ### **HappyHorse-参考生视频** > 仅输出计费，计费规则请参见[视频生成](#d809366847gza)。 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） happyhorse-1.1-r2v -中国内地 - 720P 原价0.9元/秒 限时6折 @@ -13378,8 +12724,6 @@ happyhorse-1.1-r2v happyhorse-1.0-r2v -中国内地 - 720P 原价0.9元/秒 限时8折 @@ -13492,11 +12836,33 @@ happyhorse-1.0-r2v 原价1.6元/秒 限时8折 +#### 日本（东京） + +**模型 ID（Model ID）** + +**服务部署范围** + +**输出视频分辨率** + +**输出单价** + +happyhorse-1.1-r2v + +全球 + +720P + +原价0.9元/秒 限时6折 + +1080P + +原价1.2元/秒 限时6折 + ### **HappyHorse-视频编辑** **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） @@ -13504,20 +12870,16 @@ happyhorse-1.0-r2v **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输入和输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） happyhorse-1.0-video-edit -中国内地 - 720P 原价0.9元/秒 限时8折 @@ -13600,32 +12962,52 @@ happyhorse-1.0-video-edit 原价1.6元/秒 限时8折 +#### 日本（东京） + +计费规则：输入视频和输出视频均计费，按**视频秒数**计费，失败不计费也不占用免费额度。 + +**模型 ID（Model ID）** + +**服务部署范围** + +**输出视频分辨率** + +**输入和输出单价** + +happyhorse-1.0-video-edit + +全球 + +720P + +原价0.9元/秒 限时8折 + +1080P + +原价1.6元/秒 限时8折 + ### **万相-文生视频** > 仅输出计费，计费规则请参见[视频生成](#d809366847gza)。 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-t2v-2026-06-12 -中国内地 - 720P 0.6元/秒 @@ -13638,8 +13020,6 @@ wan2.7-t2v-2026-06-12 wan2.7-t2v-2026-04-25 -中国内地 - 720P 0.6元/秒 @@ -13652,8 +13032,6 @@ wan2.7-t2v-2026-04-25 wan2.7-t2v -中国内地 - 720P 0.6元/秒 @@ -13666,8 +13044,6 @@ wan2.7-t2v wan2.6-t2v -中国内地 - 720P 0.6元/秒 @@ -13680,8 +13056,6 @@ wan2.6-t2v wan2.5-t2v-preview -中国内地 - 480P 0.3元/秒 @@ -13698,8 +13072,6 @@ wan2.5-t2v-preview wan2.2-t2v-plus -中国内地 - 480P 0.14元/秒 @@ -13712,8 +13084,6 @@ wan2.2-t2v-plus wanx2.1-t2v-turbo -中国内地 - 480P 0.24元/秒 @@ -13726,8 +13096,6 @@ wanx2.1-t2v-turbo wanx2.1-t2v-plus -中国内地 - 720P 0.70元/秒 @@ -13902,14 +13270,12 @@ wan2.6-t2v **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频类型** **输出视频分辨率** @@ -13918,12 +13284,10 @@ wan2.6-t2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-i2v-2026-04-25 -中国内地 - 有声视频 720P @@ -13938,8 +13302,6 @@ wan2.7-i2v-2026-04-25 wan2.7-i2v -中国内地 - 有声视频 720P @@ -13998,14 +13360,12 @@ wan2.7-i2v **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频类型** **输出视频分辨率** @@ -14014,12 +13374,10 @@ wan2.7-i2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.6-i2v-flash -中国内地 - 有声视频 `audio=true` @@ -14048,8 +13406,6 @@ wan2.6-i2v-flash wan2.6-i2v -中国内地 - 有声视频 720P @@ -14064,8 +13420,6 @@ wan2.6-i2v wan2.5-i2v-preview -中国内地 - 有声视频 480P @@ -14084,8 +13438,6 @@ wan2.5-i2v-preview wan2.2-i2v-flash -中国内地 - 无声视频 480P @@ -14104,8 +13456,6 @@ wan2.2-i2v-flash wan2.2-i2v-plus -中国内地 - 无声视频 480P @@ -14120,8 +13470,6 @@ wan2.2-i2v-plus wanx2.1-i2v-turbo -中国内地 - 无声视频 480P @@ -14136,8 +13484,6 @@ wanx2.1-i2v-turbo wanx2.1-i2v-plus -中国内地 - 无声视频 720P @@ -14342,26 +13688,22 @@ wan2.6-i2v **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.2-kf2v-flash -中国内地 - 480P 0.10元/秒 @@ -14378,8 +13720,6 @@ wan2.2-kf2v-flash wanx2.1-kf2v-plus -中国内地 - 720P 0.70元/秒 @@ -14419,8 +13759,6 @@ wan2.1-kf2v-plus **模型 ID（Model ID）** -**服务部署范围** - **输出视频类型** **输出视频分辨率** @@ -14429,12 +13767,10 @@ wan2.1-kf2v-plus **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-r2v-2026-06-12 -中国内地 - 有声视频 720P @@ -14449,8 +13785,6 @@ wan2.7-r2v-2026-06-12 wan2.7-r2v -中国内地 - 有声视频 720P @@ -14465,8 +13799,6 @@ wan2.7-r2v wan2.6-r2v-flash -中国内地 - 有声视频 `audio=true` @@ -14495,8 +13827,6 @@ wan2.6-r2v-flash wan2.6-r2v -中国内地 - 有声视频 720P @@ -14647,7 +13977,7 @@ wan2.6-r2v **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） @@ -14655,20 +13985,16 @@ wan2.6-r2v **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输入和输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-videoedit -中国内地 - 720P 0.6元/秒 @@ -14683,20 +14009,16 @@ wan2.7-videoedit **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx2.1-vace-plus -中国内地 - 720P 0.70元/秒 @@ -14756,26 +14078,20 @@ wan2.1-vace-plus **模型 ID（Model ID）** -**服务部署范围** - **单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.2-s2v-detect -中国内地 - 输入图像：0.004元/张 200张 wan2.2-s2v -中国内地 - 输出视频： - 480P：0.5元/秒 @@ -14791,33 +14107,29 @@ wan2.2-s2v **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频模式** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.2-animate-move -中国内地 - 标准模式`wan-std` 0.4元/秒 50秒 -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） 专业模式`wan-pro` @@ -14851,33 +14163,29 @@ wan2.2-animate-move **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输出视频模式** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.2-animate-mix -中国内地 - 标准模式`wan-std` 0.6元/秒 50秒 -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） 专业模式`wan-pro` @@ -14918,34 +14226,26 @@ wan2.2-animate-mix **模型 ID（Model ID）** -**服务部署范围** - **单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） animate-anyone-detect-gen2 -中国内地 - 输入图像：0.004元/张 200张 animate-anyone-template-gen2 -中国内地 - 输出视频：0.08元/秒 1800秒（30分钟） animate-anyone-gen2 -中国内地 - 输出视频：0.08元/秒 1800秒（30分钟） @@ -14961,26 +14261,20 @@ animate-anyone-gen2 **模型 ID（Model ID）** -**服务部署范围** - **单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） emo-detect-v1 -中国内地 - 输入图像：0.004元/张 200张 emo-v1 -中国内地 - 输出视频： - 1:1画幅视频：0.08元/秒 @@ -15001,26 +14295,20 @@ emo-v1 **模型 ID（Model ID）** -**服务部署范围** - **单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） liveportrait-detect -中国内地 - 输入图像：0.004元/张 200张 liveportrait -中国内地 - 输出视频：0.02元/秒 1800秒（30分钟） @@ -15036,26 +14324,20 @@ liveportrait **模型 ID（Model ID）** -**服务部署范围** - **单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） emoji-detect-v1 -中国内地 - 输入图像：0.004元/张 200张 emoji-v1 -中国内地 - 输出视频：0.08元/秒 1800秒（30分钟） @@ -15068,18 +14350,14 @@ emoji-v1 **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） videoretalk -中国内地 - 0.08元/秒 1800秒（30分钟） @@ -15092,20 +14370,16 @@ videoretalk **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） video-style-transform -中国内地 - 540P 0.2元/秒 @@ -15126,8 +14400,6 @@ video-style-transform **模型 ID（Model ID）** -**服务部署范围** - **输出视频类型** **输出视频分辨率** @@ -15138,8 +14410,6 @@ video-style-transform pixverse/pixverse-c1-t2v -中国内地 - 有声视频 `audio=true` @@ -15184,8 +14454,6 @@ pixverse/pixverse-c1-t2v pixverse/pixverse-v6-t2v -中国内地 - 有声视频 `audio=true` @@ -15230,8 +14498,6 @@ pixverse/pixverse-v6-t2v pixverse/pixverse-v5.6-t2v -中国内地 - 有声视频 `audio=true` @@ -15276,8 +14542,6 @@ pixverse/pixverse-v5.6-t2v pixverse/pixverse-v5.6-it2v -中国内地 - 有声视频 `audio=true` @@ -15328,8 +14592,6 @@ pixverse/pixverse-v5.6-it2v **模型 ID（Model ID）** -**服务部署范围** - **输出视频类型** **输出视频分辨率** @@ -15340,8 +14602,6 @@ pixverse/pixverse-v5.6-it2v pixverse/pixverse-c1-it2v -中国内地 - 有声视频 `audio=true` @@ -15386,8 +14646,6 @@ pixverse/pixverse-c1-it2v pixverse/pixverse-v6-it2v -中国内地 - 有声视频 `audio=true` @@ -15432,8 +14690,6 @@ pixverse/pixverse-v6-it2v pixverse/pixverse-v5.6-it2v -中国内地 - 有声视频 `audio=true` @@ -15484,8 +14740,6 @@ pixverse/pixverse-v5.6-it2v **模型 ID（Model ID）** -**服务部署范围** - **输出视频类型** **输出视频分辨率** @@ -15496,8 +14750,6 @@ pixverse/pixverse-v5.6-it2v pixverse/pixverse-c1-kf2v -中国内地 - 有声视频 `audio=true` @@ -15542,8 +14794,6 @@ pixverse/pixverse-c1-kf2v pixverse/pixverse-v6-kf2v -中国内地 - 有声视频 `audio=true` @@ -15588,8 +14838,6 @@ pixverse/pixverse-v6-kf2v pixverse/pixverse-v5.6-kf2v -中国内地 - 有声视频 `audio=true` @@ -15640,8 +14888,6 @@ pixverse/pixverse-v5.6-kf2v **模型 ID（Model ID）** -**服务部署范围** - **输出视频类型** **输出视频分辨率** @@ -15652,8 +14898,6 @@ pixverse/pixverse-v5.6-kf2v pixverse/pixverse-c1-r2v -中国内地 - 有声视频 `audio=true` @@ -15698,8 +14942,6 @@ pixverse/pixverse-c1-r2v pixverse/pixverse-v5.6-r2v -中国内地 - 有声视频 `audio=true` @@ -15750,16 +14992,12 @@ pixverse/pixverse-v5.6-r2v **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[**（注）**](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) pixverse/pixverse-lipsync -中国内地 - 0.12元/秒 无免费额度 @@ -15772,8 +15010,6 @@ pixverse/pixverse-lipsync **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** @@ -15782,8 +15018,6 @@ pixverse/pixverse-lipsync pixverse/pixverse-motioncontrol -中国内地 - 360P 0.27元/秒 @@ -15806,16 +15040,12 @@ pixverse/pixverse-motioncontrol **模型 ID（Model ID）** -**服务部署范围** - **输出单价** **免费额度**[**（注）**](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) pixverse/pixverse-upscale -中国内地 - 0.15元/秒 无免费额度 @@ -15828,8 +15058,6 @@ pixverse/pixverse-upscale **模型 ID（Model ID）** -**服务部署范围** - **输出视频类型** **输出视频分辨率** @@ -15840,8 +15068,6 @@ pixverse/pixverse-upscale kling/kling-v3-video-generation -中国内地 - 无声视频 720P @@ -15866,8 +15092,6 @@ kling/kling-v3-video-generation kling/kling-v3-omni-video-generation -中国内地 - 无声视频（无参考视频） 720P @@ -15908,8 +15132,6 @@ kling/kling-v3-omni-video-generation **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** @@ -15918,8 +15140,6 @@ kling/kling-v3-omni-video-generation vidu/viduq3-pro\_text2video -中国内地 - 540P 0.3125元/秒 @@ -15936,8 +15156,6 @@ vidu/viduq3-pro\_text2video vidu/viduq3-turbo\_text2video -中国内地 - 540P 0.25元/秒 @@ -15954,8 +15172,6 @@ vidu/viduq3-turbo\_text2video vidu/viduq2\_text2video -中国内地 - 540P 0.1125元/秒 @@ -15978,8 +15194,6 @@ vidu/viduq2\_text2video **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** @@ -15988,8 +15202,6 @@ vidu/viduq2\_text2video vidu/viduq3-pro-fast\_img2video -中国内地 - 720P 0.375元/秒 @@ -16002,8 +15214,6 @@ vidu/viduq3-pro-fast\_img2video vidu/viduq3-pro\_img2video -中国内地 - 540P 0.3125元/秒 @@ -16020,8 +15230,6 @@ vidu/viduq3-pro\_img2video vidu/viduq3-turbo\_img2video -中国内地 - 540P 0.25元/秒 @@ -16038,8 +15246,6 @@ vidu/viduq3-turbo\_img2video vidu/viduq2-pro\_img2video -中国内地 - 540P 0.15625元/秒 @@ -16056,8 +15262,6 @@ vidu/viduq2-pro\_img2video vidu/viduq2-turbo\_img2video -中国内地 - 540P 0.0875元/秒 @@ -16074,8 +15278,6 @@ vidu/viduq2-turbo\_img2video vidu/viduq2-pro-fast\_img2video -中国内地 - 720P 0.1元/秒 @@ -16094,8 +15296,6 @@ vidu/viduq2-pro-fast\_img2video **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** @@ -16104,8 +15304,6 @@ vidu/viduq2-pro-fast\_img2video vidu/viduq3-pro\_start-end2video -中国内地 - 540P 0.3125元/秒 @@ -16122,8 +15320,6 @@ vidu/viduq3-pro\_start-end2video vidu/viduq3-turbo\_start-end2video -中国内地 - 540P 0.25元/秒 @@ -16140,8 +15336,6 @@ vidu/viduq3-turbo\_start-end2video vidu/viduq2-pro\_start-end2video -中国内地 - 540P 0.15625元/秒 @@ -16158,8 +15352,6 @@ vidu/viduq2-pro\_start-end2video vidu/viduq2-turbo\_start-end2video -中国内地 - 540P 0.0875元/秒 @@ -16182,8 +15374,6 @@ vidu/viduq2-turbo\_start-end2video **模型 ID（Model ID）** -**服务部署范围** - **输出视频分辨率** **输出单价** @@ -16192,8 +15382,6 @@ vidu/viduq2-turbo\_start-end2video vidu/viduq3-ad\_reference2video -中国内地 - 720P 0.75元/秒 @@ -16206,8 +15394,6 @@ vidu/viduq3-ad\_reference2video vidu/viduq3-drama\_reference2video -中国内地 - 1080P 0.875元/秒 @@ -16216,8 +15402,6 @@ vidu/viduq3-drama\_reference2video vidu/viduq3-mix\_reference2video -中国内地 - 720P 0.78125元/秒 @@ -16230,8 +15414,6 @@ vidu/viduq3-mix\_reference2video vidu/viduq3\_reference2video -中国内地 - 540P 0.3125元/秒 @@ -16248,8 +15430,6 @@ vidu/viduq3\_reference2video vidu/viduq3-turbo\_reference2video -中国内地 - 540P 0.15625元/秒 @@ -16266,8 +15446,6 @@ vidu/viduq3-turbo\_reference2video vidu/viduq2-pro\_reference2video -中国内地 - 540P 0.25元/秒 @@ -16284,8 +15462,6 @@ vidu/viduq2-pro\_reference2video vidu/viduq2\_reference2video -中国内地 - 540P 0.21875元/秒 @@ -16310,8 +15486,6 @@ vidu/viduq2\_reference2video **模型 ID（Model ID）** -**服务部署范围** - **3D任务类型** **输出规格** @@ -16320,8 +15494,6 @@ vidu/viduq2\_reference2video Tripo/Tripo-H3.1 -中国内地 - 文生3D 标准版+无贴图 @@ -16376,8 +15548,6 @@ Tripo/Tripo-H3.1 Tripo/Tripo-P1.0 -中国内地 - 文生3D 无贴图 @@ -16416,18 +15586,14 @@ Tripo/Tripo-P1.0 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3.7-text-embedding -中国内地 - 0.5元 100万Token @@ -16436,8 +15602,6 @@ text-embedding-v4 > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0.5元 100万Token @@ -16446,8 +15610,6 @@ text-embedding-v3 > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0.5元 50万Token @@ -16456,8 +15618,6 @@ text-embedding-v2 > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0.7元 50万Token @@ -16466,24 +15626,18 @@ text-embedding-v1 > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 -中国内地 - 0.7元 50万Token text-embedding-async-v2 -中国内地 - 0.7元 2000万Token text-embedding-async-v1 -中国内地 - 0.7元 2000万Token @@ -16516,13 +15670,11 @@ text-embedding-v3 **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **文本** @@ -16530,8 +15682,6 @@ text-embedding-v3 qwen3-vl-embedding -中国内地 - 0.7元 1.8元 @@ -16540,14 +15690,10 @@ qwen3-vl-embedding qwen2.5-vl-embedding -中国内地 - 100万Token tongyi-embedding-vision-plus -中国内地 - 0.5元 0.5元 @@ -16556,8 +15702,6 @@ tongyi-embedding-vision-plus tongyi-embedding-vision-flash -中国内地 - 0.15元 0.15元 @@ -16566,8 +15710,6 @@ tongyi-embedding-vision-flash multimodal-embedding-v1 -中国内地 - 0.7元 0.9元 @@ -16582,24 +15724,20 @@ multimodal-embedding-v1 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-vl-rerank -中国内地 - 文本输入：0.7元 图片输入：1.8元 @@ -16608,16 +15746,12 @@ qwen3-vl-rerank qwen3-rerank -中国内地 - 文本输入：0.5元 100万Token gte-rerank-v2 -中国内地 - 文本输入：0.8元 100万Token @@ -16646,8 +15780,6 @@ qwen3-rerank **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** @@ -16656,8 +15788,6 @@ qwen3-rerank farui-plus -中国内地 - 20元 20元 @@ -16672,20 +15802,16 @@ farui-plus **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） tongyi-intent-detect-v3 -中国内地 - 0.4元 1元 @@ -16698,28 +15824,24 @@ tongyi-intent-detect-v3 **说明** -以下模型仅在中国内地服务部署范围下有免费额度，其他服务部署范围下均无免费额度。 +以下模型仅在华北2（北京）下有免费额度，其他地域均无免费额度。 #### 华北2（北京） **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-plus-character > [Session Cache](https://help.aliyun.com/zh/model-studio/role-play#6034f997cde74)享有折扣 -中国内地 - 0.8元 2元 @@ -16730,8 +15852,6 @@ qwen-flash-character > [Session Cache](https://help.aliyun.com/zh/model-studio/role-play#6034f997cde74)享有折扣 -中国内地 - 0.25元 1.5元 @@ -16742,8 +15862,6 @@ qwen-flash-character-2026-02-26 > [Session Cache](https://help.aliyun.com/zh/model-studio/role-play#6034f997cde74)享有折扣 -中国内地 - 0.18元 1.5元 @@ -16850,20 +15968,16 @@ qwen-plus-character **模型 ID（Model ID）** -**服务部署范围** - **输入单价（每百万Token）** **输出单价（每百万Token）** **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） gui-plus -中国内地 - 1.5元 4.5元 @@ -16872,8 +15986,6 @@ gui-plus gui-plus-2026-02-26 -中国内地 - ## 错误码 如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-training-and-deployment-billing.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-training-and-deployment-billing.md index 8bc376aa..9f893367 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-training-and-deployment-billing.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-training-and-deployment-billing.md @@ -474,9 +474,9 @@ wan2.2-kf2v-flash - 后付费时，如果账户欠费，部署的资源将继续保留并计费 24 小时，在这 24 小时内服务仍可正常使用。超过 24 小时后系统停止计费，模型部署进入欠费状态，底层资源将被删除，但模型部署任务仍会保留。补足欠费后，系统将重新分配资源并恢复使用（恢复后继续产生费用）。如果您不希望继续产生费用，可删除模型部署任务，删除成功后将不再计费。 -当模型输入超过最长输入 Token 或 超出购买的 TPM 量时，相关调用将自动切换为当前模型的按量付费模式。此时，推理性能可能下降，将受业务空间中当前快照模型的公共流量的管控，[费用](https://help.aliyun.com/zh/model-studio/model-pricing)按模型调用（按量付费）标准计收。 +当模型输入超过最长输入 Token 时，相关调用将自动切换为当前模型的按量付费模式；超出购买的 TPM 量时，按创建时选择的溢出策略处理（「自动溢出」切换为按量付费，「仅使用 PTU 容量」返回 429）。此时，推理性能可能下降，将受业务空间中当前快照模型的公共流量的管控，[费用](https://help.aliyun.com/zh/model-studio/model-pricing)按模型调用（按量付费）标准计收。 -- 此时，调用 API 返回 Header 将包含：`x-dashscope-ptu-overflow:true`。 +- 此时（仅「自动溢出」策略下），调用 API 返回 Header 将包含：`x-dashscope-ptu-overflow:true`。 - TPM 统计请前往：[模型监控（北京）](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)。 @@ -820,15 +820,33 @@ MU2 x 8 ¥240,288 +MU3 x 8 + +¥1,096 + +¥527,752 + 千问3.6-35B-A3B qwen3.6-35b-a3b -MU8 x 1 +MU1 x 8 -¥47 +¥432 -¥22,400 +¥208,944 + +MU2 x 8 + +¥504 + +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 MU9 x 1 @@ -846,6 +864,12 @@ MU1 x 2 ¥52,236 +MU3 x 8 + +¥1,096 + +¥527,752 + 千问3.6-Plus-2026-04-02 qwen3.6-plus-2026-04-02 @@ -866,12 +890,6 @@ PD分离模式：¥417,888 qwen3.5-397b-a17b -MU2 x 8 - -¥504 - -¥240,288 - MU3 x 8 MU3 x 16（PD分离模式） @@ -884,6 +902,12 @@ PD分离模式：¥2,192 PD分离模式：¥1,055,504 +MU6 x 16 + +¥400 + +¥193,424 + 千问3.5-122B-A10B qwen3.5-122b-a10b @@ -894,11 +918,11 @@ MU1 x 4 ¥104,472 -MU2 x 8 +MU6 x 16 -¥504 +¥400 -¥240,288 +¥193,424 千问3.5-35B-A3B @@ -916,10 +940,40 @@ MU2 x 8 ¥240,288 +MU3 x 8 + +¥1,096 + +¥527,752 + +MU9 x 1 + +¥51 + +¥24,600 + 千问3.5-27B qwen3.5-27b +MU2 x 8 + +¥504 + +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 + +MU8 x 1 + +¥47 + +¥22,400 + MU9 x 1 ¥51 @@ -936,6 +990,12 @@ MU1 x 2 ¥52,236 +MU2 x 2 + +¥126 + +¥60,072 + MU8 x 1 ¥47 @@ -962,12 +1022,24 @@ MU1 x 2 qwen3.5-plus-2026-02-15 +MU1 x 8 + MU1 x 16（PD分离模式） +¥432 + PD分离模式：¥864 +¥208,944 + PD分离模式：¥417,888 +MU2 x 8 + +¥504 + +¥240,288 + MU3 x 8 MU3 x 16（PD分离模式） @@ -996,41 +1068,25 @@ MU2 x 8 ¥240,288 -千问3-Next-80B-A3B-Instruct - -qwen3-next-80b-a3b-instruct - -MU1 x 2 - -¥108 - -¥52,236 - 千问3-32B qwen3-32b -MU1 x 4 - -¥216 - -¥104,472 - -MU6 x 4 +MU6 x 16 -¥100 +¥400 -¥48,356 +¥193,424 -千问3-30B-A3B +千问3-30B-A3B-Thinking-2507 -qwen3-30b-a3b +qwen3-30b-a3b-thinking-2507 -MU9 x 2 +MU1 x 2 -¥102 +¥108 -¥49,200 +¥52,236 千问3-8B @@ -1080,12 +1136,6 @@ MU1 x 2 ¥52,236 -MU5 x 1 - -¥21 - -¥10,139 - 千问3-Embedding-0.6B qwen3-embedding-0.6b @@ -1154,6 +1204,16 @@ MU5 x 1 ¥10,139 +千问2.5-开源版-72B + +qwen2.5-72b-instruct + +MU1 x 8 + +¥432 + +¥208,944 + 千问2.5-开源版-32B qwen2.5-32b-instruct @@ -1190,26 +1250,6 @@ MU5 x 1 ¥10,139 -千问2.5-开源版-3B - -qwen2.5-3b-instruct - -MU5 x 1 - -¥21 - -¥10,139 - -千问-Flash-2025-07-28 - -qwen-flash-2025-07-28 - -MU1 x 4 - -¥216 - -¥104,472 - 千问-Plus-2025-07-28 qwen-plus-2025-07-28 @@ -1266,6 +1306,12 @@ GLM-5.1 glm-5.1 +MU2 x 8 + +¥504 + +¥240,288 + MU3 x 16（PD分离模式） PD分离模式：¥2,192 @@ -1298,6 +1344,16 @@ PD分离模式：¥800 PD分离模式：¥386,848 +GLM-4.7-Flash + +glm-4.7-flash + +MU3 x 16（PD分离模式） + +PD分离模式：¥2,192 + +PD分离模式：¥1,055,504 + ###### DeepSeek **模型名称** @@ -1318,11 +1374,11 @@ DeepSeek-v4-Flash deepseek-v4-flash -MU1 x 8 +MU3 x 8 -¥432 +¥1,096 -¥208,944 +¥527,752 DeepSeek-v3.2 @@ -1350,16 +1406,6 @@ PD分离模式：¥480,576 **最小计费：天** -MiniMax-M2.5 - -MiniMax-M2.5 - -MU1 x 16（PD分离模式） - -PD分离模式：¥864 - -PD分离模式：¥417,888 - Kimi-K2.5 kimi-k2.5 @@ -1402,15 +1448,21 @@ MU2 x 8 **最小计费：天** -千问3-VL-235B-A22B-Instruct +千问3-VL-32B-Instruct -qwen3-vl-235b-a22b-instruct +qwen3-vl-32b-instruct -MU1 x 4 +MU2 x 8 -¥216 +¥504 -¥104,472 +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 千问3-VL-8B-Instruct @@ -1422,6 +1474,12 @@ MU1 x 2 ¥52,236 +MU5 x 1 + +¥21 + +¥10,139 + 千问3-VL-4B-Instruct qwen3-vl-4b-instruct @@ -1442,6 +1500,16 @@ MU5 x 1 ¥10,139 +千问3-VL-Embedding-2B + +qwen3-vl-embedding-2b + +MU5 x 1 + +¥21 + +¥10,139 + 千问3-VL-Flash-2025-10-15 qwen3-vl-flash-2025-10-15 @@ -1472,16 +1540,6 @@ MU6 x 4 ¥48,356 -千问VL-OCR-2025-11-20 - -qwen-vl-ocr-2025-11-20 - -MU6 x 4 - -¥100 - -¥48,356 - ###### 千问 Omni **模型名称** @@ -1562,6 +1620,14 @@ MU5 **元/千Token** +千问3.5-27B 邀测中 + +qwen3.5-27b + +¥0.0006 + +¥0.0048 + 千问3-32B qwen3-32b @@ -1592,6 +1658,16 @@ qwen3-8b 思考模式：¥0.005 +千问3-4B-Instruct-2507 + +qwen3-4b-instruct-2507 + +¥0.0003 + +非思考模式：¥0.0012 + +思考模式：¥0.003 + 千问2.5-开源版-72B qwen2.5-72b-instruct @@ -1624,6 +1700,14 @@ qwen2.5-7b-instruct ¥0.001 +千问2-开源版-7B + +qwen2-7b-instruct + +¥0.001 + +¥0.002 + ##### 千问VL **基础模型** @@ -1670,6 +1754,14 @@ qwen2.5-vl-7b-instruct ¥0.005 +千问2.5-VL-3B-Instruct + +qwen2.5-vl-3b-instruct + +¥0.0012 + +¥0.0036 + ### **图像生成模型-万相** 经过SFT-LoRA高效微调的万相图像生成模型，部署免费，调用按微调的基础模型的标准调用价格计费。模型训练和部署流程请参见[图像生成模型调优](https://help.aliyun.com/zh/model-studio/wan-image-generation-finetune-guide)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/new-free-quota.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/new-free-quota.md index 6e121bcc..85b1cbde 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/new-free-quota.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/new-free-quota.md @@ -4,17 +4,17 @@ **说明** -仅华北2（北京）地域且服务部署范围为[中国内地](https://help.aliyun.com/zh/model-studio/regions/#080da663a75xh)的模型享有免费额度，其他地域和部署范围无免费额度。 +仅华北2（北京）地域模型享有免费额度，其他地域无免费额度。 ## 规则说明 ### 有效期 -免费额度的有效期为 30～90 天，从开通阿里云百炼或模型申请通过之日起计算。额度到期或耗尽后，继续调用模型推理服务将[产生计费](https://help.aliyun.com/zh/model-studio/billing-for-model-studio)。 +免费额度的有效期为 90 天，从开通阿里云百炼、模型发布或模型申请通过之日起计算（以较晚者为准）。额度到期或耗尽后，继续调用模型推理服务将[产生计费](https://help.aliyun.com/zh/model-studio/billing-for-model-studio)。 **重要** -自**2025年9月8日11点**起，首次开通阿里云百炼的用户，获赠的新人免费额度有效期调整为 90 天，在此之前已开通的用户不受影响，详情参考[阿里云百炼新人免费额度有效期调整通知](https://help.aliyun.com/zh/model-studio/new-free-quota-validity-adjustment)。 +**2025年9月8日11点**前已开通阿里云百炼的用户，免费额度有效期可能不足90天；在此之后开通的用户有效期为90天。详情参考[阿里云百炼新人免费额度有效期调整通知](https://help.aliyun.com/zh/model-studio/new-free-quota-validity-adjustment)。 免费额度过期后自动失效，不支持补发、延期或重置： @@ -55,7 +55,7 @@ ## 获取免费额度 -访问[阿里云百炼-中国内地版](https://bailian.console.aliyun.com/#/model-market)，阅读并同意协议后，系统将自动开通阿里云百炼并发放**免费推理额度**。 +访问[阿里云百炼-华北2（北京）地域](https://bailian.console.aliyun.com/#/model-market)，阅读并同意协议后，系统将自动开通阿里云百炼并发放**免费推理额度**。 > 如果未弹出服务协议，表示您已经开通过阿里云百炼且获得免费额度。 @@ -212,9 +212,12 @@ 免费额度列显示**无免费额度**或**免费额度**区域不显示，可能由以下原因之一导致： -- **免费额度已到期或耗尽**：免费额度的有效期为 30～90 天，从开通阿里云百炼或模型申请通过之日起计算，到期或耗尽后将不再显示，继续调用模型将产生计费。 -- **该模型所在地域或服务部署范围不享有免费额度**：仅华北2（北京）地域且服务部署范围为中国内地的模型、以及仅新加坡地域且服务部署范围为国际的模型享有免费额度，其他地域和部署范围无免费额度。 +- **免费额度已到期或耗尽**：免费额度的有效期为 90 天，从开通阿里云百炼、模型发布或模型申请通过之日起计算（以较晚者为准），到期或耗尽后将不再显示，继续调用模型将产生计费。 + +- **该模型所在地域或服务部署范围不享有免费额度**：仅华北2（北京）地域的模型以及仅新加坡地域的模型享有免费额度，其他地域和部署范围无免费额度。 + - **该模型本身不提供免费额度**：部分模型不参与新人免费额度发放。 + ### 使用哪种 API Key 才能消耗免费额度？ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/savings-plan-and-resource-package.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/savings-plan-and-resource-package.md index 5ce9dfe5..a1867096 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/savings-plan-and-resource-package.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/savings-plan-and-resource-package.md @@ -80,7 +80,7 @@ AI 通用型节省计划是针对大模型按量付费使用场景设计的折 - C 类：qwen3.6-max-preview、DeepSeek、Kimi、GLM、MiniMax、HappyHorse - > 三方直供模型不支持抵扣，详情参见[三方直供模型支持抵扣 AI 通用型节省计划吗？](#85a29cab67489) + > 三方直供模型不支持抵扣，其中 DeepSeek、Kimi、GLM 已有阿里云直供版本可抵扣，MiniMax 暂无阿里云直供版本。详情参见[三方直供模型支持抵扣 AI 通用型节省计划吗？](#85a29cab67489) **每月承诺消费金额范围** @@ -802,7 +802,7 @@ ASR模型按秒计费，TTS模型按字符计费，请前往[百炼控制台](ht - 根据[退订规则](https://help.aliyun.com/zh/user-center/cancel-subscription/)，预付费商品未发生使用的部分，可按未使用额度费用[申请退款](https://billing-cost.console.aliyun.com/refund/refund?commodityType=RESOURCE_PLANS&refundType=NOREASON_REFUND)；已使用的部分则无法退款。 -**使用限制**：资源包按模型名称严格匹配，**跨版本或子型号不通用**，请以资源包购买页标注的适用模型为准。例如，qwen-plus 资源包不支持抵扣 qwen-max 或 qwen-turbo 的调用费用；若需要同时覆盖多个模型版本的调用费用，建议选择 AI 通用型节省计划。 +**使用限制**：资源包按模型名称严格匹配，**跨版本或子型号不通用**，请以资源包购买页标注的适用模型为准。例如，qwen-plus 资源包不支持抵扣 qwen-max 的调用费用；若需要同时覆盖多个模型版本的调用费用，建议选择 AI 通用型节省计划。 ### **大语言模型推理资源包** @@ -812,48 +812,36 @@ ASR模型按秒计费，TTS模型按字符计费，请前往[百炼控制台](ht [大语言模型推理资源包 qwen-max](https://common-buy.aliyun.com/?commodityCode=sfm_llminference2_dp_cn#/buy) -[大语言模型推理资源包 qwen-turbo](https://common-buy.aliyun.com/?commodityCode=sfm_llminference3_dp_cn#/buy) - **适用地域** 华北2（北京） 华北2（北京） -华北2（北京） - **适用模型** qwen-plus 及 qwen-plus-latest的实时推理服务（[非思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)） qwen-max的实时推理服务（[非思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)） -qwen-turbo的实时推理服务（[非思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)） - **包含输入和输出总Tokens** 1,200万/1.1亿 1,800万/3,900万/3.9亿/11.7亿/19.5亿 -3,500万/3.5亿/17.5亿/35亿 - **价格（元）** 11.66/114.4 57.6/125/1250/3750/6250 -11.45/114.45/572.25/1144.5 - **有效期** 自购买日起生效，有效期可选 3 个月、6 个月或 1 年。 自购买之日起有效期为 1 年。 -自购买之日起有效期为 1 年。 - **使用限制** - **qwen-plus**、**qwen-plus-latest** @@ -866,7 +854,7 @@ qwen-turbo的实时推理服务（[非思考模式](https://help.aliyun.com/zh/m - [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)、[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)、[模型调优](https://help.aliyun.com/zh/model-studio/model-training-overview)、[模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-introduction)产生的费用。 -- **qwen-max**、**qwen-turbo** +- **qwen-max** - 仅支持抵扣实时推理产生的费用（[非思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)，包含输入和输出），不支持抵扣[Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)、[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)、[模型调优](https://help.aliyun.com/zh/model-studio/model-training-overview)、[模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-introduction)产生的费用。 @@ -895,7 +883,7 @@ qwen-turbo的实时推理服务（[非思考模式](https://help.aliyun.com/zh/m **图像编辑**：qwen-image-edit-plus -**资源包容量 (生成图片张数)** +**资源包容量（生成图片张数）** 80/400 @@ -962,7 +950,7 @@ qwen-turbo的实时推理服务（[非思考模式](https://help.aliyun.com/zh/m ### **三方直供模型支持抵扣 AI 通用型节省计划吗？** -[C 类模型](#ho1f5x10wuun0)中，阿里直供的模型支持抵扣，三方直供的模型不支持抵扣。可以在[百炼模型广场](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)中通过模型卡片右上角标识（如"阿里直供"或"三方直供"标签）判断。 +[C 类模型](#ho1f5x10wuun0)中，阿里直供的模型支持抵扣，三方直供的模型不支持抵扣。目前，DeepSeek、Kimi、GLM 有阿里云直供版本，可通过 AI 通用型节省计划抵扣；MiniMax 暂无阿里云直供版本，暂不支持通过节省计划抵扣。可以在[百炼模型广场](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)中通过模型卡片右上角标识（如"阿里直供"或"三方直供"标签）查看最新的直供模型列表。 ### **购买节省计划后如何使用？** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md index 03d7af80..0a979d61 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md @@ -1,7 +1,5 @@ # 常见问题 -汇总 Token Plan 团队版 Coding Plan 使用过程中的常见问题与解决方法。 - ## **使用阿里云 AI 助理** 推荐使用[阿里云 AI 助理](https://www.aliyun.com/ai-assistant/)，其知识库整合了阿里云官方帮助文档。 @@ -44,7 +42,7 @@ 3. API Key 复制不完整、有空格或拼写错误 -1. [使用套餐专属 API Key](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart#2782cf93b1w8h) +1. [使用套餐 API Key](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart#2782cf93b1w8h) 2. 前往[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)确认订阅是否过期 @@ -512,7 +510,7 @@ OpenClaw 开启思考模式的步骤。 - 确保已填写有效的 API Key（非空、无多余空格）。 - - 若使用 Coding Plan，请确保使用的是专属 Key（`sk-sp-xxx`）。 + - 若使用 Coding Plan，请确保使用的是套餐 Key（`sk-sp-xxx`）。 - 若 Web UI 中显示的是 `__OPENCLAW_REDACTED__`，表示 API Key 已保存，无需重填；若显示为空或 `YOUR_API_KEY`，则需重新填写。 @@ -535,11 +533,11 @@ OpenClaw 开启思考模式的步骤。 1. OpenClaw 配置错误 - 若 Base URL 或模型提供商配置有误，导致请求未进入 Coding Plan 专属通道，而是被路由到了 通用的API 调用，从而触发限流。 + 若 Base URL 或模型提供商配置有误，导致请求未进入 Coding Plan 通道，而是被路由到了 通用的API 调用，从而触发限流。 - 若使用 Coding Plan 套餐，请核对[OpenClaw配置文件](https://help.aliyun.com/zh/model-studio/openclaw#0c6a73ae73mqr)中的 `models`、`agents`、`gateway`（含嵌套字段），确保与文档配置一致。例如：模型服务提供商的结构为`{ "models": { "providers": { "bailian": {...} } } }` 。 - - 若当前未使用 Coding Plan 套餐，建议切换至 Coding Plan 以获取专属额度。 + - 若当前未使用 Coding Plan 套餐，建议切换至 Coding Plan 以获取套餐额度。 2. 超出套餐限额：在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)查看套餐用量情况。 @@ -596,6 +594,10 @@ OpenClaw 启动时出现以下报错信息： 2. 确认 Coding Plan 的 API Key 配置在 `models.providers.bailian.apiKey`，详情请参见[OpenClaw](https://help.aliyun.com/zh/model-studio/openclaw)。 +### **Coding Plan 支持 Claude Code 的** [**Agent Teams**](https://code.claude.com/docs/en/agent-teams) **功能吗？** + +暂不支持。使用时会报错"model `claude-opus-4-6` is not supported."。 + ### **Coding Plan 支持 OpenAI Responses API 吗？** 不支持。Coding Plan 仅支持 OpenAI Chat Completions 协议和 Anthropic Messages 协议，不支持 OpenAI Responses API。 @@ -604,11 +606,11 @@ OpenClaw 启动时出现以下报错信息： 可能原因： -- **API Key 格式错误**：API Key 填写为空、格式不正确、复制不完整，或在复制时误带了多余的空格。请确认 API Key 为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)（以 `sk-sp-` 开头），复制完整且无空格。 +- **API Key 格式错误**：API Key 填写为空、格式不正确、复制不完整，或在复制时误带了多余的空格。请确认 API Key 为 Coding Plan [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)（以 `sk-sp-` 开头），复制完整且无空格。 -- **Coding Plan 订阅状态已过期或失效：** Coding Plan 专属 API Key依赖于套餐的订阅状态。如果 Coding Plan 套餐已到期或失效，对应的专属 Key 将无法继续使用。请确保Coding Plan 订阅状态是否仍然有效。 +- **Coding Plan 订阅状态已过期或失效：** Coding Plan API Key 依赖于套餐的订阅状态。如果 Coding Plan 套餐已到期或失效，对应的 API Key 将无法继续使用。请确保Coding Plan 订阅状态是否仍然有效。 -- **使用了错误的Base URL**：配置了 Coding Plan 专属 API Key（以 `sk-sp-` 开头），但 Base URL 仍保留为阿里云百炼通用地址（如 [https://dashscope](https://dashscope).aliyuncs.com/compatible-mode/v1）。请根据[接入的AI工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)，将 Base URL 替换为下表中Coding Plan 专属地址。 +- **使用了错误的Base URL**：配置了 Coding Plan API Key（以 `sk-sp-` 开头），但 Base URL 仍保留为阿里云百炼通用地址（如 [https://dashscope](https://dashscope).aliyuncs.com/compatible-mode/v1）。请根据[接入的AI工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)，将 Base URL 替换为下表中 Coding Plan 地址。 **工具** @@ -652,7 +654,7 @@ OpenClaw 启动时出现以下报错信息： https://coding\-intl.dashscope.aliyuncs.com/v1 -- **使用了错误的 API Key**：配置了 Coding Plan 的专属 Base URL，但 API Key 误填了阿里云百炼的通用 API Key（以 `sk-` 开头）。请将 API Key 更新为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)。 +- **使用了错误的 API Key**：配置了 Coding Plan 的 Base URL，但 API Key 误填了阿里云百炼的通用 API Key（以 `sk-` 开头）。请将 API Key 更新为 Coding Plan [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)。 - **OpenClaw历史配置缓存导致配置错误：**请删除`~/.openclaw/agents/main/agent/models.json`文件中的`providers.bailian`配置项，并重启OpenClaw。 @@ -680,11 +682,11 @@ Coding Plan 仅限在编程工具（如 Claude Code、Qwen Code 等）中使用 如果开通 Coding Plan 后仍产生扣费或欠费，可能有以下原因： -1. **未正确配置专属 API Key 和 Base URL（最常见原因）** +1. **未正确配置套餐 API Key 和 Base URL（最常见原因）** - 原因：如果在 AI 工具中配置的是通用 API Key（格式为`sk-xxx`）和通用 Base URL（不含 coding 关键字），系统会将其识别为按量计费调用，产生按量计费的账单。 - - 解决方案：请务必使用 Coding Plan 专属配置。**API Key** 的格式必须为 `sk-sp-xxx`，**Base URL** 必须包含 `coding` 关键字（如 `https://coding.dashscope.aliyuncs.com/xxx`）。详情请参见[获取套餐专属 API Key 和 Base URL](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart#2782cf93b1w8h)。 + - 解决方案：请务必使用 Coding Plan 套餐配置。**API Key** 的格式必须为 `sk-sp-xxx`，**Base URL** 必须包含 `coding` 关键字（如 `https://coding.dashscope.aliyuncs.com/xxx`）。详情请参见[获取套餐专属 API Key 和 Base URL](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart#2782cf93b1w8h)。 2. **账单结算延时导致欠费（费用产生于开通Coding Plan套餐前）** @@ -694,7 +696,7 @@ Coding Plan 仅限在编程工具（如 Claude Code、Qwen Code 等）中使用 3. **同时配置了 Coding Plan和通用API调用凭证，但误用了通用API调用** - - 原因：若工具中同时保留了通用和专属两套配置，部分工具（如 OpenClaw）会自动路由到通用凭证进行请求，从而产生扣费。 + - 原因：若工具中同时保留了通用和套餐两套配置，部分工具（如 OpenClaw）会自动路由到通用凭证进行请求，从而产生扣费。 - 解决方案：建议在工具中移除通用 API 配置，并确保选用[Coding Plan 支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan#2bbc4faf2ej0e)。例如在 OpenCode 中，请选择供应商标为 **Model Studio Coding Plan** 的模型。 @@ -940,7 +942,7 @@ Coding Plan 套餐中包含的模型（如 glm-5、qwen3.5-plus 等）均为完 API Key 格式 -`sk-sp-xxx`（专属 Key） +`sk-sp-xxx`（套餐 Key） `sk-xxx`（百炼通用 Key） diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md index 1ab71393..57403f95 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md @@ -50,11 +50,11 @@ Coding Plan 整合了千问、GLM、Kimi 、MiniMax顶级模型，并兼容主 - 每月**90,000** 次请求 -- **限时优惠：**活动已结束，当前价格以下单页为准。 +- **限时优惠**：活动已结束，当前价格以下单页为准。 - **限量抢购**：名额有限、先到先得。每日 09:30:00（UTC+08:00）补充，可前往[Coding Plan 页面](https://www.aliyun.com/benefit/scene/codingplan)抢购。 -- **额度消耗：**单次提问将按实际“模型调用次数”扣除额度。简单任务约消耗 5-10 次，复杂任务约 10-30+ 次，实际消耗受任务难度、上下文及工具使用影响。在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)可以查看用量。 +- **额度消耗：**单次提问将按实际“模型调用次数”扣除额度。简单任务约消耗 5-10 次，复杂任务约 10-30+ 次，实际消耗受任务难度、上下文及工具使用影响。在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)可以查看用量。 - **额度恢复**： @@ -84,7 +84,7 @@ Coding Plan 整合了千问、GLM、Kimi 、MiniMax顶级模型，并兼容主 您需要获取并配置套餐专属的 API Key 和 Base URL，才能正确使用并抵扣套餐额度。 -- **API Key**：在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)，获取Coding Plan 专属 API Key（格式为`sk-sp-xxxxx`）。 +- **API Key**：在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)，获取Coding Plan 专属 API Key（格式为`sk-sp-xxxxx`）。 - **Base URL**：后续需在 AI 工具中配置以下其中一个Base URL（因工具而异），具体操作请参见对应的AI工具文档。 @@ -99,33 +99,33 @@ Coding Plan 专属的 API Key 和 Base URL 与百炼按量计费的 API Key（`s ### **步骤三：接入AI工具** - [**OpenClaw**开源、自托管个人 AI 助手](https://help.aliyun.com/zh/model-studio/openclaw) +[**OpenClaw**开源、自托管个人 AI 助手](https://help.aliyun.com/zh/model-studio/openclaw) - [**Hermes Agent**开源 AI 代理框架，内置自学习循环](https://help.aliyun.com/zh/model-studio/hermes-agent) +[**Hermes Agent**开源 AI 代理框架，内置自学习循环](https://help.aliyun.com/zh/model-studio/hermes-agent) - [**Claude Code**AI 终端编码助手，支持自然语言编程](https://help.aliyun.com/zh/model-studio/claude-code) +[**Claude Code**AI 终端编码助手，支持自然语言编程](https://help.aliyun.com/zh/model-studio/claude-code) - [**OpenCode**开源 AI 编程代理工具](https://help.aliyun.com/zh/model-studio/opencode) +[**OpenCode**开源 AI 编程代理工具](https://help.aliyun.com/zh/model-studio/opencode) - [**Cursor**AI 原生代码编辑器](https://help.aliyun.com/zh/model-studio/cursor) +[**Cursor**AI 原生代码编辑器](https://help.aliyun.com/zh/model-studio/cursor) - [**Codex**OpenAI 推出的命令行编程工具](https://help.aliyun.com/zh/model-studio/codex) +[**Codex**OpenAI 推出的命令行编程工具](https://help.aliyun.com/zh/model-studio/codex) - [**Qwen Code**开源命令行 AI 编码工具](https://help.aliyun.com/zh/model-studio/qwen-code) +[**Qwen Code**开源命令行 AI 编码工具](https://help.aliyun.com/zh/model-studio/qwen-code) - [**QwenPaw**开源个人 AI 助手，支持本地与云端部署](https://help.aliyun.com/zh/model-studio/qwenpaw) +[**QwenPaw**开源个人 AI 助手，支持本地与云端部署](https://help.aliyun.com/zh/model-studio/qwenpaw) - [**Cherry Studio**多模型桌面客户端](https://help.aliyun.com/zh/model-studio/cherry-studio) +[**Cherry Studio**多模型桌面客户端](https://help.aliyun.com/zh/model-studio/cherry-studio) - [**Chatbox**跨平台 AI 桌面客户端](https://help.aliyun.com/zh/model-studio/chatbox) +[**Chatbox**跨平台 AI 桌面客户端](https://help.aliyun.com/zh/model-studio/chatbox) - [**Cline**VS Code 扩展，智能代码补全和调试](https://help.aliyun.com/zh/model-studio/cline) +[**Cline**VS Code 扩展，智能代码补全和调试](https://help.aliyun.com/zh/model-studio/cline) - [**Qoder**面向真实软件开发的 Agentic 编码平台](https://help.aliyun.com/zh/model-studio/qoder-agent) +[**Qoder**面向真实软件开发的 Agentic 编码平台](https://help.aliyun.com/zh/model-studio/qoder-agent) - [**Lingma**阿里云推出的智能编码辅助工具](https://help.aliyun.com/zh/model-studio/lingma-agent) +[**Lingma**阿里云智能编码助手，提供独立 IDE](https://help.aliyun.com/zh/model-studio/lingma-agent) - [**Kilo CLI**轻量高性能命令行编程工具](https://help.aliyun.com/zh/model-studio/kilo-cli) +[**Kilo CLI**轻量高性能命令行编程工具](https://help.aliyun.com/zh/model-studio/kilo-cli) [··· **更多工具**其他编程工具](https://help.aliyun.com/zh/model-studio/more-tools) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md similarity index 77% rename from skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md rename to skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md index 23651b2d..dad44bd5 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md @@ -1,16 +1,16 @@ # 添加视觉理解能力 -百炼 Coding Plan 中的部分模型（qwen3.6-plus、qwen3.5-plus、kimi-k2.5）原生支持视觉理解，可直接处理图片输入。对于 glm-5、MiniMax-M2.5 等纯文本模型，可通过添加本地 Skill 使其获得视觉能力。 +Token Plan 支持的部分模型（qwen3.7-plus 等）原生支持视觉理解，可直接处理图片输入。对于 glm-5、MiniMax-M2.5 等纯文本模型，可通过添加本地 Skill 使其获得视觉能力。 **说明** -运行图片理解 Skill 会消耗 Coding Plan 额度，无其他收费项。 +运行图片理解 Skill 会消耗 Token Plan Credits，无其他收费项。 ## 前提条件 -1. 已订阅 [Coding Plan](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)，详情请参见[快速开始](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart)。 +1. 已订阅 [Token Plan](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription)。 -2. 已在 Coding Plan 工具中完成接入配置，且能正常对话，详情请参见[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。 +2. 已在 AI 工具中完成接入配置，且能正常对话，详情请参见[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。 ## 视觉支持情况 @@ -21,11 +21,13 @@ **说明** -- qwen3.6-plus +- qwen3.8-max-preview + +- qwen3.7-plus -- qwen3.5-plus +- qwen3.6-plus -- kimi-k2.5 +- kimi-k2.5 等 是 @@ -51,7 +53,7 @@ ## 方法 1：直接使用视觉模型（推荐） -qwen3.6-plus、qwen3.5-plus 和 kimi-k2.5 具备视觉理解能力。如果经常需要处理图片，直接切换到这些模型是最简单、推荐的做法。 +qwen3.7-plus 等模型具备视觉理解能力。如果经常需要处理图片，直接切换到这些模型是最简单、推荐的做法。 **工具** @@ -59,15 +61,15 @@ qwen3.6-plus、qwen3.5-plus 和 kimi-k2.5 具备视觉理解能力。如果经 Claude Code -`/model qwen3.6-plus`或`/model qwen3.5-plus`或 `/model kimi-k2.5` +`/model qwen3.7-plus`或`/model qwen3.6-plus`或`/model qwen3.5-plus`或 `/model kimi-k2.5` OpenCode -`/models`→ 搜索并选择`qwen3.6-plus`或`qwen3.5-plus`或`kimi-k2.5` +`/models`→ 搜索并选择`qwen3.7-plus`或`qwen3.6-plus`或`qwen3.5-plus`或`kimi-k2.5` Qwen Code -`/model`→ 选择`qwen3.6-plus`或`qwen3.5-plus`或`kimi-k2.5` +`/model`→ 选择`qwen3.7-plus`或`qwen3.6-plus`或`qwen3.5-plus`或`kimi-k2.5` 更多编程工具中的模型切换方式请参考[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。切换后可直接在对话中引用图片路径，或拖拽/粘贴图片。 @@ -91,9 +93,9 @@ Qwen Code --- name: image-analyzer description: 帮助没有视觉能力的模型进行图像理解。当需要分析图像内容、提取图片中的信息、文字、界面元素，或理解截图、图表、架构图等任何视觉内容时，使用此技能，传入图片路径即可获得描述信息。 - model: qwen3.6-plus + model: qwen3.7-plus --- - qwen3.6-plus具有视觉理解能力，请直接使用qwen3.6-plus模型进行图片理解。 + qwen3.7-plus具有视觉理解能力，请直接使用qwen3.7-plus模型进行图片理解。 ``` 创建完成后的目录结构如下： @@ -111,7 +113,7 @@ Qwen Code 2. 下载[aliyun.png](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260225/hxwnny/aliyun.png)到项目目录下，并提问：`请加载image-analyzer skill，描述一下 aliyun.png banner位置是什么信息。`可收到如下回复： - ![image.png](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5928202771/p1054884.png) + aliyun.png 为阿里云官网首页截图，banner 区域标题为 **Coding Plan 已支持 Qwen3.5**，正文介绍阿里云百炼支持 Qwen3.5、Kimi-k2.5、GLM-4.7 等模型，新客首月仅 7.9 元，页面提供**立即订阅**和**在线咨询**入口。 ### OpenCode @@ -128,13 +130,13 @@ Qwen Code **说明** - model 字段必须使用 OpenCode 配置文件中定义的 provider 和模型名称。参考 [OpenCode](https://help.aliyun.com/zh/model-studio/opencode) 文档的配置示例，应为`bailian-coding-plan/qwen3.6-plus`。 + model 字段必须使用 OpenCode 配置文件中定义的 provider 和模型名称。参考 [OpenCode](https://help.aliyun.com/zh/model-studio/opencode) 文档的配置示例，应为`bailian-token-plan/qwen3.7-plus`。 ``` --- description: Analyzes images using a vision-capable model. Use this agent when the user needs to understand image content, extract information from screenshots, diagrams, UI mockups, or any visual content. Invoke with @image-analyzer followed by the image path and your question. mode: subagent - model: bailian-coding-plan/qwen3.6-plus + model: bailian-token-plan/qwen3.7-plus tools: write: false edit: false @@ -156,7 +158,18 @@ Qwen Code 2. 下载[aliyun.png](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260225/hxwnny/aliyun.png)到项目目录下，通过`@`唤起`image-analyzer`并提问：`@image-analyzer，描述一下 aliyun.png banner位置是什么信息。`可收到如下回复： - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6472262771/p1055847.png) + ``` + Banner位置包含： + 左侧： + - 汉堡菜单图标 + - 阿里云橙色logo和"阿里云"文字 + - 导航菜单：大模型、产品、解决方案、权益、定价、云市场、伙伴、服务、了解阿里云 + 右侧： + - 搜索框(显示"大模型") + - 图标：蓝色圆圈、地球、耳机 + - 链接：文档、备案、控制台 + Build · glm-5 · 37.0s + ``` ## **常见问题** @@ -167,17 +180,17 @@ Qwen Code **解决方案**：在 OpenCode 配置文件的模型定义中添加 `modalities` 字段，将 `input` 设为 `["text", "image"]`，如下所示： -> 将sk-sp-xxx替换为Coding Plan API Key。 +> 将sk-sp-xxx替换为Token Plan API Key。 ``` { "$schema": "https://opencode.ai/config.json", "provider": { - "bailian-coding-plan-test": { + "bailian-token-plan": { "npm": "@ai-sdk/anthropic", - "name": "Model Studio Coding Plan", + "name": "Model Studio Token Plan", "options": { - "baseURL": "https://coding.dashscope.aliyuncs.com/apps/anthropic/v1", + "baseURL": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", "apiKey": "sk-sp-xxx" }, "models": { @@ -255,7 +268,7 @@ Qwen Code "mode": "merge", "providers": { "bailian": { - "baseUrl": "https://coding.dashscope.aliyuncs.com/v1", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", "apiKey": "YOUR_API_KEY", "api": "openai-completions", "models": [ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md new file mode 100644 index 00000000..88c71bd8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md @@ -0,0 +1,79 @@ +# 接入 Harness 工具 + +Token Plan 支持的部分 Qwen 模型内置 Harness 工具，可为 AI 编程工具扩展联网搜索、代码解释器、网页抓取等能力。 + +**说明** + +适用于 Token Plan，不适用于 Coding Plan。 + +## **工具概览** + +**工具** + +**说明** + +联网搜索 + +检索互联网信息，结合搜索结果生成回答 + +代码解释器 + +在沙箱环境中编写与运行 Python 代码，用于数学计算、数据分析等场景 + +网页抓取 + +访问指定 URL 并提取内容，为大模型提供所需信息 + +以图搜图 + +根据输入图片从互联网搜索视觉相似的图片，适用于以图找同款、视觉内容溯源等场景 + +文搜图 + +根据文本描述从互联网搜索相关图片，适用于可视化问答、配图推荐等场景 + +## **支持的模型和工具** + +### **个人版** + +**模型** + +**支持的工具** + +qwen3.8-max-preview + +联网搜索、代码解释器、网页抓取、以图搜图、文搜图 + +qwen3.7-max + +联网搜索、代码解释器、网页抓取 + +qwen3.7-plus + +联网搜索、代码解释器、网页抓取、以图搜图、文搜图 + +### **团队版** + +**模型** + +**支持的工具** + +qwen3.8-max-preview + +联网搜索、代码解释器、网页抓取、以图搜图、文搜图 + +qwen3.7-max + +联网搜索、代码解释器、网页抓取 + +qwen3.7-plus + +联网搜索、代码解释器、网页抓取、以图搜图、文搜图 + +## **费用说明** + +Harness 工具按成功调用次数计费，费用从套餐 Credits 中抵扣。 + +## **使用方式** + +将 AI 编程工具的模型切换为上述支持 Harness 的 Qwen 模型，在对话中直接提问即可。模型会根据问题自动调用相应的内置工具，无需额外配置。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md index 4daf85ea..4822f70b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md @@ -1,14 +1,6 @@ # 接入多模态生成模型 -图像生成模型需通过工具的扩展机制（Skill、Slash Command 或 Agent）接入。 - -## **前提：获取套餐专属凭证** - -在控制台「我的订阅」打开 Token Plan 套餐详情页，接入信息卡片展示套餐专属 API Key（以 `sk-sp-` 为前缀，掩码显示），支持生成、重置与复制 API Key。 - -**说明** - -套餐详情页「可使用模型」以文本、编程模型为主；图像生成模型不在该列表展示，需通过 `multimodal-generation` API 调用。 +Token Plan 中的图像生成、视频生成、语音合成模型需通过工具的扩展机制（Skill、Slash Command 或 Agent）接入。 ## **示例：在 Claude Code 中接入图像生成模型** @@ -16,7 +8,7 @@ ### **步骤一：创建 Slash Command** -将套餐专属 API Key（以 `sk-sp-` 为前缀）配置为环境变量 `$ANTHROPIC_AUTH_TOKEN`，供后续 curl 鉴权使用。 +将套餐 API Key（以 `sk-sp-` 为前缀）配置为环境变量 `$ANTHROPIC_AUTH_TOKEN`，供后续 curl 鉴权使用。 在项目根目录创建 `.claude/commands/text-to-image.md`，写入以下内容： @@ -27,7 +19,7 @@ ## 步骤 -1. 从用户需求中提取 prompt（图片描述）、model、size（默认 1024*1024）。若用户明确指定了模型（如“模型=wan2.7-image”或“用 wan2.7-image 画”），必须严格使用用户指定的模型名，不要回退到默认模型；仅当用户未指定模型时才使用默认 qwen-image-2.0。常用图像生成模型有 qwen-image-2.0、qwen-image-2.0-pro、wan2.7-image、wan2.7-image-pro、z-image-turbo 等，完整列表以百炼模型列表为准。 +1. 从用户需求中提取 prompt（图片描述）、model、size（默认 1024*1024）。若用户明确指定了模型（如“模型=wan2.7-image”或“用 wan2.7-image 画”），必须严格使用用户指定的模型名，不要回退到默认模型；仅当用户未指定模型时才使用默认 qwen-image-2.0。常用图像生成模型有 qwen-image-2.0、qwen-image-2.0-pro、wan2.7-image、wan2.7-image-pro 等，完整列表以百炼模型列表为准。 2. 调用 API 生成图片（使用 Bash 工具执行 curl）： @@ -55,9 +47,117 @@ curl -s -X POST "https://token-plan.cn-beijing.maas.aliyuncs.com/api/v1/services 在 Claude Code 中输入 `/text-to-image 画一只猫`。如需使用默认模型以外的图像生成模型，在指令中写明模型名即可，例如 `/text-to-image 用 wan2.7-image 画一只猫`。 +## **示例：在 Claude Code 中接入视频生成模型** + +以 Claude Code 为例，通过 Slash Command 接入视频生成模型。视频生成为异步接口，流程为"提交任务 → 轮询状态 → 下载视频"。 + +### **步骤一：创建 Slash Command** + +将套餐 API Key（以 `sk-sp-` 为前缀）配置为环境变量 `$ANTHROPIC_AUTH_TOKEN`，供后续 curl 鉴权使用。 + +在项目根目录创建 `.claude/commands/text-to-video.md`，写入以下内容： + +``` +调用 Token Plan 文生视频 API，根据描述生成视频并自动下载到本地。 + +用户需求：$ARGUMENTS + +## 步骤 + +1. 从用户需求中提取 prompt（视频描述）、model（默认 happyhorse-1.1-t2v）、resolution（默认 720P）、ratio（默认 16:9）、duration（默认 5 秒）。若用户明确指定了模型（如"模型=happyhorse-1.0-t2v"），必须严格使用用户指定的模型名。 + +2. 使用 Bash 工具执行以下脚本，一次性完成提交任务、等待完成、下载视频： + +```bash +#!/bin/bash +set -e + +TASK_RESPONSE=$(curl -s -X POST "https://token-plan.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis" \ + -H "X-DashScope-Async: enable" \ + -H "Authorization: Bearer $ANTHROPIC_AUTH_TOKEN" \ + -H "Content-Type: application/json" \ + -d '{ + "model": "", + "input": {"prompt": ""}, + "parameters": {"resolution": "", "ratio": "", "duration": } + }') + +TASK_ID=$(echo "$TASK_RESPONSE" | grep -o '"task_id":"[^"]*"' | head -1 | cut -d'"' -f4) +if [ -z "$TASK_ID" ]; then echo "提交失败: $TASK_RESPONSE"; exit 1; fi +echo "任务已提交，ID: $TASK_ID，等待生成..." + +while true; do + sleep 15 + STATUS_RESPONSE=$(curl -s "https://token-plan.cn-beijing.maas.aliyuncs.com/api/v1/tasks/$TASK_ID" \ + -H "Authorization: Bearer $ANTHROPIC_AUTH_TOKEN") + STATUS=$(echo "$STATUS_RESPONSE" | grep -o '"task_status":"[^"]*"' | cut -d'"' -f4) + if [ "$STATUS" = "SUCCEEDED" ]; then + VIDEO_URL=$(echo "$STATUS_RESPONSE" | grep -o '"video_url":"[^"]*"' | cut -d'"' -f4) + OUTPUT="generated_$(date +%Y%m%d_%H%M%S).mp4" + curl -s -o "$OUTPUT" "$VIDEO_URL" + echo "视频已下载: $(pwd)/$OUTPUT" + exit 0 + elif [ "$STATUS" = "FAILED" ]; then + echo "生成失败: $STATUS_RESPONSE"; exit 1 + fi + echo "生成中..." +done +``` + +3. 向用户展示生成的视频文件路径。 +``` + +### **步骤二：生成视频** + +在 Claude Code 中输入 `/text-to-video 一只白色的猫在阳台上晒太阳`。如需使用其他视频生成模型，在指令中写明模型名即可，例如 `/text-to-video 用 happyhorse-1.1-r2v 生成一只猫跳跃的视频`。 + +## **示例：在 Claude Code 中接入语音合成模型** + +以 Claude Code 为例，通过 Slash Command 接入语音合成模型。语音合成为同步接口，支持非流式和流式两种调用模式。 + +### **步骤一：创建 Slash Command** + +将套餐 API Key（以 `sk-sp-` 为前缀）配置为环境变量 `$ANTHROPIC_AUTH_TOKEN`，供后续 curl 鉴权使用。 + +在项目根目录创建 `.claude/commands/text-to-speech.md`，写入以下内容： + +``` +调用 Token Plan 语音合成 API，将文本转为语音文件。 + +用户需求：$ARGUMENTS + +## 步骤 + +1. 从用户需求中提取 text（待合成文本）、voice（音色，默认 longanhuan_v3.6）、format（音频格式，默认 mp3）、sample_rate（采样率，默认 24000）。若用户明确指定了模型，使用用户指定的模型名；否则默认使用 qwen-audio-3.0-tts-plus。 + +2. 调用 API 合成语音（使用 Bash 工具执行 curl）： + +``` +curl -s -X POST "https://token-plan.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/SpeechSynthesizer" \ + -H "Authorization: Bearer $ANTHROPIC_AUTH_TOKEN" \ + -H "Content-Type: application/json" \ + -o "speech_$(date +%Y%m%d_%H%M%S)." \ + -d '{ + "model": "", + "input": { + "text": "", + "voice": "", + "format": "", + "sample_rate": + } + }' +``` + +3. 向用户展示生成的音频文件路径。 +``` + +### **步骤二：合成语音** + +在 Claude Code 中输入 `/text-to-speech 你好，欢迎使用百炼`。 + ## **其他工具** -控制台套餐详情页「快速接入 AI 编程工具」入口提供 Qwen Code、Qoder、OpenClaw、Claude Code、OpenCode 等工具的接入文档。不同工具的扩展机制和配置文件路径如下表所示，该表为支持扩展机制的主流 AI 编程工具示例。将上述 Claude Code 示例中的配置内容保存到对应路径即可。 +不同工具的扩展机制和配置文件路径如下表所示，该表为支持扩展机制的主流 AI 编程工具示例。将上述 Claude Code 示例中的配置内容保存到对应路径即可。 工具 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md deleted file mode 100644 index aaf9db77..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md +++ /dev/null @@ -1,253 +0,0 @@ -# 工具调用 - -Token Plan 团队版支持通过模型内置工具和 MCP 服务两种方式为 AI 编程工具扩展能力，如联网搜索、代码解释器、网页抓取等。 - -## **工具概览** - -Token Plan 团队版提供两种方式接入工具： - -- **模型内置工具**：qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash 模型的 Responses API 内置了联网搜索、代码解释器、网页抓取、以图搜图、文搜图五种工具。启用后，模型会在需要时自动调用相应工具。 - -- **MCP 服务**：其他模型（如 deepseek-v3.2、glm-5 等）可通过百炼 MCP 广场的 MCP 服务获取工具能力。本文以联网搜索 MCP 为例说明接入方式，其他 MCP 服务的接入方式类似。 - - -## **费用说明** - -### **模型内置工具** - -qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash 模型内置工具的费用可通过 Token Plan 团队版抵扣，内置工具不额外收费，产生的 token 消耗统一从套餐 Credits 中抵扣。具体价格以[控制台模型详情页](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3.6-plus)为准。 - -### **MCP 服务** - -百炼 MCP 广场提供联网搜索、代码解释器、网页抓取等 MCP 服务。联网搜索 MCP 全部用户前 2000 次调用免费，免费额度用尽后按 29 元/千次计费；其他 MCP 服务部分**限时免费**，每月提供一定免费额度。具体价格以[MCP 广场](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/mcp-market)各服务详情页为准。 - -## **使用方式** - -### **使用 qwen3.7-max / qwen3.7-plus / qwen3.6-plus / qwen3.6-flash 模型内置工具** - -将 AI 工具的模型设置为 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus` 或 `qwen3.6-flash`，在对话中直接提问即可。模型会根据问题自动调用相应的内置工具： - -**工具** - -**说明** - -联网搜索 - -检索互联网信息，结合搜索结果生成回答 - -代码解释器 - -调用模型时启用内置的 Python 代码解释器，可使模型在沙箱环境里编写与运行 Python 代码，以解决数学计算、数据分析等复杂问题。 - -网页抓取 - -网页抓取工具可以访问指定 URL 并提取内容，为大模型提供所需信息。 - -以图搜图 - -图搜图工具使模型能够根据输入图片从互联网搜索视觉相似的图片，并基于搜索结果进行分析和推理，适用于以图找同款、视觉内容溯源等场景。 - -文搜图 - -文搜图工具使模型能够根据文本描述从互联网搜索相关图片，并基于图片内容进行描述和推理，适用于可视化问答、配图推荐等场景。 - -### **通过 MCP 服务接入工具** - -其他模型可通过百炼 MCP 广场的 MCP 服务获取工具能力。以下以联网搜索 MCP 为例说明接入方式。 - -#### **前提条件** - -已获取[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。此处的 API Key 为百炼通用 API Key（格式为 sk-xxx），用于调用 MCP 服务，与 Token Plan 团队版专属 API Key（格式为 sk-sp-xxx）不同。 - -#### **开通 MCP 服务** - -1. 进入百炼的[MCP 广场](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/mcp-market)，找到需要的 MCP 服务（如联网搜索）。 - -2. 点击**立即开通**，确认开通。 - -3. 开通成功后，获取以下配置信息： - - - **Streamable HTTP Endpoint**：MCP 服务的连接地址。 - - - **API Key**：即百炼 API Key，是 MCP 服务的鉴权密钥。 - - -#### **接入工具** - -将 MCP 服务添加到 AI 编程工具中。以下以联网搜索 MCP 为例，示例中的 `YOUR_API_KEY` 需替换为百炼 API Key。接入其他 MCP 服务时，将 Endpoint 地址替换为对应服务的地址即可。 - -## OpenClaw - -1. 在终端执行如下命令安装 MCPorter。 - - ``` - npm install -g mcporter - ``` - -2. 在终端执行如下命令启用 MCPorter。 - - ``` - openclaw config set skills.entries.mcporter.enabled true - ``` - -3. 在 `~/.openclaw/workspace` 目录下，执行如下命令添加联网搜索 MCP。 - - ``` - mcporter config add WebSearch https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp --transport http --header "Authorization=Bearer YOUR_API_KEY" - ``` - -4. 执行如下命令确认 MCP 已安装。 - - ``` - mcporter list - ``` - -5. 执行如下命令使配置生效。 - - ``` - openclaw gateway restart - ``` - -6. 发送提问 `用 mcporter 搜索阿里云的新闻` 即可看到搜索结果。 - - -## OpenCode - -1. 在配置文件 `~/.config/opencode/opencode.json` 中写入 MCP 配置信息。 - - ``` - { - "mcp": { - "WebSearch": { - "type": "remote", - "httpUrl": "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp", - "headers": { - "Authorization": "Bearer YOUR_API_KEY" - } - } - } - } - ``` - - 若 `opencode.json` 中已有其他配置（如 provider），将 mcp 字段合并到现有配置中即可。 - -2. 在终端执行以下命令进入 OpenCode。 - - ``` - opencode - ``` - -3. 在对话框执行 `/mcps` 确认 `websearch` 状态为 Enabled。确认后按 Esc 退出。 - -4. 发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - - -## Claude Code - -1. 在终端执行以下命令添加联网搜索 MCP 服务。 - - ``` - claude mcp add WebSearch https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp -t http -H "Authorization: Bearer YOUR_API_KEY" - ``` - - 终端返回 `Added SSE MCP server xx` 即表示添加成功。 - -2. 执行以下命令进入 Claude Code。 - - ``` - claude - ``` - -3. 在对话框执行 `/mcp` 命令，确认 `websearch` 的状态为 connected。 - -4. 按 Esc 退出 MCP 列表后，发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - - -## Qwen Code - -1. 在终端执行以下命令添加联网搜索 MCP。 - - ``` - qwen mcp add WebSearch \ - -t http \ - "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp" \ - -H "Authorization: Bearer YOUR_API_KEY" - ``` - -2. 在终端执行以下命令进入 Qwen Code。 - - ``` - qwen - ``` - -3. 在对话框执行 `/mcp` 命令确认 MCP 连接状态。 - -4. 发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - - -## Kilo CLI - -1. 在配置文件 `~/.config/kilo/opencode.json` 中写入 MCP 配置信息。 - - ``` - { - "mcp": { - "websearch": { - "type": "remote", - "url": "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp", - "headers": { - "Authorization": "Bearer YOUR_API_KEY" - } - } - } - } - ``` - - 若 `opencode.json` 中已有其他配置（如 provider），将 mcp 字段合并到现有配置中即可。 - -2. 在终端执行以下命令查看 MCP 状态。Connected 即表示连接成功。 - - ``` - kilocode mcp list - ``` - -3. 在终端执行以下命令进入 Kilo CLI。 - - ``` - kilo - ``` - -4. 发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - - -## Kilo Code IDE 插件 - -1. 打开 Kilo Code IDE 插件，配置联网搜索 MCP 信息。 - - ``` - { - "mcpServers": { - "websearch": { - "type": "streamable-http", - "url": "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp", - "headers": { - "Authorization": "Bearer YOUR_API_KEY" - } - } - } - } - ``` - - 当联网搜索的 MCP 状态显示为绿色时，表示添加成功。 - -2. 返回对话界面，发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md similarity index 51% rename from skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md rename to skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md index 3208f230..5fd78221 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md @@ -1,6 +1,6 @@ # 联网搜索 -在 Coding Plan 支持的编程工具中添加联网搜索工具，使模型能够检索实时信息。 +在 Token Plan 支持的编程工具中添加联网搜索工具，使模型能够检索实时信息。 ## 适用范围 @@ -8,11 +8,11 @@ ## 前提条件 -1. 已订阅 [Coding Plan](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)，详情请参见[快速开始](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart)。 +1. 已订阅 [Token Plan](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription)，详情请参见[快速开始](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart)。 -2. 已在 Coding Plan 工具（如 Claude Code、Qwen Code）中完成接入配置，且能正常对话，详情请参见[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。 +2. 已在 Token Plan 支持的工具（如 Claude Code、Qwen Code）中完成接入配置，且能正常对话，详情请参见[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。 -3. 已获取[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。此处的 API Key 为百炼通用 API Key（格式为 sk-xxx），用于调用 MCP 服务，与 Coding Plan 专属 API Key（格式为 sk-sp-xxx）不同。 +3. 已获取[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。此处的 API Key 为百炼通用 API Key（格式为 sk-xxx），用于调用 MCP 服务，与 Token Plan 专属 API Key（格式为 sk-sp-xxx）不同。 ## 开通或升级联网搜索 MCP @@ -27,6 +27,8 @@ - 联网搜索 MCP 全部用户前 2000 次调用免费，免费额度用尽后按 29 元/千次计费。如果使用第三方 MCP 服务，该 MCP 服务可能收费，以 MCP 服务的介绍信息为准。 + - 部分MCP服务支持**个人FC资源部署**，按实际调用时长和次数计费，适用于需要专属资源、指定资源地域等场景。 + 3. 开通成功后，可以获取以下配置信息： 1. **Streamable HTTP Endpoint**：MCP 服务的连接地址。联网搜索 MCP 的连接地址为`https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`。 @@ -38,10 +40,12 @@ 1. 进入百炼的[MCP广场](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/mcp-market)，找到**联网搜索** MCP 服务。 -2. 单击右侧**取消开通**，再单击**立即开通**。 +2. 单击右侧**取消开通**，再单击**立即开通** > **确认开通**。 - 联网搜索 MCP 全部用户前 2000 次调用免费，免费额度用尽后按 29 元/千次计费。如果使用第三方 MCP 服务，该 MCP 服务可能收费，以 MCP 服务的介绍信息为准。 + - 部分MCP服务支持**个人FC资源部署**，按实际调用时长和次数计费，适用于需要专属资源、指定资源地域等场景。 + 3. 重新开通成功后，即完成协议升级，可以获取以下配置信息： 1. **Streamable HTTP Endpoint**：MCP 服务的连接地址。联网搜索 MCP 的连接地址为`https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`。 @@ -78,8 +82,11 @@ ``` mcporter list ``` - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9753812771/p1055340.png) + ``` + mcporter 0.7.3 — Listing 1 server(s) (per-server timeout: 30s) + - websearch (1 tool, 2.4s) + [ok] Listed 1 server (1 healthy). + ``` 5. 在终端执行如下命令使配置生效。 @@ -89,7 +96,7 @@ 6. 发送提问`用 mcporter 搜索阿里云的新闻`即可看到搜索结果。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9753812771/p1055129.png) + 系统返回包含 4 条阿里云相关新闻的卡片，每条新闻显示标题、来源及发布时间，底部提示是否需要抓取某篇新闻的详细内容。 ## OpenCode @@ -120,13 +127,27 @@ 3. 在对话框执行`/mcps`确认`websearch`状态是否为 Enabled。确认后按 Esc 退出。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4753812771/p1054875.png) - 4. 发送提问`用websearch MCP搜索阿里云的新闻`即可看到搜索结果。 > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9228202771/p1054876.png) + ``` + 已为您搜索到最新的阿里云相关新闻： + 主要新闻摘要 + 1. 阿里云2026年战略目标 + 阿里云目标在2026年拿下中国AI云市场增量的80%，重点发力MaaS（模型即服务）和GPU算力服务。 + 目前阿里云上调用大模型API的客户中，有七成同时使用其GPU算力服务。 + 2. 阿里云服务器2026年优惠活动 + - 最便宜服务器：38元/年（2核2G、200M峰值带宽） + - 99元/年 ECS服务器（新老用户同享，续费同价） + - 199元/年企业专享服务器（2核4G、5M带宽） + 3. AI自动化工具生态 + 阿里云推出OpenClaw（原Clawdbot）AI + 自动化代理工具部署方案，支持一键部署到轻量应用服务器，可与阿里云百炼大模型无缝联动。 + 4. 企业智能客服 + 瓴羊Quick Service智能客服平台基于通义千问大模型，提供大模型驱动的语义理解、多轮对话、 + 知识库智能生成等功能。 + ``` ## Claude Code @@ -155,17 +176,54 @@ 3. 在对话框执行`/mcp`命令，确认`websearch`的状态为 connected。首次添加可能需要等待状态从 connecting 变成 connected。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054987.png) - 若连接状态显示 failed，请选中该 MCP 并选择 Reconnect 重连。若重试 1-2 次仍失败，请核实配置。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0613342771/p1055684.png) + ``` + /mcp + WebSearch MCP Server + Status: [ok] connected + Auth: [ok] authenticated + URL: https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp + xxx + Capabilities: tools + Tools: 1 tools + 1. View tools + 2. Re-authenticate + 3. Clear authentication + > 4. Reconnect + 5. Disable + ``` 4. 按 Esc 退出 MCP 列表后，发送提问`用websearch MCP搜索阿里云的新闻`即可看到搜索结果。 > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054990.png) + ``` + > 用websearch MCP搜索阿里云的新闻 + * websearch – bailian_web_search (MCP)(query: "阿里云 新闻", count: 10) + └ { + "status": 0, + "pages": [ + … +51 lines (ctrl+o to expand) + ] + } + * 搜索结果已返回，以下是阿里云的最新新闻摘要： + 阿里云近期新闻 + 1. 编程模型订阅服务推出（2026 年 2 月 25 日） + 阿里云百炼推出 Coding Plan 编程模型订阅服务，包含四款开源模型 API： + - Qwen3.5、GLM-5、MiniMax M2.5、Kimi K2.5 + - 用户可在 Qwen Code、Claude Code、Cline 等 AI 工具上无缝切换使用 + - 降低开发者模型选型与接入成本 + 2. 千问 3.5 新模型开源（2026 年 2 月 25 日） + 阿里发布三款中型千问 3.5 新模型： + - Qwen3.5-35B-A3B、Qwen3.5-122B-A10B、Qwen3.5-27B + - 基于 Qwen3.5-35B-A3B 的托管模型 Qwen3.5-Flash 已上线阿里云百炼 + - 每百万 Token 输入低至 0.2 元 + 3. 市场表现亮眼 + - 千问（Qwen）大模型调用量占比跃升至 32.1%，排名中国企业级大模型第一 + - 2026 财年第二季度收入 398.24 亿元，同比增长 34% + - AI 相关产品收入连续第九个季度实现三位数增长 + ``` ## Qwen Code @@ -197,13 +255,80 @@ 3. 在`Qwen Code`对话框执行`/mcp`命令以确认 MCP 连接状态。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4753812771/p1054878.png) + ``` + > /mcp + Configured MCP servers: + * WebSearch – Ready (1 tool) + Tools: + – mcp__WebSearch__bailian_web_search + [tip] Tips: + – Use /mcp desc to show server and tool descriptions + – Use /mcp schema to show tool parameter schemas + – Use /mcp nodesc to hide descriptions + – Use /mcp auth to authenticate with OAuth-enabled servers + – Press Ctrl+T to toggle tool descriptions on/off + ``` 4. 发送提问`用 websearch MCP 搜索阿里云的新闻`即可看到搜索结果。 > 为了避免和其他工具混淆，建议提问时明确提及`websearch MCP`。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9228202771/p1054879.png) + ``` + * 搜索结果已返回，我需要整理并总结其中与阿里云相关的主要新闻信息，以清晰的方式呈现给用户。 + * 已为您搜索到阿里云相关新闻，以下是主要内容： + 阿里云最新动态 + 1. 阿里云千问大模型表现亮眼 + - 2025 年下半年，阿里云千问（Qwen）在中国企业级大模型市场占比跃升至32.1%，几乎翻倍（上半年为 + 17.7%），位居第一 + - 字节豆包占比 21.3% 排名第二，DeepSeek 占比 18.4% 排名第三 + 2. AI 战略投入 + - 阿里豪掷3800 亿元押注 AI 基础设施 + - 2026 年，阿里云将 AI 相关收入目标增速上调至200% + - 2026 年 2 月推出30 亿元"春节请客计划"，联合淘宝、飞猪、盒马等以免单形式发放福利 + 3. Coding Plan 服务更新 + - 阿里云 Coding Plan 订阅服务新增支持 Qwen 3.5-Plus、GLM-4.7、Kimi-K2.5 等编程模型 + - 新用户首月首购享两折优惠 + 4. 技术进展 + - 除夕夜开源的千问 Qwen3.5-Plus 采用混合注意力机制和稀疏 MoE 架构 + - 总参数 3970 亿，激活 170 亿，推理速度达主流模型的 2 倍 + 需要我为您搜索更具体的阿里云新闻主题吗？ + > 输入您的消息或 @ 文件路径 + 按 ? 查看快捷键 1.8% context used + ``` + + +## Cline + +1. 点击MCP Servers，进入Configure页面，点击Configure MCP Servers。 + + ![2026-02-26\_13-58-09](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054977.jpg) + +2. 在配置文件`cline_mcp_settings.json`中写入 MCP 配置信息，并将`YOUR_API_KEY`替换为上一步骤获取的 API Key。配置完成后，保存文件。 + + ``` + { + "mcpServers": { + "websearch": { + "type": "http", + "url": "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp", + "headers": { + "Authorization": "Bearer YOUR_API_KEY" + }, + "disabled": false + } + } + } + ``` + +3. 配置完成后，在Configure页面下可以查看MCP的相关信息。 + + ![2026-02-26\_14-05-44](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054984.jpg) + +4. 新开对话，发送提问`用websearch MCP搜索阿里云的新闻`即可看到搜索结果。 + + > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 + + ![2026-02-26\_14-10-05](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054989.jpg) ## Kilo CLI @@ -231,8 +356,14 @@ ``` kilocode mcp list ``` - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0248612771/p1055115.png) + ``` + MCP Servers + │ + * [ok] websearch connected + │ https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp + │ + └ 1 server(s) + ``` 3. 在终端中执行以下命令进入 Kilo CLI。 @@ -244,14 +375,31 @@ > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9228202771/p1054891.png) + ``` + 以下是阿里云的最新新闻要点： + 阿里云千问大模型市场份额跃升第一 + - 市场份额翻倍：根据沙利文2026年2月报告，阿里云千问（Qwen）占比跃升至*32.1%*，相较2025年上半年的17.7%几乎翻倍，超越字节豆包（21.3%）和DeepSeek(18.4%)，成为企业级大模型市场第一 + - 日均调用量激增：2025年下半年中国企业级大模型日均调用量达37.0万亿tokens，较上半年增长263% + AI基础设施投资战略 + - 3800亿元投资：阿里宣布豪掷3800亿元押注AI基础设施 + - 收入目标：2026年将AI相关收入目标增速上调至200% + - 市场目标：阿里云目标拿下中国AI云市场增量的80% + 千问模型技术进展 + - 开源Qwen3.5-Plus：除夕夜开源，采用混合注意力机制和稀疏MoE架构 + - 高效能设计：总参数3970亿，仅激活170亿，推理速度达到主流模型的2倍 + - 春节免单计划：千问启动30亿元"春节请客计划"，联合淘宝、飞猪、盒马等生态业务发放福利 + 其他动态 + - Coding Plan上新：支持千问3.5、GLM-4.7、Kimi-K2.5等编程模型，新用户首月两折优惠 + - 云服务器优惠：轻量应用服务器38元/年起，200M带宽不限流量 + Code · kimi-k2.5 · 49.8s + ``` ## Kilo Code IDE 插件 1. 打开Kilo Code IDE插件配置联网搜索 MCP 信息，并将`YOUR_API_KEY`替换为上一步骤获取的 API Key。配置完成后，保存文件。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3100902771/p1055026.png) + 依次单击左上角齿轮图标，在侧边栏选择 **Agent Behaviour**，单击 **MCP Servers** 页签，然后单击底部 **Edit Global MCP** 按钮，打开 `mcp_settings.json` 配置文件，将以下 JSON 内容粘贴并保存。 ``` { @@ -269,14 +417,10 @@ 当联网搜索的MCP状态显示为绿色时，表示添加成功。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3100902771/p1055021.png) - 2. 返回对话界面，发送提问`用websearch MCP搜索阿里云的新闻`即可看到搜索结果。 > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3100902771/p1055023.png) - ## 常见问题 @@ -292,7 +436,7 @@ - 如果使用的 URL 为 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/sse`，说明开通的是旧版 SSE 协议，请将协议[升级至Streamable HTTP](#da7cd47fa8pcd)。 -3. **API Key 错误**：请确认使用了有效的百炼通用 API Key（格式为 sk-xxx，非 Coding Plan 专属 API Key），并已在命令中正确替换 YOUR\_API\_KEY。 +3. **API Key 错误**：请确认使用了有效的百炼通用 API Key（格式为 sk-xxx，非 Token Plan 专属 API Key），并已在命令中正确替换 YOUR\_API\_KEY。 4. **免费额度用尽**：联网搜索 MCP 全部用户前 2000 次调用免费，免费额度用尽后按 29 元/千次计费，请确认账户余额充足。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-faq.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-faq.md deleted file mode 100644 index 76b285c1..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-faq.md +++ /dev/null @@ -1,292 +0,0 @@ -# 常见问题 - -Token Plan 团队版常见问题汇总，涵盖购买、使用、计量和性能相关的问题解答。 - -## **Token Plan 团队版和 Coding Plan 有什么区别？** - -**Token Plan 团队版** - -**Coding Plan** - -适用场景 - -一人公司/团队/企业日常办公 - -个人开发场景 - -支持的模型 - -文本生成、图像生成模型 - -文本生成模型 - -计费方式 - -按 Token 消耗抵扣 Credits - -按模型调用次数 - -使用频次 - -无每 5 小时/每周限额 - -每 5 小时/每周限额 - -API Key 和 Base URL - -在[管理后台](https://tokenplan-enterprise.bailian.aliyunportal.com)生成专属 API Key，Base URL 详见[快速开始](https://help.aliyun.com/zh/model-studio/token-plan-quickstart) - -在[Coding Plan 页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/coding-plan)获取专属 API Key 和专属 Base URL - -高峰期性能 - -多租户隔离 - -高峰期间可能排队 - -数据安全 - -承诺不使用数据训练模型 - -用户数据授权 - -## **接入与调用** - -### **如何在编程工具中使用图像生成模型？** - -图像生成模型使用独立的接口，无法通过文本模型的 Base URL 直接调用。需要通过工具的 Skill 或扩展机制接入，具体配置方法请参见[接入多模态生成模型](https://help.aliyun.com/zh/model-studio/token-plan-multimodal-gen)。 - -### **常见报错及解决方案** - -**报错信息** - -**可能原因** - -**解决方案** - -**401 InvalidApiKey: No API-key provided.** - -请求头中未携带 API Key（`Authorization: Bearer` 或 `x-api-key` 均未传）。 - -在管理后台生成 API Key，并在工具中完成配置。 - -**401 InvalidApiKey: Invalid API-key provided.** - -1. 误用了百炼通用 API Key（sk-xxx 格式）或 Coding Plan 的 API Key - -2. Token Plan 团队版订阅过期 - -3. API Key 复制不完整或包含空格 - - -1. 确认使用的是 Token Plan 专属 API Key，确保完整且无空格。 - -2. 确认订阅是否过期。 - -3. 如仍报错，重置 API Key，重置后使用新 Key 配置。 - - -**404 model 'xxx' not found or not supported** - -**400 Model not exist.** - -1. 模型名称拼写错误或大小写错误 - -2. 模型 ID 不在套餐支持列表中 - - -1. 确认模型名称区分大小写，与套餐支持的模型 ID 一致。 - -2. 检查所选套餐是否包含该模型。 - - -**401 invalid access token or token expired** - -误用了 Coding Plan 或其他套餐的 Base URL - -Anthropic 兼容端点：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` - -OpenAI 兼容端点：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - -**401 Incorrect API key provided** - -误用了百炼通用 Base URL（dashscope.aliyuncs.com） - -Anthropic 兼容端点：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` - -OpenAI 兼容端点：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - -**400 InvalidParameter: Range of input length should be \[1, xxx\]** - -输入内容（含对话历史、代码上下文等）超出模型的最大上下文长度 - -新建会话清空历史，或使用工具自带的上下文压缩命令（如 Claude Code 的 `/compact`、Qwen Code 的 `/clear`）。也可切换上下文窗口更大的模型。 - -**400 InvalidParameter: url error, please check url！** - -Base URL 路径与协议不匹配。例如把 OpenAI 兼容路径配在 Anthropic 端点上，或反之。 - -按工具实际使用的协议选择对应的端点： - -- Anthropic 兼容协议（Claude Code 等）：以 `/apps/anthropic` 结尾。 - -- OpenAI 兼容协议（Cursor、Qwen Code 等）：以 `/compatible-mode/v1` 结尾。 - - -**400 InvalidParameter: Range of max\_tokens should be \[1, xxxx\]** - -请求中的 `max_tokens`（或工具配置中的最大输出长度）超出当前模型支持的最大输出 Token 数。 - -将 `max_tokens` 调整为不超过报错信息中提示的上限值。 - -**400 invalid\_parameter\_error: The thinking\_budget parameter must be a positive integer and not greater than xxxxx** - -工具配置中的思维链长度（如 `thinking_budget`、`budgetTokens`）超过当前模型支持的上限。各模型上限不同，以报错中的数值为准。 - -将思维链长度调整为不超过报错提示的上限值，或在不支持思考模式的模型上移除该配置项。 - -**400 data\_inspection\_failed: Input text data may contain inappropriate content.** - -输入或输出命中平台内容安全策略。 - -修改输入内容后重新提交。如多次触发，调整提示词避免敏感话题。 - -**429 API-Key Requests rate limit exceeded, please try again later.** - -短时间内请求过于密集，触发模型调用限流。 - -等待一分钟后重试；如频繁触发请降低请求频率，并确认 API Key 未被他人共享使用。 - -**429 Throttling.AllocationQuota: Allocated quota exceeded, please increase your quota limit.** - -**insufficient\_quota: You exceeded your current quota, please check your plan and billing details.** - -该报错可能由以下两种原因触发： - -**套餐额度已用尽**：坐席额度和共享用量包均已耗尽。 - -**触发模型调用限流**：即使套餐额度充足，每秒或每分钟消耗的 Token 数（TPS/TPM）超过模型限流阈值也会触发。限流按主账号维度计算，账号下所有 RAM 子账号、业务空间和 API Key 的调用量合并计算；即使每分钟总调用量未超限，短时间内的请求激增也可能触发。 - -**额度已用尽**：可加购坐席（加购后需将新坐席分配给成员后再使用）、加购共享用量包，或等待下一计费周期额度自动重置。 - -**触发限流**：等待约一分钟后重试，并采用平滑请求策略（如匀速调度、指数退避或请求队列缓冲）避免瞬时高峰。 - -**Connection error** - -Base URL 域名拼写错误或网络连接异常 - -检查 Base URL 域名拼写及网络连接。 - -## **产品功能相关** - -### **Token Plan 团队版的 API Key 能与其他套餐或普通 API 混用吗？** - -不能。Token Plan 团队版、Coding Plan 和百炼按量计费三者的 API Key 和 Base URL 互不相通，请勿混用。误用其他 API Key 不会抵扣 Token Plan 团队版的套餐额度。 - -### **能在多个工具中使用同一订阅吗？** - -可以。同一 API Key 可在全部兼容的 AI 编程和智能体工具中使用，额度共享消耗。每个成员持有独立的 API Key，不可共享给其他成员。 - -### **有哪些使用限制？** - -仅限在兼容的 AI 编程和智能体工具中交互式使用，不可用于自动化脚本或应用后端。违规使用可能导致订阅暂停或 API Key 封禁。 - -### **团队管理入口在哪里？** - -阿里云主账号或 RAM 用户登录[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)后，在左侧菜单进入**我的订阅**，通过订阅卡片的**设置**、**用量分析**、**分配座席**入口进入成员管理与设置面板；也可点击**进入管理平台**跳转独立管理平台。通过 SSO 或钉钉加入的成员，通过管理员分发的**管理平台地址**（形如 tokenplan-enterprise.bailian.aliyunportal.com）登录管理平台。详见[访问入口](https://help.aliyun.com/zh/model-studio/token-plan-team#tp05-enter)。 - -### **成员如何获取 API Key？** - -管理员在管理后台创建成员账号并分配席位后，为成员生成 API Key。成员无法自行生成，需联系管理员获取。详见[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)。 - -### **回收席位、修改角色、移出组织有什么区别？** - -这三个操作都在成员管理页面执行，但作用范围不同： - -- **回收席位**：撤销成员的席位使用权并将席位释放回席位池。成员失去使用权，但仍留在组织中，可被重新分配席位。 - -- **修改角色**：变更成员的权限（如管理员/普通成员），不影响席位分配和组织归属。 - -- **移出组织**：将成员从团队完全移除，席位自动回收至席位池，成员从成员列表消失。 - - -### **为什么 API Key 只能查看一次，丢失后如何处理？** - -为避免团队间 API Key 混用导致计费混乱，API Key 仅在首次生成或重置时显示，后续无法再次查看或复制。若 API Key 丢失，在**成员管理**页面找到对应成员，点击**重置**生成新 Key，原 Key 立即失效，需在工具中重新配置。 - -## **购买相关** - -### **可以同时购买多个套餐吗？** - -每个阿里云账号限购一个订阅，同一订阅下每种坐席类型均可购买多个。共享用量包可叠加购买，单次最多 1000 个。 - -### **可以单独购买共享用量包吗？** - -不可以。共享用量包是 Token Plan 团队版的附加商品，需先订阅 Token Plan 团队版坐席套餐后，才能购买共享用量包。 - -### **套餐是否支持退订？** - -支持按席位退订。在控制台**Token Plan 订阅详情页**的订阅明细中，点击对应席位的**退订**，已有用量消耗的席位不可退订。也可勾选多个席位后点击**批量退订**。退款原路退回支付账户，预计 1-3 个工作日到账。详见[订阅管理](https://help.aliyun.com/zh/model-studio/token-plan-overview#tp01-sub-mgmt)。 - -### **阿里云账号欠费是否影响 Token Plan 团队版的使用？** - -Token Plan 团队版为预付费订阅产品，只要套餐额度未用尽且订阅仍在有效期内，阿里云账号欠费不影响 Token Plan 团队版的正常使用。 - -### **如何加购坐席？** - -在 Token Plan 订阅详情页的订阅明细中，点击**加购座席**，选择坐席类型和数量后提交订单。加购后需将新坐席分配给成员才能使用。详见[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)。 - -### **如何关闭或开启自动续费？** - -在 Token Plan 订阅详情页的订阅明细中，点击**关闭自动续费**或**开启自动续费**。关闭后订阅到期不自动续费，需手动续费。 - -### **套餐变更与退订重购有哪些限制？** - -Token Plan 团队版与 Coding Plan 是两个独立的订阅计划，不支持相互转换： - -- **套餐互转限制**：不支持将已购买的 Token Plan 团队版更换为 Coding Plan，也不支持将 Coding Plan 直接转换为 Token Plan 团队版（即使补差价也不行）。可以同时订阅这两个计划，各自独立计费。 - -- **退订重购注意事项**：退款并重新购买后，API Key 和 Base URL 会发生变更，需在工具中重新配置新的专属 API Key 才能正常使用。 - - -## **计量相关** - -### **Credits 抵扣规则是什么？** - -Token Plan 团队版实际消耗取决于每次请求中输入 Token、缓存 Token 和输出 Token 的组合。优先从坐席额度抵扣，坐席额度用尽后从共享用量包抵扣，全部用尽后服务暂停至下一计费周期或购买共享用量包补充额度。 - -### **如何查看用量？** - -在[Token Plan 订阅详情页](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)可查看套餐和共享用量包的用量详情。管理员还可在[管理后台](https://tokenplan-enterprise.bailian.aliyunportal.com)的用量分析页面查看全部成员的消耗明细。 - -### **用量如何重置？** - -坐席额度在每个订阅月到期时重置，未用完的额度不累积到下月。共享用量包额度购买后有效期为 1 个月，到期后需重新购买，不随座席额度按月重置。 - -### **超出限额之后怎么办？** - -坐席额度用尽后自动从共享用量包抵扣；全部额度用尽后服务暂停。可通过以下方式恢复： - -- 购买共享用量包补充额度。 - -- 等待下一计费周期额度自动重置。 - - -### **续费后为什么 Credits 没有增加？** - -Token Plan 团队版的坐席额度按订阅周期计算，每个订阅月到期时自动重置。续费（续订）仅延长订阅有效期或预定下一计费周期的额度，**不会叠加补充至当前计费周期**。 - -若当前周期额度已用尽且需立即恢复服务，续订下月额度无法即时补充，可通过以下方式恢复： - -- 购买共享用量包补充额度。 - -- 升级至更高规格的坐席。 - -- 加购坐席（加购后需将新坐席分配给成员方可使用）。 - - -## **数据安全** - -### **数据安全如何保障？** - -Token Plan 团队版承诺不使用对话数据训练模型，传输过程采用 HTTPS 加密，并基于多租户隔离架构保障企业级数据隔离。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-overview.md index 8abf5759..9043d571 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-overview.md @@ -1,299 +1,187 @@ -# Token Plan（团队版）概述 +# Token Plan 概述 -Token Plan 团队版是阿里云百炼推出的 AI 大模型订阅服务，以 Credits 统一计量，支持文本生成与图像生成模型，兼容主流 AI 编程与智能体工具，提供团队管理后台、数据安全保障，调用平稳运行。 +Token Plan 是阿里云百炼推出的 AI 大模型订阅服务，以 Credits 统一计量，支持多种 AI 编程和智能体工具。Token Plan 提供个人版和团队版两个版本，满足从个人开发者到企业团队的不同需求。 **说明** -Token Plan 团队版目前仅支持**华北2（北京）**地域。 +Token Plan 目前仅支持**华北2（北京）**地域，请在[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan)左上角将地域切换至**华北2（北京）**后购买并使用。 ## **产品简介** -Token Plan 团队版整合千问和三方模型，支持文本生成与图像生成。通过 Credits 统一计量，同一订阅可在多种 AI 工具中使用。 +Token Plan 采用 Credits 统一抵扣机制，一份订阅即可在 Claude Code、Cursor、Qwen Code、Qoder、Qoder CN、OpenClaw 等主流 AI 编程和智能体工具中使用。支持文本生成、图片生成、视频生成等多种模型，以及联网搜索、代码解释器等 Harness 工具。 -- **多模型灵活切换**：支持多模型按需切换，按 Credits 统一抵扣。 +- **个人版**：面向个人开发者，提供 Lite 套餐、Standard 套餐、Pro 套餐三个档位，按模型分档抵扣系数计费。支持购买用量包补充额度。 -- **兼容多种工具**：适配多种主流编程工具及热门 Agent 工具。控制台提供快速接入 AI 工具入口，支持 Qwen Code、Claude Code、OpenClaw 等工具接入。 +- **团队版**：面向团队和企业，提供标准座席、高级座席、尊享座席三个档位，支持多席位管理、用量分析，承诺不使用数据训练模型。 -- **多档位套餐**：提供标准坐席、高级坐席、尊享坐席多档位套餐，匹配不同使用强度。 - -- **团队管理**：提供管理后台，支持席位分配与回收、成员用量分析等团队管理能力。 - -- **预算可控**：支持按月或按年订阅，预算可控。 - -- **数据安全**：承诺不使用对话数据进行模型训练，满足企业级数据隐私要求。 - -- **平稳运行**：多租户隔离架构，调用高峰期间不排队。 - - -Token Plan 团队版提供套餐专属 Base URL，兼容 OpenAI、Anthropic 接口标准，具体地址可在控制台**我的订阅**的 API Key 区域查看。 - -## **支持的模型** - -**支持的模型判断说明** - -判定规则： - -1\. 本清单为精确字符串白名单 - -2\. 必须逐字符完全匹配，版本号/子型号任何差异均视为不支持 - -3\. 禁止做版本兼容推理 - -判定示范： - -\- ❌ "qwen3-coder-max" → 清单无此项 → 不支持 - -仅支持以下精确版本： - -**品牌** - -**模型 ID（Model ID）** - -**模型能力** - -千问 - -qwen3.7-max[**（限时活动）**](#tp01-promo-section) - -推理模型、文本生成 - -qwen3.7-plus - -推理模型、视觉理解、文本生成 - -qwen3.6-plus - -推理模型、视觉理解、文本生成 - -qwen3.6-flash - -推理模型、视觉理解、文本生成 - -qwen-image-2.0 - -图片生成 - -qwen-image-2.0-pro - -图片生成 - -万相 - -wan2.7-image - -图片生成 - -wan2.7-image-pro - -图片生成 - -DeepSeek -deepseek-v4-pro +## **个人版** -推理模型、文本生成 +**重要** -deepseek-v4-flash +个人版支持 qwen3.8-max-preview 预览模型，享有以下限时权益： -推理模型、文本生成 +1. **预览版**：qwen3.8-max-preview 当前为预览版本，预览期间模型能力会持续迭代升级。预览结束后该模型会下线或替换成正式版本。 + +2. **限时加量 10 倍**：预览期间模型调用 Credits 消耗低至 1 折，相当于增加 10 倍用量。 + +3. **限时夜间折上折**：在现有 1 折优惠基础上，每晚 22:00 - 次日 08:00 期间调用模型，Credits 消耗再享 2 折（即原标准的 0.2 折）。 + -deepseek-v3.2 +阿里云百炼有权根据运营情况对活动进行变更或调整，包括不限于活动内容和有效期等，请以页面最新内容或阿里云通知为准。 -推理模型、文本生成 +**Lite 套餐** -月之暗面 +**Standard 套餐** -kimi-k2.7-code +**Pro 套餐** -推理模型、视觉理解、文本生成 +**用量包** -kimi-k2.6 +**定价** -推理模型、视觉理解、文本生成 +原价 60 元/月 +限时 **39 元/月** -kimi-k2.5 +原价 180 元/月 +限时 **139 元/月** -推理模型、视觉理解、文本生成 +原价 600 元/月 +限时 **499 元/月** -智谱 AI +100 元/个/月 +20,000 Credits/个 -glm-5.2 +**5 小时限额** -推理模型、文本生成 +700 Credits -glm-5.1 +3,000 Credits -推理模型、文本生成 +12,000 Credits -glm-5 +无限制 -推理模型、文本生成 +**每 7 天限额** -MiniMax +2,500 Credits -MiniMax-M2.5 +10,000 Credits -推理模型、文本生成 +40,000 Credits -## **套餐与定价** +无限制 -前往 [Token Plan 团队版购买页面](https://common-buy.aliyun.com/token-plan/)选择坐席类型、数量和订阅周期，完成订阅。主账号和 RAM 账号均可订阅。 +**并发 Agent** -订阅周期支持按月购买、按年购买、连续包月包年。 +1-2 个 -### **限时活动** +3-4 个 -即日起至 2026 年 7 月 22 日 23:59（UTC+8），qwen3.7-max 模型 Credits 消耗减半，同时支持隐式缓存。 +6-8 个 -### **Token Plan 团队版** +\- -提供标准坐席、高级坐席、尊享坐席三个档位，匹配不同使用强度。 +**购买条件** -席位（坐席）是 Token Plan 团队版的最小订阅单位，代表一个团队成员的使用名额。管理员在[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)中将席位分配给成员后，系统自动为该成员生成专属的 API Key。每个席位绑定一个成员、对应一个 API Key，不可共享。 +\- -**坐席类型** +需有效订阅，最多 5 个 -**价格** +**模型** -**额度** +支持文本生成、图像生成、视频生成等多种模型（[查看完整列表](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview#tpp01-models)） -**适用场景** +**Harness 工具** -标准坐席 +支持多种 Harness 工具，以 Credits 统一抵扣（[查看详情](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview#tpp01-harness)） -¥198/坐席/月 +## **团队版** -25,000 Credits/坐席/月 +**重要** -轻度使用 AI 辅助的团队成员 +团队版支持 qwen3.8-max-preview 预览模型，享有以下限时权益： -高级坐席 +1. **预览版**：qwen3.8-max-preview 当前为预览版本，预览期间模型能力会持续迭代升级。预览结束后该模型会下线或替换成正式版本。 + +2. **限时加量 10 倍**：预览期间模型调用 Credits 消耗低至 1 折，相当于增加 10 倍用量。 + -¥698/坐席/月 +阿里云百炼有权根据运营情况对活动进行变更或调整，包括不限于活动内容和有效期等，请以页面最新内容或阿里云通知为准。 -100,000 Credits/坐席/月 +**标准座席 Standard** -日常高频使用 AI 编程或办公的团队成员 +**高级座席 Pro** -尊享坐席 +**尊享座席 Max** -¥1,398/坐席/月 +**共享用量包 Extra Bundle** -250,000 Credits/坐席/月 +**定价** -重度依赖 AI 的核心开发者或高强度使用者 +原价 198 元/座席/月 +限时 **150 元/座席/月** -### Token Plan 团队版 - 共享用量包 +原价 698 元/座席/月 +限时 **550 元/座席/月** -跨坐席共享的弹性用量包，当个别坐席用量超出套餐额度时，可从共享用量包中抵扣。每个共享用量包有效期为 1 个月，到期未使用的额度自动清零。持有多个共享用量包时，优先抵扣最近到期的用量包。 +**1,398 元/座席/月** -**档位** +5,000 元/个/月 -**价格** +**每月总额度** -**额度** +25,000 Credits/座席/月 -Token Plan 团队版 - 共享用量包 +100,000 Credits/座席/月 -¥5,000/个 +250,000 Credits/座席/月 625,000 Credits/个 -## **订阅管理** - -在[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)的**我的订阅**页面管理订阅： - -- **加购席位**：点击**加购座席**，新加席位与现有订阅统一到期，费用和 Credits 额度均按剩余时长折算。 - -- **升级席位**：在**订阅明细**中点击席位的**升级**，将低档位席位升级为高档位，按差价补缴费用。升级当月的 Credits 额度按**高低档位差值**、依本计费周期的剩余时长折算后补充到当月可用额度；自下一计费周期起按新档位的完整额度发放。实际到账额度以控制台**我的订阅**页面为准。 - -- **退订席位**：在**订阅明细**中点击席位的**退订**，按席位维度退订；已有用量消耗的席位不可退订。退款原路退回支付账户，预计 1-3 个工作日到账。 - -- **批量操作**：勾选多个席位后点击**批量升级**或**批量退订**。 - -- **续费**：点击**续费**按钮，续费周期与订阅时一致（按月订阅则按月续费，按年订阅则按年续费），到期前完成可避免服务中断。 - -- **自动续费**：点击**开启自动续费**，确认后次日生效，到期前 9 天系统按订阅周期自动扣款续费。如需关闭，点击**关闭自动续费**并确认，关闭后停止到期前自动扣款续费，订阅到期后自动停订。 - -- **加购共享用量包**：在**共享用量包**区域点击**前往购买**。 - - -## **Credits 计费机制** - -### **计费说明** - -单次请求消耗的 Credits **并非固定值**，由模型类型、Token 用量、思考模式及工具调用等动态决定。其中 Token 用量会随多轮对话累积的上下文（历史消息、代码、工具返回、检索内容等）持续增长，且部分模型按**上下文长度阶梯计费**（上下文越长，单价档位可能越高），因此同一模型在不同请求下的消耗可能相差较大。实际消耗以[控制台订阅页用量明细](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan)为准。 - -### **计算示例** +**5 小时限额** -以 qwen3.6-plus 为例，预估单次请求消耗明细如下（不同模型的单价不同，实际以账单为准）： +无限制 -**Token 类型** +**7 天限额** -**数量** +无限制 -**消耗 Credits** +**模型** -输入 tokens +支持文本生成、图像生成、视频生成等多种模型（[查看完整列表](https://help.aliyun.com/zh/model-studio/token-plan-team-overview#tpt01-models)） -8,349 +**Harness 工具** -1.67 +支持多种 Harness 工具，以 Credits 统一抵扣（[查看详情](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview#tpp01-harness)） -缓存 tokens +**团队管理** -40,794 +支持多席位管理和用量分析（[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team-management)） -0.82 +## **套餐限额** -输出 tokens +### **个人版** -573 +个人版采用 5 小时和 7 天两层固定窗口限额，限额单位为 Credits： -0.69 - -**合计** - -**约 3.18 Credits** - -**说明** - -上表**仅为单次请求的示例**，并不代表每次请求都固定消耗约 3 Credits。在 AI 编程、智能体等**多轮对话**场景中，每次请求都会携带累积的上下文（历史对话、代码、工具返回等），随着对话进行输入 Token 持续增多，单次消耗的 Credits 也会**相应上升**；若模型按上下文长度阶梯计费，长上下文可能进入更高价位档，消耗速度进一步加快。 - -如需控制消耗，建议：任务切换或话题变更时**及时开启新会话**、清理无关历史，以缩短上下文；对长文档、大代码库按需拆分输入；并在上述控制台订阅页用量明细中关注各模型的实时消耗趋势。 - -**说明** - -上表以 qwen3.6-plus 展示 Token 类型分布，其中**“缓存 tokens”一行为通用 Token 分类示意**。**Token Plan 团队版中，隐式缓存的命中取决于所选模型是否支持**——目前仅 qwen3.7-max 在限时活动期间支持隐式缓存（详见上文[限时活动](#tp01-promo-section)章节）；使用其他模型时，输入 Token 不会进入隐式缓存抵扣，此时“缓存 tokens”一行不适用。 - -### **抵扣顺序** - -1. 优先从坐席套餐的月度额度中抵扣。 - -2. 坐席额度用尽后，从共享用量包中抵扣。持有多个共享用量包时，优先抵扣最近到期的用量包。 +- **5 小时限额**：自首次调用起开启 5 小时计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 5 小时后额度重置。 -3. 全部额度用尽后，服务将暂停至下一计费周期或购买共享用量包补充额度。 +- **7 天限额**：自首次调用起开启 7 天计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 7 天后额度重置。 -## **查看额度消耗情况** +任一层限额触顶即暂停服务，可购买用量包补充额度继续使用，或等待对应窗口周期结束后额度重置。窗口期内未用完的额度不结转至下一周期。 -**通过控制台**：登录[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)，在**我的订阅**页面查看总额度使用百分比、重置时间、团队席位分配情况，以及各席位与共享用量包的状态和到期时间。 +### **团队版** -**通过团队管理平台**：在**用量分析**页面，可查看近 1、7、30 天的 Credits 消耗趋势、各模型用量，以及每个成员的消耗明细。详见[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)。 +团队版采用月度总额度制，无 5 小时和 7 天窗口限额。每个座席的月度额度在计费周期内可用，到期未使用的额度不结转。超出月度总额度后调用将被阻断，可购买共享用量包补充额度。 -## **使用细则** +## **常见问题** -1. **使用范围**：仅限在兼容的 AI 编程和智能体工具中交互式使用，不可用于自动化脚本或应用后端。违规使用可能导致订阅暂停或 API Key 封禁。 - -2. **数据安全**：Token Plan 团队版不会使用对话数据训练模型。 - -3. **账号规范**：API Key 仅限已分配席位的成员本人使用，不可共享或公开泄露。 - -4. **退订与退款**：在控制台**我的订阅**页面按席位退订，已有用量消耗的席位不可退订。退款原路退回支付账户，预计 1-3 个工作日到账。 - -5. **服务地域**：Token Plan 团队版目前仅在特定地域提供服务，如需从海外调用，请确认符合当地法律法规要求。关于百炼支持的地域和服务部署范围，请参见[选择地域和服务部署范围](https://help.aliyun.com/zh/model-studio/regions/)。 - +### **个人版和团队版可以同时购买吗？** + +可以。同一阿里云账号可以同时持有个人版和团队版，各自独立计费。 -## 错误码 +### **关于 Coding Plan** -如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 +Coding Plan 和 Token Plan 是两个独立的订阅产品，两者之间无法迁移或升级。Coding Plan Lite 已于 2026 年 3 月 20 日停止新购，2026 年 4 月 13 日停止续费和升级；Coding Plan Pro 为限量抢购，库存售罄后不再补充。推荐使用 **Token Plan**，支持更多模型和 Harness 工具。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md new file mode 100644 index 00000000..2e703bf7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md @@ -0,0 +1,186 @@ +# 常见问题 + +Token Plan 个人版的额度、购买、订阅和接入常见问题。 + +## **额度与限额** + +### 5 小时限额和 7 天限额是什么意思？ + +Token Plan 个人版采用 5 小时和每 7 天两层固定窗口限额，限额单位为 Credits。两层限额独立计算，任一层触顶即暂停服务，需等待对应窗口周期结束后额度重置。窗口期内未用完的额度不结转至下一周期。 + +- **5 小时限额**：自首次调用起开启 5 小时计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 5 小时后额度重置。 + + 例如：Standard 套餐的 5 小时限额为 3,000 Credits。您在 7 月 20 日 10:00 首次调用，系统开启窗口（10:00 ~ 15:00）。10:00 消耗 2,000 Credits，11:00 消耗 1,000 Credits，累计 3,000 Credits 触顶暂停。需等到 15:00 窗口结束，额度重置为 3,000 Credits，服务恢复。 + +- **7 天限额**：自首次调用起开启 7 天计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 7 天后额度重置。 + + 例如：Standard 套餐的 7 天限额为 10,000 Credits。您在 7 月 20 日首次调用，系统开启窗口（7 月 20 日 ~ 7 月 27 日）。7 月 20 日消耗 4,000 Credits，7 月 22 日消耗 6,000 Credits，累计 10,000 Credits 触顶暂停。需等到 7 月 27 日窗口结束，额度重置为 10,000 Credits，服务恢复。 + + +各档位限额如下： + +**档位** + +**5 小时限额** + +**7 天限额** + +Lite 套餐 + +700 Credits + +2,500 Credits + +Standard 套餐 + +3,000 Credits + +10,000 Credits + +Pro 套餐 + +12,000 Credits + +40,000 Credits + +### 5 小时限额到了但7 天限额还有余量，还能继续使用吗？ + +不能。任一层限额触顶即暂停服务，可购买用量包补充额度、使用额度重置功能，或等待对应窗口周期结束后额度自动重置。 + +### 7 天限额是固定日期重置吗？ + +不是。7 天限额采用固定窗口机制，自首次调用起计时 7 天，到期后额度重置。重置时间取决于您首次调用的时间，而非固定的日历日期（如每周一）。 + +### 额度用完了怎么办？ + +限额用完后调用会被阻断，不会按量计费。恢复方式： + +- 等待额度释放。 + +- 升级套餐。 + +- 购买用量包，获得不受 5 小时和 7 天限额约束的额外额度。 + + +### 用量包和套餐是什么关系？需要先买套餐吗？ + +用量包是套餐的补充，提供不受套餐限额约束的额外 Credits，需先订阅有效套餐后才能购买，最多同时持有 5 个。 + +### 用量包的额度有 5 小时/周限制吗？ + +没有。用量包额度不受套餐的 5 小时和 7 天窗口限额约束，购买后即可使用。 + +### 用量包有效期多久？ + +用量包有效期为 1 个月，到期后未使用的额度自动作废，不支持退款。 + +## **接入报错** + +### 常见报错及解决方案 + +**报错信息** + +**可能原因** + +**解决方案** + +401 InvalidApiKey: No API-key provided. + +请求头中未携带 API Key + +生成 API Key 并在工具中完成配置 + +401 InvalidApiKey: Invalid API-key provided. + +误用了按量计费的 API Key 或 Coding Plan 的 Key；订阅过期；Key 复制不完整 + +确认使用 Token Plan 个人版专属 API Key，确保完整且无空格 + +404 model 'xxx' not found or not supported + +模型名称拼写错误或不在支持列表 + +确认模型名称区分大小写，与套餐支持的模型 ID 一致。 + +401 invalid access token or token expired + +误用了 Coding Plan 或其他计费模式的 Base URL + +使用 Token Plan 个人版专属 Base URL + +401 Incorrect API key provided + +误用了百炼通用 Base URL（dashscope.aliyuncs.com） + +使用 Token Plan 个人版专属 Base URL + +429 Requests rate limit exceeded + +短时间内请求过于密集 + +等待一分钟后重试，降低请求频率 + +429 Allocated quota exceeded + +5 小时或7 天限额用尽 + +等待窗口释放额度 + +## **并发与性能** + +### 最多支持多少个 Agent 并发？ + +并发能力与套餐档位相关： + +**档位** + +**建议并发** + +Lite 套餐 + +可同时支持 1-2 个 Agent 并发运行 + +Standard 套餐 + +可同时支持 3-4 个 Agent 并发运行 + +Pro 套餐 + +可同时支持 6-8 个 Agent 并发运行 + +### 高峰期响应会变慢吗？ + +高峰期可能出现排队等待。如需更稳定的吞吐，可升级到更高档位或使用团队版。 + +## **使用规则** + +### "禁止 API 生产自动化调用"具体是什么意思？ + +Token Plan 个人版仅供个人通过官方指定工具（如 Cursor、Claude Code、Windsurf 等）进行交互式开发。不允许将 API Key 用于生产环境的自动化服务、批量脚本或后台定时任务等非交互场景。 + +### 多人共用一个账号可以吗？ + +不可以。Token Plan 个人版限单人使用，不允许多人共用同一账号或 API Key。如需多人协作，请使用 Token Plan 团队版。 + +## **购买与订阅** + +### RAM 用户可以使用 Token Plan 吗？ + +可以，需由主账号完成以下授权： + +1. 在 [RAM 控制台](https://ram.console.aliyun.com/)为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略，同时授予 `AliyunBSSReadOnlyAccess` 系统策略。 + +2. 在百炼控制台[账号管理](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)页面，为该 RAM 用户分配管理员或订阅套餐权限。 + + +### 可以升配吗？升配后额度怎么算？ + +支持从低档位升级到更高档位。升级按剩余时长补缴差价，升级后每 5 小时限额和每 7 天限额立即提升至新档位对应额度。 + +### 可以降配吗？ + +不支持降配。如需更换为更低档位，可在订阅到期后重新购买。 + +### 自动续费怎么取消？ + +登录[百炼控制台 Token Plan](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan) 页面，在订阅管理中关闭自动续费。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md new file mode 100644 index 00000000..5a8b7f3d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md @@ -0,0 +1,261 @@ +# 概述 + +[Token Plan 个人版](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)是面向个人开发者的 AI 大模型订阅服务，以 Credits 统一计量，支持文本、多模态模型及 Harness 工具，适配主流 AI 编程和智能体工具。 + +**说明** + +Token Plan 个人版目前仅支持**华北2（北京）**地域。 + +## **核心特性** + +- **Credits 统一计量**：通过 Credits 统一抵扣不同模型和 Harness 工具的费用。 + +- **多模态模型支持**：覆盖文本生成、推理、视觉理解、图片生成、语音合成、语音识别、视频生成等能力。 + +- **Harness 工具集成**：支持联网搜索、文搜图、图搜图、网页抓取、代码解释器。 + +- **用量包**：超出限额后可购买用量包，不受限制继续使用。 + +- **兼容多种工具**：适配 Claude Code、Cursor、Qwen Code、Qoder、Qoder CN、OpenClaw 等主流 AI 编程和智能体工具。 + + +## **套餐档位与定价** + +**Lite 套餐** + +**Standard 套餐** + +**Pro 套餐** + +**用量包** + +**定价** + +原价 60 元/月 +限时 **39 元/月** + +原价 180 元/月 +限时 **139 元/月** + +原价 600 元/月 +限时 **499 元/月** + +100 元/个/月 +20,000 Credits/个 + +**每 7 天限额** + +2,500 Credits + +10,000 Credits + +40,000 Credits + +无限制 + +**每 5 小时限额** + +700 Credits + +3,000 Credits + +12,000 Credits + +无限制 + +**并发 Agent** + +1-2 个 + +3-4 个 + +6-8 个 + +\- + +**权益** + +文本、视觉等多模态模型 + +联网搜索等 Harness 工具 + +适配主流工具并持续扩展 + +享受 Lite 套餐所有权益 + +4x Lite 套餐用量 + +享受 Standard 套餐所有权益 + +16x Lite 套餐用量 + +更高的并发上限 + +不受 5 小时和 7 天限额约束 + +需先订阅套餐后购买 + +最多同时持有 5 个 + +- **每 5 小时限额**：自首次调用起开启 5 小时计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 5 小时后额度重置。 + +- **每 7 天限额**：自首次调用起开启 7 天计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 7 天后额度重置。 + + +任一层限额触顶即暂停服务，可购买用量包补充额度继续使用，或等待对应窗口周期结束后额度重置。窗口期内未用完的额度不结转至下一周期。 + +## **Credits 计费机制** + +### **计费说明** + +单次消耗的 Credits 由模型类型、Token 用量、思考模式及工具调用等动态决定，实际消耗以[控制台订阅页](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)用量详情为准。 + +### **抵扣顺序** + +1. 每次调用消耗的 Credits 同时计入每 5 小时限额和每 7 天限额。 + +2. 任一层限额（5 小时或 7 天）触顶后，服务暂停，可购买用量包补充额度、使用额度重置功能，或等待对应窗口周期结束后额度自动重置。 + + +### **额度重置** + +可自主使用额度重置功能，将 5 小时/7 天限额归零重新计算。额度重置后，当前窗口内已消耗的 Credits 清零，限额从零开始重新累计。 + +## **支持的模型** + +**重要** + +个人版支持 qwen3.8-max-preview 预览模型，享有以下限时权益： + +1. **预览版**：qwen3.8-max-preview 当前为预览版本，**预览期间模型能力会持续迭代升级**。预览结束后该模型会下线或替换成正式版本。 + +2. **限时加量 10 倍**：限时活动期间，模型调用 Credits 消耗低至 1 折，相当于增加 10 倍用量。 + +3. **限时夜间折上折**：在现有 1 折优惠基础上，每晚 22:00 - 次日 08:00 期间调用模型，Credits 消耗再享 2 折（即原标准的 0.2 折）。 + + +阿里云百炼有权根据运营情况对活动进行变更或调整，包括不限于活动内容和有效期等，请以页面最新内容或阿里云通知为准。 + +**品牌** + +**模型 ID（Model ID）** + +**模型能力** + +千问 + +qwen3.8-max-preview + +推理模型、视觉理解、文本生成 + +qwen3.7-max + +推理模型、文本生成 + +qwen3.7-plus + +推理模型、视觉理解、文本生成 + +qwen3.6-flash + +推理模型、视觉理解、文本生成 + +qwen-audio-3.0-tts-plus + +语音合成 + +智谱 AI + +glm-5.2 + +推理模型、文本生成 + +DeepSeek + +deepseek-v4-pro + +推理模型、文本生成 + +万相 + +wan2.7-image + +图片生成 + +wan2.7-image-pro + +图片生成 + +HappyHorse + +happyhorse-1.1-i2v + +视频生成 + +happyhorse-1.1-t2v + +视频生成 + +happyhorse-1.1-r2v + +视频生成 + +## **支持的 Harness 工具** + +**工具能力** + +**工具名称** + +联网搜索 + +web\_search + +文搜图 + +t2i\_search + +图搜图 + +i2i\_search + +网页抓取 + +web\_extractor + +代码解释器 + +code\_interpreter + +## **订阅管理** + +### **升级** + +支持从低档位升级到更高档位。升级按剩余时长补缴差价，升级后每 5 小时限额和每 7 天限额立即提升至新档位对应额度。 + +### **续费** + +续费支持切换续费周期，也可一次续费多个周期。 + +- **手动续费**：支持选择不同续费时长，可续费多个周期。 + +- **自动续费**：开启后到期前系统自动扣款续费。 + + +续费仅延长订阅有效期，不会叠加补充至当前计费周期的额度。 + +### **其他** + +- 个人版暂不支持退订。 + +- 订阅到期后重新购买，API Key 会发生变更，需在工具中重新配置。新 API Key 可在控制台[**我的订阅**](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)页面的 API Key 区域获取。 + + +## **订阅前须知** + +1. **严禁 API 调用**：仅限在编程工具和智能体工具（如 Claude Code、Cursor、Qwen Code、Qoder、Qoder CN、OpenClaw 等）中使用，禁止以 API 调用的形式用于自动化脚本、自定义应用程序后端或任何非交互式批量调用场景。将套餐 API Key 用于允许范围之外的调用将被视为违规或滥用，可能会导致订阅被暂停或 API Key 被封禁。 + +2. **数据使用授权**：使用 Token Plan 个人版期间，模型输入以及模型生成的内容将用于服务改进与模型优化。停止使用 Token Plan 个人版服务可终止后续数据授权，但终止授权的范围不涵盖已授权使用的数据。详细条款请参见[阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html)第 5.2 条。 + +3. **账号使用规范**：套餐为订阅人专享使用，禁止共享。账号共享可能导致订阅权益受限。 + +4. **购买限制**：同一实名认证主体限购一份，可同时购买个人版和团队版。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md new file mode 100644 index 00000000..10116c46 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md @@ -0,0 +1,81 @@ +# 快速开始 + +三步完成 Token Plan 个人版订阅和接入：选择套餐、获取 API Key、配置 AI 工具。 + +## **步骤一：订阅 Token Plan 个人版** + +访问 [Token Plan 个人版购买页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)，选择套餐档位和订阅周期，完成订阅。 + +购买须知： + +- **RAM 用户授权**：RAM 用户使用 Token Plan 前，需由主账号完成以下授权： + + 1. 在 [RAM 控制台](https://ram.console.aliyun.com/)为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略，同时授予 `AliyunBSSReadOnlyAccess` 系统策略。 + + 2. 在百炼控制台[账号管理](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)页面，为该 RAM 用户分配管理员或订阅套餐权限。 + + +## **步骤二：获取 API Key 和 Base URL** + +- **API Key**：订阅完成后，在 Token Plan 控制台的**我的订阅**页面生成 API Key。API Key 仅在生成时完整显示一次，请立即复制并妥善保存。 + +- **Base URL**：根据 AI 工具支持的协议，选择对应的 Base URL。 + + +**协议** + +**Base URL** + +OpenAI 兼容 + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +Anthropic 兼容 + +`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` + +**重要** + +Token Plan 的 API Key 以 `sk-sp-` 开头，与百炼通用 API Key（`sk-` 开头）格式不同，两者不可混用。Token Plan、Coding Plan 和按量付费的 API Key 与 Base URL 完全隔离，必须配套使用。 + +## **步骤三：接入 AI 工具** + +将 API Key 和 Base URL 配置到 AI 工具中，即可开始使用。 + +[**OpenClaw**开源、自托管个人 AI 助手](https://help.aliyun.com/zh/model-studio/openclaw) + +[**Hermes Agent**开源 AI 代理框架，内置自学习循环](https://help.aliyun.com/zh/model-studio/hermes-agent) + +[**Claude Code**AI 终端编码助手，支持自然语言编程](https://help.aliyun.com/zh/model-studio/claude-code) + +[**OpenCode**开源 AI 编程代理工具](https://help.aliyun.com/zh/model-studio/opencode) + +[**Cursor**AI 原生代码编辑器](https://help.aliyun.com/zh/model-studio/cursor) + +[**Codex**OpenAI 推出的命令行编程工具](https://help.aliyun.com/zh/model-studio/codex) + +[**Qwen Code**开源命令行 AI 编码工具](https://help.aliyun.com/zh/model-studio/qwen-code) + +[**QwenPaw**开源个人 AI 助手，支持本地与云端部署](https://help.aliyun.com/zh/model-studio/qwenpaw) + +[**Cherry Studio**多模型桌面客户端](https://help.aliyun.com/zh/model-studio/cherry-studio) + +[**Chatbox**跨平台 AI 桌面客户端](https://help.aliyun.com/zh/model-studio/chatbox) + +[**Cline**VS Code 扩展，智能代码补全和调试](https://help.aliyun.com/zh/model-studio/cline) + +[**Qoder**面向真实软件开发的 Agentic 编码平台](https://help.aliyun.com/zh/model-studio/qoder-agent) + +[**Lingma**阿里云智能编码助手，提供独立 IDE](https://help.aliyun.com/zh/model-studio/lingma-agent) + +[**Kilo CLI**轻量高性能命令行编程工具](https://help.aliyun.com/zh/model-studio/kilo-cli) + +[··· **更多工具**其他编程工具](https://help.aliyun.com/zh/model-studio/more-tools) + +## **可选：接入多模态生成模型** + +Token Plan 个人版支持多模态生成模型（wan2.7-image、happyhorse-1.1-t2v 等）。多模态生成模型使用独立的接口，需要通过 AI 工具的 Skill 或扩展机制接入，详见[接入多模态生成模型](https://help.aliyun.com/zh/model-studio/token-plan-multimodal-gen)。 + +## **可选：接入 Harness 工具** + +通过 Harness 工具调用，模型可以在对话中调用联网搜索、文搜图、图搜图、网页抓取、代码解释器等扩展能力。当前仅 qwen3.7、qwen3.8 支持原生工具调用，通过 Responses API 直接调用，Harness 工具按抵扣系数消耗 Credits。详见[接入 Harness 工具](https://help.aliyun.com/zh/model-studio/token-plan-harness-tool)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md new file mode 100644 index 00000000..d4334f0f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md @@ -0,0 +1,211 @@ +# 常见问题 + +Token Plan 团队版的常见问题解答，包括产品选择、Credits 计费与额度、模型与工具兼容性、使用限制、购买续费与退订等。 + +## **产品定位与套餐选择** + +### **个人版和团队版有什么区别？我该买哪个？** + +**对比项** + +**个人版** + +**团队版** + +适用场景 + +个人开发者 + +团队/企业 + +额度机制 + +5 小时 + 7 天固定窗口限额 + +固定月额度 + +团队管理 + +不支持 + +席位分配与回收、成员用量分析、SSO 接入 + +数据安全 + +数据使用遵循服务协议 + +承诺不使用对话数据训练模型 + +高峰期性能 + +高峰期可能出现等待 + +多租户隔离，高峰期不排队 + +个人开发者、日常使用 AI 编程工具，选择个人版即可；团队需要多人协作、统一管理席位和用量、对数据安全有更高要求，选择团队版。 + +### **个人版和团队版能同时买吗？额度是分开算的还是共享的？** + +可以同时购买。同一阿里云账号可以同时持有个人版和团队版，各自独立计费，额度不共享。 + +## **Credits 计费与额度** + +### **团队版的额度机制是怎样的？** + +团队版采用固定月额度，无 5 小时/7 天窗口限制。各坐席类型的月度额度： + +- **标准坐席**：25,000 Credits/坐席/月 + +- **高级坐席**：100,000 Credits/坐席/月 + +- **尊享坐席**：250,000 Credits/坐席/月 + + +坐席额度在每个订阅月到期时重置，未用完的额度不累积到下月。 + +### **坐席额度用完了怎么办？** + +坐席额度用尽后调用会被阻断，不会按量计费。恢复方式： + +- 等待下一个订阅月额度自动重置。 + +- 购买共享用量包（625,000 Credits/个，有效期 1 个月），团队内全部成员共享使用。 + + +## **模型与工具兼容性** + +### **Token Plan 团队版支持哪些模型？** + +团队版支持文本生成、推理、视觉理解、图片生成和视频生成模型。具体模型列表和抵扣系数请参见控制台的模型列表页面。 + +### **支持 Cursor / Claude Code / Cline 等第三方工具吗？** + +支持。Token Plan 兼容 OpenAI 和 Anthropic 协议，任何支持自定义 Base URL 和 API Key 的工具均可接入，包括 Cursor、Claude Code、Qwen Code、Qoder、Qoder CN、Cline、OpenClaw、Cherry Studio、Chatbox 等。具体配置方法请参见快速开始中的接入 AI 工具部分。 + +### **团队版的 API Key 能用在个人版上吗？** + +不能。个人版和团队版各自生成独立的 API Key，不可混用。系统会根据 API Key 自动识别对应的套餐。 + +### **Harness 工具是什么？** + +Harness 工具是模型内置的扩展能力，包括联网搜索、文搜图、图搜图、网页抓取、代码解释器等。团队版支持 Harness 工具，调用时按工具抵扣系数消耗 Credits。当前仅 qwen3.7 和 qwen3.8 系列模型支持原生 Harness 工具调用。 + +## **使用限制** + +### **高峰期性能如何？** + +团队版基于多租户隔离架构，调用高峰期间不排队。 + +## **购买、续费与退订** + +### **支持 RAM 子账号购买吗？** + +支持。RAM 子账号使用前，需主账号完成以下授权： + +1. 在 RAM 控制台为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略。 + +2. 在百炼控制台账号管理页面，为该 RAM 用户分配管理员或订阅套餐权限。 + + +### **可以升配吗？升配后额度怎么算？** + +支持坐席升配。升配后立即生效，限额按新坐席类型执行。升配需补缴差价（按剩余天数折算）。 + +### **可以降配吗？** + +不支持降配。如需使用更低坐席类型，可在当前订阅到期后重新订阅。 + +### **自动续费怎么取消？** + +登录[费用中心 > 续费管理](https://usercenter2.aliyun.com/finance/renew-manage)，找到 Token Plan 团队版订单，关闭自动续费。 + +### **续费时可以更换订阅时长吗？** + +不可以。续费仅支持按原订阅时长续费。如需更换订阅时长，可在订阅到期后重新购买。 + +### **续费后需要重新配置 API Key 吗？** + +不需要。续费仅延长当前订阅的有效期，不会影响已分配席位的 API Key 和 Base URL，成员无需在工具中重新配置。仅当订阅到期后重新购买、或退订席位后重新购买时，系统才会生成新的 API Key 和 Base URL，此时需在工具中重新配置，可在控制台**我的订阅**页面的 API Key 区域获取。 + +### **限时优惠的计费规则是什么？** + +限时优惠仅适用于包月订阅的新购、续费和自动续费，包年订阅和升级坐席不参与。 + +加购坐席时按剩余时长折算费用，实际收费取折算金额与限时价中的较低值。 + +**示例**：标准坐席原价 ¥198/月，限时价 ¥150/月。 + +- 加购时按剩余时长折算为 ¥99，低于限时价，实际收取 ¥99。 + +- 加购时按剩余时长折算为 ¥165，高于限时价，实际收取 ¥150。 + + +## **与现有产品的关系** + +### **已有个人版，再买团队版会冲突吗？** + +不冲突。两者可以同时持有，各自独立计费。使用时根据 API Key 自动匹配对应套餐。 + +## **接入报错** + +### **常见报错及解决方案** + +**报错信息** + +**可能原因** + +**解决方案** + +401 InvalidApiKey: No API-key provided. + +请求头中未携带 API Key + +生成 API Key 并在工具中完成配置 + +401 InvalidApiKey: Invalid API-key provided. + +误用了按量计费的 API Key 或 Coding Plan 的 Key；订阅过期；Key 复制不完整 + +确认使用 Token Plan 团队版专属 API Key，确保完整且无空格 + +404 model 'xxx' not found or not supported + +模型名称拼写错误或不在支持列表 + +确认模型名称区分大小写，与套餐支持的模型 ID 一致 + +401 invalid access token or token expired + +误用了 Coding Plan 或其他计费模式的 Base URL + +使用 Token Plan 团队版专属 Base URL + +401 Incorrect API key provided + +误用了百炼通用 Base URL（dashscope.aliyuncs.com） + +使用 Token Plan 团队版专属 Base URL + +400 Range of input length should be \[1, xxx\] + +输入内容超出模型最大上下文长度 + +新建会话清空历史，或使用工具的上下文压缩命令 + +400 调用图像/视频生成模型报错（如 qwen-image-2.0、wan2.7-image） + +图像、视频生成模型使用独立接口，无法通过文本模型的 Base URL 直接调用 + +通过工具的 Skill、Slash Command 或 Agent 扩展机制接入，详见[接入多模态生成模型](https://help.aliyun.com/zh/document_detail/6546109.html) + +429 Requests rate limit exceeded + +短时间内请求过于密集，触发模型限流 + +等待一分钟后重试，降低请求频率 + +429 Allocated quota exceeded + +坐席月度额度用尽 + +购买共享用量包或等待下月额度重置 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md new file mode 100644 index 00000000..bc5280b9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md @@ -0,0 +1,183 @@ +# 团队管理 + +在 Token Plan 控制台或管理平台中添加和管理团队成员、分配和回收席位、监控 Credits 用量。 + +## **访问入口** + +**阿里云主账号或 RAM 用户**：登录 Token Plan 控制台，在**我的订阅**页面进行团队管理：在**团队版**卡片点击**设置**修改组织名称、登录方式（SSO/钉钉）等；在**订阅明细**区域点击**分配座席**进入成员管理。 + +**说明** + +RAM 用户使用 Token Plan 前，需由主账号完成以下授权： + +1. 在 RAM 控制台为该 RAM 用户授予以下系统策略：`AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理），用于使用 Token Plan，权限范围按实际需要选择；同时授予 `AliyunBSSFullAccess`（费用中心权限），用于查看**我的订阅**页面中主账号已订阅的 Token Plan 套餐与用量。缺少费用中心权限时，RAM 用户登录后**我的订阅**页面将显示为空。 + +2. 在百炼控制台**账号管理**页面，为该 RAM 用户分配**管理员**或**订阅套餐**权限。 + + +通过 **SSO 或钉钉**加入的成员：通过管理员分发的**管理平台地址**登录管理平台。管理平台地址可在**设置**页面的**基本信息**区域获取。 + +## **角色与权限** + +**角色** + +**权限** + +所有者 + +添加或移除成员、分配或回收席位、修改成员角色、查看全部成员和模型的用量 + +管理员 + +权限范围与所有者相同。管理员由所有者授予，可被移除或降级。 + +成员 + +使用管理员分配的 API Key 和 Base URL 调用模型 + +## **成员管理** + +### **添加成员** + +- **手动添加**（不能登录管理平台，仅供 API 调用）：在**成员管理**页面点**添加成员**，在弹窗中填用户名（仅支持英文字母、数字、下划线）和角色，可选择同时分配席位。在该成员的操作列点**分配席位**选择席位版本，分配后系统自动生成 API Key，连同 Base URL 发给成员即可调用模型。 + +- **SSO 或钉钉登录**（成员可登录管理平台，自管席位和 API Key）：完成 SAML 或钉钉接入配置后，成员从登录页对应入口登录即自动加入。 + + +### **修改成员角色** + +在**成员管理**页面，找到目标成员，点击操作列的**修改角色**，在弹窗中选择新角色（管理员或成员）后保存。所有者角色不可修改。 + +### **重置 API Key** + +在**成员管理**页面，找到目标成员，点击操作列的**重置**。重置后原 API Key 立即失效，新 API Key 需重新发给成员。 + +### **移出成员** + +在**成员管理**页面，找到目标成员，点击操作列的**移出组织**。移出后席位自动回收、API Key 立即失效。 + +## **SAML 接入** + +通过标准 SAML 2.0 对接企业 IdP，对应登录页的**SSO**入口。配置后，成员用 IdP 账号登录管理平台即自动加入组织。 + +### **SAML 配置** + +**前提条件**：组织内有成员时无法编辑 SSO 配置，需先全部移出。 + +1. 登录 Token Plan 控制台，在**我的订阅**页面的**团队版**卡片点击**设置**，找到**SSO 配置**区域。 + +2. 从企业 IdP 获取 IdP 信息（IdP Entity ID、IdP SSO URL、IdP Certificate）。点击**编辑**，填入自定义的 SP Entity ID 和上述 IdP 信息，保存。 + +3. 保存后系统自动生成 ACS URL。将 SP Entity ID 和 ACS URL 填入企业 IdP 的 SSO 应用配置中。 + +4. 在**基本信息**区域复制**管理平台地址**，分享给团队成员。成员访问该地址，在登录页选择 SSO 登录方式即可加入组织。 + + +**SP 信息**（百炼侧） + +**参数** + +**说明** + +SP Entity ID + +百炼在 SSO 流程中的唯一标识，自定义填写，需同步填入企业 IdP 的 SSO 应用配置。 + +ACS URL + +IdP 认证成功后回传响应的地址，保存 SSO 配置后由百炼自动生成，需填入企业 IdP。 + +SP Certificate + +百炼侧 SAML 签名证书，保存 SSO 配置后由系统自动生成，用于企业 IdP 验证百炼侧响应签名，需同步配置到企业 IdP 信任链。 + +**IdP 信息**（企业侧，需从企业 IdP 获取后填入） + +**参数** + +**说明** + +IdP Entity ID + +企业身份提供商的唯一标识。 + +IdP SSO URL + +企业 IdP 的登录入口地址，成员登录时跳转到此地址进行认证。 + +IdP Certificate + +企业 IdP 的签名证书，百炼用来验证响应确实来自该企业。 + +### **配置示例：阿里云 IDaaS** + +**前提条件**：已开通阿里云 IDaaS EIAM 实例。 + +1. **在 IDaaS 中创建 SAML 应用**：登录 IDaaS 实例管理平台，进入**应用管理**，点击**添加应用**，选择**标准 SAML 2.0**应用模板，填写应用名称后创建。 + +2. **从 IDaaS 获取 IdP 信息并填入百炼**：在 IDaaS 应用的**单点登录**页面底部的**应用配置信息**区域，复制 IdP 信息，填入 Token Plan 控制台的 SSO 配置区域。 + +3. **将 SP Entity ID 和 ACS URL 填入 IDaaS**：保存后百炼 SSO 配置区域显示自动生成的 ACS URL，在 IDaaS 应用的**登录访问** > **单点登录**页面填入对应参数。 + +4. **在 IDaaS 中创建账户并授权**：在 IDaaS 的**账户管理**中为团队成员创建账户，在 SAML 应用的**登录访问** > **应用授权**页面添加授权。 + +5. **分享管理平台地址给成员**：在**设置**页面的**基本信息**区域复制**管理平台地址**，成员访问该地址选择 SSO 登录即可加入组织。 + + +## **钉钉接入** + +使用钉钉作为身份系统的企业可直接接入，对应登录页的**钉钉**入口。配置后，成员用钉钉账号登录管理平台即自动加入组织。 + +**前提条件**：已在钉钉中创建企业，并将待登录的成员加入企业。 + +1. **创建钉钉企业内部应用**：登录钉钉开发者后台，进入**应用开发** > **企业内部应用** > **钉钉应用**，点击**创建应用**。 + +2. **获取应用凭证**：在应用详情页的**基础信息** > **凭证与基础信息**，记录 Client ID 和 Client Secret。 + +3. **配置回调域名**：在**开发配置** > **安全设置**，在**重定向 URL（回调域名）**中填入 `https://account-enterprise.bailian.aliyunportal.com/api/v1/auth/dingtalk/callback`。 + +4. **开通通讯录读权限**：在应用的**权限管理**页面，开通**通讯录个人信息读权限**。 + +5. **发布应用**：在**应用发布** > **版本管理与发布**页面，创建新版本并发布。 + +6. **在 Token Plan 管理平台填入凭证**：在 Token Plan 控制台**我的订阅**页面的**团队版**卡片点击**设置**，在**SSO 配置**区域切换到**钉钉**选项卡，填入配置名称、钉钉应用 AppKey 和 AppSecret。 + +7. **把管理平台地址分享给成员**：在**设置**页面的**基本信息**区域复制**管理平台地址**，成员访问该地址选择钉钉登录即可加入组织。 + + +## **席位操作** + +### **查看席位状态** + +在控制台**我的订阅**页面的**团队座席**区域查看各档位席位的已分配数量和总数；在**订阅明细**区域可查看每个席位的状态和到期时间。 + +### **分配席位** + +1. 在**成员管理**页面，找到目标成员，点击操作列的**分配席位**。 + +2. 在弹窗中选择席位版本，点击**确定**。 + + +分配后系统自动生成 API Key，连同 Base URL 发给成员即可调用模型。 + +### **回收席位** + +在**成员管理**页面，找到目标成员，点击操作列的**回收席位**，在确认弹窗中点**确定**。回收后席位转为未分配，原成员将无法使用席位的 Credits；重新分配后系统会为新成员生成新的 API Key。 + +### **加购席位** + +在 Token Plan 控制台的**我的订阅**页面，点击**加购座席**，选择席位档位和数量后提交订单。新加席位与现有订阅统一到期，费用按剩余时长折算。 + +### **升级席位** + +在 Token Plan 控制台的**我的订阅**页面，在**订阅明细**中找到目标席位，点击**升级**，选择更高档位后提交订单。需要批量操作时，勾选多个席位后点击**批量升级**。升级按剩余时长补缴差价。 + +## **用量分析** + +用量分析可从 Token Plan 控制台的我的订阅页面或管理平台进入。在**用量分析**页面，所有者可查看： + +- **用量趋势**：近 1、7、30 天的 Credits 消耗趋势图。 + +- **模型用量**：组织内各模型的 Credits 消耗。 + +- **成员用量**：各成员的 Credits 消耗。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md new file mode 100644 index 00000000..d747e619 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md @@ -0,0 +1,318 @@ +# 概述 + +Token Plan 团队版是阿里云百炼推出的 AI 大模型订阅服务，以 Credits 统一计量，支持文本生成、图像生成与视频生成模型，兼容主流 AI 编程与智能体工具，提供团队管理后台、数据安全保障，调用平稳运行。 + +**说明** + +Token Plan 团队版目前仅支持**华北2（北京）**地域。 + +## **产品简介** + +Token Plan 团队版整合千问和三方模型，支持文本生成、图像生成与视频生成。通过 Credits 统一计量，同一订阅可在多种 AI 工具中使用。 + +- **多模型灵活切换**：支持多模型按需切换，按 Credits 统一抵扣。 + +- **兼容多种工具**：适配多种主流编程工具及热门 Agent 工具。控制台提供快速接入 AI 工具入口，支持 Qwen Code、Claude Code、Qoder、Qoder CN、OpenClaw 等工具接入。 + +- **多档位套餐**：提供标准坐席、高级坐席、尊享坐席多档位套餐，匹配不同使用强度。 + +- **团队管理**：提供管理后台，支持席位分配与回收、成员用量分析等团队管理能力。 + +- **预算可控**：支持按月或按年订阅，预算可控。 + +- **数据安全**：承诺不使用对话数据进行模型训练，满足企业级数据隐私要求。 + +- **平稳运行**：多租户隔离架构，调用高峰期间不排队。 + + +Token Plan 团队版提供套餐专属 Base URL，兼容 OpenAI、Anthropic 接口标准，具体地址可在控制台**我的订阅**的 API Key 区域查看。 + +## **支持的模型** + +**重要** + +团队版支持 qwen3.8-max-preview 预览模型，享有以下限时权益： + +1. **预览版**：qwen3.8-max-preview 当前为预览版本，**预览期间模型能力会持续迭代升级**。预览结束后该模型会下线或替换成正式版本。 + +2. **限时加量 10 倍**：限时活动期间，模型调用 Credits 消耗低至 1 折，相当于增加 10 倍用量。 + + +阿里云百炼有权根据运营情况对活动进行变更或调整，包括不限于活动内容和有效期等，请以页面最新内容或阿里云通知为准。 + +**品牌** + +**模型 ID（Model ID）** + +**模型能力** + +千问 + +qwen3.8-max-preview + +推理模型、视觉理解、文本生成 + +qwen3.7-max + +推理模型、文本生成 + +qwen3.7-plus + +推理模型、视觉理解、文本生成 + +qwen3.6-plus + +推理模型、视觉理解、文本生成 + +qwen3.6-flash + +推理模型、视觉理解、文本生成 + +qwen-image-2.0 + +图片生成 + +qwen-image-2.0-pro + +图片生成 + +qwen-audio-3.0-tts-plus + +语音合成 + +万相 + +wan2.7-image + +图片生成 + +wan2.7-image-pro + +图片生成 + +DeepSeek + +deepseek-v4-pro + +推理模型、文本生成 + +deepseek-v4-flash + +推理模型、文本生成 + +deepseek-v3.2 + +推理模型、文本生成 + +月之暗面 + +kimi-k2.7-code + +推理模型、视觉理解、文本生成 + +kimi-k2.6 + +推理模型、视觉理解、文本生成 + +kimi-k2.5 + +推理模型、视觉理解、文本生成 + +智谱 AI + +glm-5.2 + +推理模型、文本生成 + +glm-5.1 + +推理模型、文本生成 + +glm-5 + +推理模型、文本生成 + +MiniMax + +MiniMax-M2.5 + +推理模型、文本生成 + +HappyHorse + +happyhorse-1.1-i2v + +视频生成 + +happyhorse-1.1-t2v + +视频生成 + +happyhorse-1.1-r2v + +视频生成 + +## **套餐与定价** + +前往 Token Plan 团队版购买页面选择坐席类型、数量和订阅周期，完成订阅。主账号和 RAM 账号均可订阅。订阅周期支持按月购买、按年购买、连续包月包年。 + +### **限时活动** + +即日起至 2026 年 7 月 22 日 23:59（UTC+8），qwen3.7-max 模型 Credits 消耗减半，同时支持隐式缓存。 + +### **Token Plan 团队版** + +提供标准坐席、高级坐席、尊享坐席三个档位，匹配不同使用强度。 + +席位（坐席）是 Token Plan 团队版的最小订阅单位，代表一个团队成员的使用名额。管理员在团队管理中将席位分配给成员后，系统自动为该成员生成专属的 API Key。每个席位绑定一个成员、对应一个 API Key，不可共享。 + +**坐席类型** + +**价格** + +**额度** + +**适用场景** + +标准坐席 + +原价 ¥198/坐席/月 +限时 **¥150/坐席/月** + +25,000 Credits/坐席/月 + +轻度使用 AI 辅助的团队成员 + +高级坐席 + +原价 ¥698/坐席/月 +限时 **¥550/坐席/月** + +100,000 Credits/坐席/月 + +日常高频使用 AI 编程或办公的团队成员 + +尊享坐席 + +¥1,398/坐席/月 + +250,000 Credits/坐席/月 + +重度依赖 AI 的核心开发者或高强度使用者 + +**重要** + +限时优惠仅适用于包月订阅的新购、续费和自动续费，包年订阅和升级坐席不参与。加购坐席时按剩余时长折算费用，实际收费取折算金额与限时价中的较低值。 + +### **Token Plan 团队版 - 共享用量包** + +跨坐席共享的弹性用量包，当个别坐席用量超出套餐额度时，可从共享用量包中抵扣。每个共享用量包有效期为 1 个月，到期未使用的额度自动清零。持有多个共享用量包时，优先抵扣最近到期的用量包。 + +**档位** + +**价格** + +**额度** + +Token Plan 团队版 - 共享用量包 + +¥5,000/个 + +625,000 Credits/个 + +## **Credits 计费机制** + +### **计费说明** + +单次消耗的 Credits 由模型类型、Token 用量、思考模式及工具调用等动态决定，实际消耗以控制台订阅页用量明细为准。 + +### **计算示例** + +以 qwen3.6-plus 为例，预估单次请求消耗明细如下（不同模型的单价不同，实际以账单为准）： + +**Token 类型** + +**数量** + +**消耗 Credits** + +输入 tokens + +8,349 + +1.67 + +缓存 tokens + +40,794 + +0.82 + +输出 tokens + +573 + +0.69 + +**合计** + +**约 3.18 Credits** + +### **抵扣顺序** + +1. 优先从坐席套餐的月度额度中抵扣。 + +2. 坐席额度用尽后，从共享用量包中抵扣。持有多个共享用量包时，优先抵扣最近到期的用量包。 + +3. 全部额度用尽后，服务将暂停至下一计费周期或购买共享用量包补充额度。 + + +## **查看额度消耗情况** + +**通过控制台**：登录 Token Plan 控制台，在**我的订阅**页面查看总额度使用百分比、重置时间、团队席位分配情况，以及各席位与共享用量包的状态和到期时间。 + +**通过团队管理平台**：在**用量分析**页面，可查看近 1、7、30 天的 Credits 消耗趋势、各模型用量，以及每个成员的消耗明细。详见团队管理。 + +## **订阅管理** + +### **加购席位** + +在 Token Plan 控制台的**我的订阅**页面，点击**加购座席**，选择席位档位和数量后提交订单。新加席位与现有订阅统一到期，费用和 Credits 额度均按剩余时长折算。 + +### **升级席位** + +在**订阅明细**中找到目标席位，点击**升级**，选择更高档位后提交订单。需要批量操作时，勾选多个席位后点击**批量升级**。升级按剩余时长补缴差价，升级新增的 Credits 额度同步按剩余时长折算后发放，实际可用额度以控制台展示为准。 + +### **续费** + +- 点击**续费**按钮，续费周期与订阅时一致（按月订阅则按月续费，按年订阅则按年续费），到期前完成可避免服务中断。 + +- **自动续费**：点击**开启自动续费**，确认后次日生效，到期前 9 天系统按订阅周期自动扣款续费。如需关闭，点击**关闭自动续费**并确认。 + + +### **退订席位** + +在**订阅明细**中点击席位的**退订**，按席位维度退订；已有用量消耗的席位不可退订。也可勾选多个席位后点击**批量退订**。退款原路退回支付账户，预计 1-3 个工作日到账。 + +### **加购共享用量包** + +在**共享用量包**区域点击**前往购买**。 + +### **注意事项** + +- 退订重购后，API Key 和 Base URL 会发生变更，需在工具中重新配置。新 API Key 和 Base URL 均可在控制台[**我的订阅**](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise)页面的 API Key 区域获取。 + +- 续费仅延长订阅有效期，不会叠加补充至当前计费周期的额度。 + +- 团队版与个人版可同时持有，各自独立计费。 + + +## **使用细则** + +1. **使用范围**：仅限在兼容的 AI 编程和智能体工具中交互式使用，不可用于自动化脚本或应用后端。违规使用可能导致订阅暂停或 API Key 封禁。 + +2. **数据安全**：Token Plan 团队版不会使用对话数据训练模型。 + +3. **账号规范**：API Key 仅限已分配席位的成员本人使用，不可共享或公开泄露。 + +4. **服务地域**：Token Plan 团队版目前仅在特定地域提供服务，如需从海外调用，请确认符合当地法律法规要求。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-quickstart.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md similarity index 64% rename from skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-quickstart.md rename to skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md index 4486ad53..8aa9ac0d 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-quickstart.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md @@ -4,16 +4,18 @@ ## **步骤一：订阅 Token Plan 团队版** -访问 [Token Plan 团队版购买页面](https://common-buy.aliyun.com/token-plan/)，选择坐席类型、数量和订阅周期（按月或按年）并完成订阅，主账号和 RAM 账号均可订阅。 +访问 [Token Plan 团队版购买页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)，选择套餐档位和订阅周期并完成订阅，主账号和 RAM 账号均可订阅。 -购买须知： - -- RAM 子账号订阅前，需主账号在 RAM 控制台授予 AliyunBailianFullAccess 权限。 +- **RAM 用户授权**：RAM 用户使用 Token Plan 前，需由主账号完成以下授权： + 1. 在 [RAM 控制台](https://ram.console.aliyun.com/)为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略，同时授予 `AliyunBSSReadOnlyAccess` 系统策略。 + + 2. 在百炼控制台[账号管理](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)页面，为该 RAM 用户分配管理员或订阅套餐权限。 + ## **步骤二：获取 API Key 和 Base URL** -- **API Key**：在 Token Plan 控制台的[成员管理页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)或管理平台创建成员账号，分配席位后，为成员生成 API Key。详见[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)。 +- **API Key**：在 Token Plan 控制台的成员管理页面或管理平台创建成员账号，分配席位后，为成员生成 API Key。详见团队管理。 API Key 使用须知： @@ -53,6 +55,8 @@ Token Plan、Coding Plan 和按量付费的 API Key 与 Base URL 完全隔离， ## **步骤三：接入 AI 工具** +将 API Key 和 Base URL 配置到 AI 工具中，即可开始使用。 + [**OpenClaw**开源、自托管个人 AI 助手](https://help.aliyun.com/zh/model-studio/openclaw) [**Hermes Agent**开源 AI 代理框架，内置自学习循环](https://help.aliyun.com/zh/model-studio/hermes-agent) @@ -83,17 +87,10 @@ Token Plan、Coding Plan 和按量付费的 API Key 与 Base URL 完全隔离， [··· **更多工具**其他编程工具](https://help.aliyun.com/zh/model-studio/more-tools) -## **可选：接入图像生成模型** +## **可选：接入多模态生成模型** -Token Plan 团队版支持图像生成模型（qwen-image-2.0、wan2.7-image 等）。图像生成模型使用独立的接口，需要通过 AI 工具的 Skill 或扩展机制接入。具体配置方法请参见[接入多模态生成模型](https://help.aliyun.com/zh/model-studio/token-plan-multimodal-gen)。 +Token Plan 团队版支持多模态生成模型（wan2.7-image、happyhorse-1.1-t2v 等）。多模态生成模型使用独立的接口，需要通过 AI 工具的 Skill 或扩展机制接入，详见[接入多模态生成模型](https://help.aliyun.com/zh/model-studio/token-plan-multimodal-gen)。 -## **可选：工具调用** - -通过接入工具调用，模型可以在对话中调用联网搜索、代码解释器等扩展能力。 - -- qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash：内置联网搜索、代码解释器、网页抓取、以图搜图、文搜图 5 个工具，通过 Responses API 直接调用。内置工具不额外收费，产生的 token 消耗统一从套餐 Credits 中抵扣。 - -- 其他模型：通过 MCP 服务接入工具。 - +## **可选：接入 Harness 工具** -详细说明请参见[工具调用](https://help.aliyun.com/zh/model-studio/token-plan-tool)。 +通过 Harness 工具调用，模型可以在对话中调用联网搜索、文搜图、图搜图、网页抓取、代码解释器等扩展能力。当前仅 qwen3.7、qwen3.8 支持原生工具调用，通过 Responses API 直接调用，Harness 工具按抵扣系数消耗 Credits。详见[接入 Harness 工具](https://help.aliyun.com/zh/model-studio/token-plan-harness-tool)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team.md deleted file mode 100644 index 9e83ccff..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team.md +++ /dev/null @@ -1,272 +0,0 @@ -# 团队管理 - -在 Token Plan 控制台或管理平台中添加和管理团队成员、分配和回收席位、监控 Credits 用量。 - -## **访问入口** - -**阿里云主账号或 RAM 用户**：登录[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)，在**我的订阅**页面进行团队管理：在**团队版**卡片点击**设置**修改组织名称、登录方式（SSO/钉钉）等；在**订阅明细**区域点击**分配座席**进入成员管理。 - -**说明** - -RAM 用户使用 Token Plan 前，需由主账号完成以下授权： - -1. 在[RAM 控制台](https://ram.console.aliyun.com/users)为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略，具体权限范围按实际需要选择。 - -2. 在百炼控制台[**账号管理**](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/user_management/user_management)页面，为该 RAM 用户分配**管理员**或**订阅套餐**权限。 - - -通过 **SSO 或钉钉**加入的成员：通过管理员分发的**管理平台地址**登录管理平台。管理平台地址可在**设置**页面的**基本信息**区域获取。 - -## **角色与权限** - -**角色** - -**权限** - -所有者 - -添加或移除成员、分配或回收席位、修改成员角色、查看全部成员和模型的用量 - -管理员 - -权限范围与所有者相同。管理员由所有者授予，可被移除或降级。 - -成员 - -使用管理员分配的 API Key 和 Base URL 调用模型 - -## **成员管理** - -### **添加成员** - -- **手动添加**（不能登录管理平台，仅供 API 调用）：在**成员管理**页面点**添加成员**，在弹窗中填用户名（仅支持英文字母、数字、下划线）和角色，可选择同时分配席位（默认不分配，需加购可点弹窗内加购更多席位）。在该成员的操作列点**分配席位**选择席位版本，分配后系统自动生成 API Key，连同 Base URL 发给成员即可调用模型。 - -- **SSO 或钉钉登录**（成员可登录管理平台，自管席位和 API Key）：完成 [SAML](#tp05-sso) 或[钉钉](#tp05-dingtalk)接入配置后，成员从登录页对应入口登录即自动加入。 - - -### **修改成员角色** - -在**成员管理**页面，找到目标成员，点击操作列的**修改角色**，在弹窗中选择新角色（管理员或成员）后保存。所有者角色不可修改。 - -### **重置 API Key** - -在**成员管理**页面，找到目标成员，点击操作列的**重置**。重置后原 API Key 立即失效，新 API Key 需重新发给成员。 - -### **移出成员** - -在**成员管理**页面，找到目标成员，点击操作列的**移出组织**。移出后席位自动回收、API Key 立即失效。 - -## **SAML 接入** - -通过标准 SAML 2.0 对接企业 IdP，对应登录页的**SSO**入口。配置后，成员用 IdP 账号登录管理平台即自动加入组织。 - -### **SAML 配置** - -**前提条件**：组织内有成员时无法编辑 SSO 配置，需先全部移出。 - -1. 登录 Token Plan 控制台，在**我的订阅**页面的**团队版**卡片点击**设置**，找到**SSO 配置**区域。 - -2. 从企业 IdP 获取 IdP 信息（IdP Entity ID、IdP SSO URL、IdP Certificate）。点击**编辑**，填入自定义的 SP Entity ID 和上述 IdP 信息，保存。 - -3. 保存后系统自动生成 ACS URL。将 SP Entity ID 和 ACS URL 填入企业 IdP 的 SSO 应用配置中。 - -4. 在**基本信息**区域复制**管理平台地址**，分享给团队成员。成员访问该地址，在登录页选择 SSO 登录方式即可加入组织。 - - -**参数说明** - -**SP 信息**（百炼侧） - -**参数** - -**说明** - -SP Entity ID - -百炼在 SSO 流程中的唯一标识，自定义填写，需同步填入企业 IdP 的 SSO 应用配置。 - -ACS URL - -IdP 认证成功后回传响应的地址，保存 SSO 配置后由百炼自动生成，需填入企业 IdP。 - -SP Certificate - -百炼侧 SAML 签名证书，保存 SSO 配置后由系统自动生成，用于企业 IdP 验证百炼侧响应签名，需同步配置到企业 IdP 信任链。 - -**IdP 信息**（企业侧，需从企业 IdP 获取后填入） - -**参数** - -**说明** - -IdP Entity ID - -企业身份提供商的唯一标识（Entity ID）。 - -IdP SSO URL - -企业 IdP 的登录入口地址，成员登录时跳转到此地址进行认证。 - -IdP Certificate - -企业 IdP 的签名证书，百炼用来验证响应确实来自该企业。 - -### **配置示例：**[**阿里云 IDaaS**](https://www.aliyun.com/product/idaas) - -**前提条件**：已开通阿里云 IDaaS EIAM 实例。如未开通，登录[IDaaS 控制台](https://yundun.console.aliyun.com/?p=idaas)，在 EIAM 云身份服务实例列表页点击**免费创建实例**。 - -1. **在 IDaaS 中创建 SAML 应用** - - 登录 IDaaS 实例管理平台，进入**应用管理**，点击**添加应用**，选择**标准 SAML 2.0**应用模板，填写应用名称后创建。 - -2. **从 IDaaS 获取 IdP 信息并填入百炼** - - 在 IDaaS 应用的**单点登录**页面底部的**应用配置信息**区域，复制 IdP 信息。然后在 Token Plan 控制台**我的订阅**页面的**团队版**卡片点击**设置**，在 **SSO 配置**区域，点击**编辑**，填入自定义的 SP Entity ID 和以下对应的 IdP 信息后保存： - - **IDaaS 应用配置信息** - - **填入百炼的字段** - - **说明** - - IdP 唯一标识（IdP Entity ID） - - IdP Entity ID - - IDaaS 在应用中的唯一标识 - - IdP SSO 地址（IdP Sign-in URL） - - IdP SSO URL - - 成员 SSO 登录时跳转到的认证地址 - - 公钥证书（Certificate） - - IdP Certificate - - 点击“复制证书内容”获取完整证书 - -3. **将 SP Entity ID 和 ACS URL 填入 IDaaS** - - 保存后，百炼 SSO 配置区域显示自动生成的 ACS URL。在 IDaaS 应用的**登录访问** > **单点登录**页面，填入以下对应参数： - - **百炼 SP 信息** - - **填入 IDaaS 的字段** - - **说明** - - SP Entity ID - - 应用唯一标识（SP Entity ID） - - 填入在百炼中自定义的 SP Entity ID - - ACS URL - - 单点登录地址（ACS URL） - - 填入百炼自动生成的 ACS URL - -4. **在 IDaaS 中创建账户并授权** - - 在 IDaaS 实例管理平台的**账户管理**中，为需要使用 Token Plan 的团队成员创建 IDaaS 账户。 - - 然后在 SAML 应用的**登录访问** > **应用授权**页面，点击**添加授权**，选择需要通过 SSO 登录的账户、组或组织机构。后续添加用户重复此步骤即可，无需重新配置 SSO 参数。 - -5. **分享管理平台地址给成员** - - 在**设置**页面的**基本信息**区域复制**管理平台地址**，分享给已授权的成员。成员访问该地址，在登录页选择 SSO 登录方式即可加入组织。 - - -## **钉钉接入** - -使用钉钉作为身份系统的企业可直接接入，对应登录页的**钉钉**入口。配置后，成员用钉钉账号登录管理平台即自动加入组织。 - -**前提条件**：已在钉钉中创建企业，并将待登录的成员加入企业。 - -1. **创建钉钉企业内部应用** - - 登录[钉钉开发者后台](https://open-dev.dingtalk.com/)，进入**应用开发** > **企业内部应用** > **钉钉应用**，点击**创建应用**。填写应用名称和描述后保存。 - -2. **获取应用凭证** - - 在应用详情页，进入**基础信息** > **凭证与基础信息**，记录 Client ID 和 Client Secret（即钉钉应用 AppKey 与 AppSecret），后续在 Token Plan 管理平台中使用。 - -3. **配置回调域名** - - 在应用详情页，进入**开发配置** > **安全设置**，在**重定向 URL（回调域名）**中填入 `https://account-enterprise.bailian.aliyunportal.com/api/v1/auth/dingtalk/callback`，保存。 - -4. **开通通讯录读权限** - - 进入应用的**权限管理**页面，开通**通讯录个人信息读权限**。该权限用于在登录时识别成员的钉钉账号。 - -5. **发布应用** - - 进入**应用发布** > **版本管理与发布**页面，创建新版本并发布。 - -6. **在 Token Plan 管理平台填入凭证** - - 在 Token Plan 控制台**我的订阅**页面的**团队版**卡片点击**设置**，在**SSO 配置**区域，切换到**钉钉**选项卡，填入配置名称、**钉钉应用 AppKey**和**钉钉应用 AppSecret**（步骤 2 中获取），保存。 - -7. **把管理平台地址分享给成员** - - 在**设置**页面的**基本信息**区域复制**管理平台地址**，分享给团队成员。成员访问该地址，在登录页选择钉钉登录方式即可加入组织。 - - -## **席位操作** - -### **查看席位状态** - -在控制台**我的订阅**页面的**团队座席**区域查看各档位席位的已分配数量和总数；在**订阅明细**区域可查看每个席位的状态和到期时间。 - -### **分配席位** - -1. 在**成员管理**页面，找到目标成员，点击操作列的**分配席位**。 - -2. 在弹窗中选择席位版本，点击**确定**。 - - -分配后系统自动生成 API Key，连同 Base URL 发给成员即可调用模型。 - -### **回收席位** - -在**成员管理**页面，找到目标成员，点击操作列的**回收席位**，在确认弹窗中点**确定**。回收后席位转为未分配，原成员将无法使用席位的 Credits；重新分配后系统会为新成员生成新的 API Key。 - -### **加购席位** - -在[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)的**我的订阅**页面，点击**加购座席**，选择席位档位和数量后提交订单。 - -新加席位与现有订阅统一到期，费用按剩余时长折算。例如订阅周期剩余 15 天，加购一个标准席位，则只需支付 15 天对应的费用。 - -### **升级席位** - -在[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)的**我的订阅**页面，在**订阅明细**中找到目标席位，点击**升级**，选择更高档位后提交订单。需要批量操作时，勾选多个席位后点击**批量升级**。 - -升级按剩余时长补缴差价。例如将标准席位升级为高级席位，只需补缴剩余时长内两个档位的差价。 - -## **用量分析** - -用量分析可从 Token Plan 控制台的我的订阅页面或管理平台进入。在**用量分析**页面，所有者可查看： - -- **用量趋势**：近 1、7、30 天的 Credits 消耗趋势图。 - -- **模型用量**：组织内各模型的 Credits 消耗。 - -- **成员用量**：各成员的 Credits 消耗。 - - -## **常见问题** - -### **席位到期后 API Key 是否仍可使用** - -席位到期后，对应的 API Key 将无法调用模型。续订套餐并重新分配席位后即可恢复使用。 - -### **回收席位后能否分配给其他成员** - -可以。回收后席位转为未分配，重新分配给其他成员后系统会生成新的 API Key。 - -### **能否为同一成员更换席位** - -每个成员同一时间只能持有一个席位。如需更换，先回收当前席位，再分配新的席位即可。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/model-training-best-practices.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/model-training-best-practices.md index 9c756531..727165e7 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/model-training-best-practices.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/model-training-best-practices.md @@ -36,7 +36,7 @@ 在阿里云百炼，**完成调优的模型必须部署后才能调用和评测**。因此，您需要首先完成模型部署，方可继续评测模型。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3946933871/CAEQSBiBgICa8texgxkiIGU5OGM2ZmE4ZjJjNDRhMWY5MmMwYmQwZmIzZWIyODM04430647_20240626141408.264.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6543415871/CAEQSBiBgICa8texgxkiIGU5OGM2ZmE4ZjJjNDRhMWY5MmMwYmQwZmIzZWIyODM04430647_20240626141408.264.svg) ## **前提条件** @@ -149,7 +149,7 @@ 2. **数据集类型**：包含**训练集**和**评测集**。每次新增数据集时，您只能选择一种类型的数据集。**训练集**由一轮或多轮的Prompt+Completion数据组成，而**评测集**仅包含Prompt数据。您需要分别创建这两种数据集。 - 3. **存储位置**：默认为**平台存储**。 + 3. **存储位置**：默认为**平台OSS存储**。 4. **导入方式**：默认为**本地上传**。 @@ -241,11 +241,11 @@ ### **新建训练任务** -在[模型调优](https://bailian.console.aliyun.com/#/efm/model_manager)页面，点击**训练新模型**，即可设置训练参数。具体操作如下： +在[模型调优](https://bailian.console.aliyun.com/#/efm/model_manager)页面，点击**创建训练任务**，即可设置训练参数。具体操作如下： 1. 前往**[模型调优](https://bailian.console.aliyun.com/#/efm/model_manager)**页面。这里展示了所有调优任务。 -2. 点击**训练新模型**，阿里云百炼将会引导您配置训练参数： +2. 点击**创建训练任务**，阿里云百炼将会引导您配置训练参数： 1. **选择模型训练方式**：阿里云百炼支持**SFT微调训练**、**DPO偏好训练**和**CPT继续预训练**三种方式，以下是选择建议： diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md index f13ea9b6..c4e229b1 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md @@ -31,7 +31,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -92,7 +93,8 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); let reasoningContent = ''; // 完整思考过程 @@ -161,7 +163,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -222,7 +225,8 @@ import os client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -240,7 +244,8 @@ import OpenAI from "openai"; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1", + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", }); const completion = await openai.chat.completions.create({ @@ -254,7 +259,8 @@ console.log(completion.choices[0].message.content); ## **curl** ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -280,7 +286,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -326,7 +333,8 @@ import process from 'process'; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); async function main() { @@ -373,7 +381,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md index fe1e76b9..84527826 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md @@ -4,19 +4,19 @@ **重要** -deepseek-v3、deepseek-v3.1、deepseek-v3.2、deepseek-v3.2-exp、deepseek-r1、deepseek-r1-0528、deepseek-r1-distill-qwen-7b/14b/32b 将于**2026年7月9日**下架。推荐转用：[qwen3.7-plus](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-plus)、[qwen3.7-max](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-max)、[qwen3.6-flash](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.6-flash)。 +deepseek-v3、deepseek-v3.1、deepseek-v3.2、deepseek-v3.2-exp、deepseek-r1、deepseek-r1-0528、deepseek-r1-distill-qwen-7b/14b/32b 将于**2026年10月10日**下架。推荐转用：[qwen3.7-plus](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-plus)、[qwen3.7-max](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-max)、[qwen3.6-flash](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.6-flash)。 ## **服务接入地址** -不同地域的服务接入地址不同，请根据您选择的地域配置对应的 Base URL。各地域可调用的模型及限流不同，请参见[限流](https://help.aliyun.com/zh/model-studio/rate-limit)文档。 +不同地域的服务接入地址不同，请根据您选择的地域配置对应的 Base URL（调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)）。各地域可调用的模型及限流不同，请参见[限流](https://help.aliyun.com/zh/model-studio/rate-limit)文档。 ## **OpenAI兼容** ## 华北2（北京） -SDK 调用配置的`base_url`：`https://dashscope.aliyuncs.com/compatible-mode/v1` +SDK 调用配置的`base_url`：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -HTTP 请求地址：`POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions` +HTTP 请求地址：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions` ## 美国（弗吉尼亚） @@ -30,31 +30,25 @@ SDK 调用配置的`base_url`：`https://{WorkspaceId}.ap-southeast-1.maas.aliyu HTTP 请求地址：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions` -调用时请将`WorkspaceId`替换为真实的 Workspace ID。 - ## 德国（法兰克福） SDK 调用配置的`base_url`：`https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1` HTTP 请求地址：`POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions` -调用时请将`WorkspaceId`替换为真实的 Workspace ID。 - ## 日本（东京） SDK 调用配置的`base_url`：`https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` HTTP 请求地址：`POST https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions` -调用时请将`WorkspaceId`替换为真实的 Workspace ID。 - ## **DashScope** ## 华北2（北京） -HTTP 请求地址为`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation` +HTTP 请求地址为`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation` -SDK 调用无需配置 `base_url`。 +SDK 调用配置的`base_url`：`dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"` ## 美国（弗吉尼亚） @@ -68,24 +62,18 @@ HTTP 请求地址为`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com SDK 调用配置的`base_url`：`dashscope.base_http_api_url = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"` -调用时请将`WorkspaceId`替换为真实的 Workspace ID。 - ## 德国（法兰克福） HTTP 请求地址为`POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation` SDK 调用配置的`base_url`：`dashscope.base_http_api_url = "https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1"` -调用时请将`WorkspaceId`替换为真实的 Workspace ID。 - ## 日本（东京） HTTP 请求地址为`POST https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation` SDK 调用配置的`base_url`：`dashscope.base_http_api_url = "https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/api/v1"` -调用时请将`WorkspaceId`替换为真实的 Workspace ID。 - ## **快速开始** deepseek-v4-pro 是 DeepSeek 系列最新模型，在编程、数学和通用任务方面表现出色。您可以通过`enable_thinking`参数在思考与非思考模式之间切换。以下示例展示如何调用思考模式的 deepseek-v4-pro 模型。 @@ -109,6 +97,7 @@ import os client = OpenAI( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) messages = [{"role": "user", "content": "你是谁"}] @@ -174,6 +163,7 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); let reasoningContent = ''; // 完整思考过程 @@ -253,6 +243,7 @@ Request ID: chatcmpl-a1b2c3d4-e5f6-7890-abcd-ef1234567890 ## **curl** ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -280,8 +271,9 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/ ``` import os +import dashscope from dashscope import Generation -# 以下为华北2（北京）地域的配置，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的配置不同。 +# 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" # 初始化请求参数 messages = [{"role": "user", "content": "你是谁？"}] @@ -357,8 +349,10 @@ import io.reactivex.Flowable; import java.lang.System; import java.util.Arrays; public class Main { - // 以下为华北2（北京）地域的配置，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的配置不同。 + static { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + } private static StringBuilder reasoningContent = new StringBuilder(); private static StringBuilder finalContent = new StringBuilder(); private static boolean isFirstPrint = true; @@ -434,6 +428,7 @@ public class Main { ## **curl** ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -469,6 +464,7 @@ import os client = anthropic.Anthropic( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", ) @@ -496,6 +492,7 @@ for event in message: ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic/v1/messages \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -514,11 +511,9 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic/v ## **推理强度（reasoning\_effort）** -deepseek-v4-pro 和 deepseek-v4-flash 默认开启思考模式。通过`reasoning_effort`参数可以调整推理强度，可选值为`high`和`max`，默认为`high`。 - -**说明** +deepseek-v4-pro 和 deepseek-v4-flash 默认开启思考模式。通过`reasoning_effort`参数可以调整推理强度，可选值为`low`、`medium`、`high`、`xhigh`和`max`，默认为`high`。 -设为`low`或`medium`时会映射为`high`，设为`xhigh`时会映射为`max`。 +其中，`low`和`medium`的效果等同于`high`；`xhigh`的效果等同于`max`。 ## **OpenAI兼容** @@ -529,6 +524,7 @@ from openai import OpenAI import os client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -545,6 +541,7 @@ print(completion.choices[0].message.content) import OpenAI from "openai"; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", }); const completion = await openai.chat.completions.create({ @@ -558,6 +555,7 @@ console.log(completion.choices[0].message.content); ## **curl** ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -572,8 +570,9 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/ ``` import os +import dashscope from dashscope import Generation -# 以下为华北2（北京）地域的配置，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的配置不同。 +# 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" response = Generation.call( api_key=os.getenv("DASHSCOPE_API_KEY"), diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md index 155b701e..7c153915 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md @@ -45,6 +45,7 @@ import os client = OpenAI( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -135,6 +136,7 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); @@ -229,6 +231,7 @@ main(); ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -291,6 +294,7 @@ import os client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -348,7 +352,8 @@ import process from 'process'; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); const tools = [ @@ -416,6 +421,7 @@ main(); ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -474,6 +480,7 @@ import os client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -502,11 +509,12 @@ print(completion.usage.prompt_tokens) # true 时少于 false ## **curl** ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ - "model": "glm-5.2", + "model": "ZHIPU/GLM-5.2", "messages": [ {"role": "user", "content": "请计算 15 * 23 是多少？"}, {"role": "assistant", "content": "15 乘以 23 等于 345。", "reasoning_content": "15 * 23 = 345"}, diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md index 0b7948dc..e5181358 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md @@ -14,9 +14,9 @@ glm-4.6、glm-4.7 将于**2026年7月9日**下架。推荐转用：[qwen3.7-plus ## 华北2（北京） -SDK 调用配置的`base_url`：`https://dashscope.aliyuncs.com/compatible-mode/v1` +SDK 调用配置的`base_url`：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -HTTP 请求地址：`POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions` +HTTP 请求地址：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions` ## 美国（弗吉尼亚） @@ -30,23 +30,42 @@ SDK 调用配置的`base_url`：`https://{WorkspaceId}.eu-central-1.maas.aliyunc HTTP 请求地址：`POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## 新加坡 SDK 调用配置的`base_url`：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` HTTP 请求地址：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions` -调用时请将`WorkspaceId`替换为真实的 [Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## DashScope ## 华北2（北京） -HTTP 请求地址：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation` +HTTP 请求地址：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation` + +SDK 调用配置的`base_url`： -SDK 调用无需配置 `base_url`。 +## **Python** + +``` +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' +``` + +## **Java** + +- **方式一：** + + ``` + import com.alibaba.dashscope.protocol.Protocol; + Generation gen = new Generation(Protocol.HTTP.getValue(), "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"); + ``` + +- **方式二：** + + ``` + import com.alibaba.dashscope.utils.Constants; + Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + ``` + ## 美国（弗吉尼亚） @@ -81,22 +100,16 @@ dashscope.base_http_api_url = 'https://dashscope-us.aliyuncs.com/api/v1' HTTP 请求地址：`POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - SDK 调用配置的`base_url`： ## **Python** -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ``` dashscope.base_http_api_url = 'https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1' ``` ## **Java** -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - - **方式一：** ``` @@ -118,7 +131,7 @@ HTTP 请求地址为`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com SDK 调用配置的`base_url`：`dashscope.base_http_api_url = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"` -调用时请将`WorkspaceId`替换为真实的 [Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **快速开始** @@ -144,7 +157,8 @@ import os client = OpenAI( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) messages = [{"role": "user", "content": "你是谁"}] @@ -233,7 +247,8 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); let reasoningContent = ''; // 完整思考过程 @@ -325,7 +340,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -525,7 +541,8 @@ public class Main { ## curl ``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -H "X-DashScope-SSE: enable" \ @@ -560,7 +577,8 @@ import os client = anthropic.Anthropic( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/apps/anthropic", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", ) message = client.messages.create( @@ -587,7 +605,8 @@ for event in message: ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/apps/anthropic/v1/messages \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic/v1/messages \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -H "anthropic-version: 2023-06-01" \ @@ -645,7 +664,8 @@ import os client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) tools = [ @@ -702,7 +722,8 @@ import process from 'process'; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); const tools = [ @@ -774,7 +795,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -867,7 +889,8 @@ for chunk in completion: ## curl ``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -H "X-DashScope-SSE: enable" \ @@ -950,7 +973,8 @@ from openai import OpenAI import os client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( model="glm-5.2", @@ -966,7 +990,8 @@ print(completion.choices[0].message.content) import OpenAI from "openai"; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1", + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", }); const completion = await openai.chat.completions.create({ model: "glm-5.2", @@ -979,7 +1004,8 @@ console.log(completion.choices[0].message.content); ## **curl** ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -1029,7 +1055,8 @@ import os client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) # 多轮对话，assistant 消息中携带 reasoning_content（历史思考过程） @@ -1057,7 +1084,8 @@ print(completion.usage.prompt_tokens) # true 时少于 false ## **curl** ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md index 23cc13c0..d942d163 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md @@ -24,22 +24,28 @@ - 如果通过SDK调用，需要[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk#8833b9274f4v8) -kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6、kimi/kimi-k2.5 均支持输入文本、图像或视频。kimi/kimi-k2.7-code-highspeed 与 kimi/kimi-k2.7-code 功能完全一致，速度提升5~6倍。kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code 为仅思考模型（`enable_thinking` 始终为 true，无法设置为 false）。kimi/kimi-k2.6、kimi/kimi-k2.5 可通过 `enable_thinking` 参数控制思考模式，默认开启思考模式： +kimi 系列模型均支持输入文本、图像或视频： -- **思考模式**（`enable_thinking: true`）：模型会输出详细的推理过程（`reasoning_content`） +1. kimi/kimi-k2.7-code-highspeed 与 kimi/kimi-k2.7-code 功能完全一致，速度提升5~6倍。 -- **非思考模式**（`enable_thinking: false` 或不设置）：直接输出结果，不包含推理过程 +2. kimi/kimi-k3、kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code 为仅思考模型。 +3. kimi/kimi-k2.6、kimi/kimi-k2.5 可通过 `enable_thinking` 参数控制思考模式，默认开启思考模式： + + - **思考模式**（`reasoning_effort: "max"`）：模型会输出详细的推理过程（`reasoning_content`） + + - **非思考模式**（`enable_thinking: false` 或不设置）：直接输出结果，不包含推理过程 + -kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6 支持通过 `preserve_thinking` 参数在多轮对话中传递思考过程，详情请参见[传递思考过程](https://help.aliyun.com/zh/model-studio/deep-thinking#jln7docdq5et5)。 +除 kimi/kimi-k2.5外，其他模型均支持通过 `preserve_thinking` 参数在多轮对话中传递思考过程，详情请参见[传递思考过程](https://help.aliyun.com/zh/model-studio/deep-thinking#jln7docdq5et5)。 -以下示例演示如何调用思考模式的 kimi-k2.6 模型进行文本生成。 +以下示例演示如何调用思考模式的 kimi/kimi-k3 模型进行文本生成。 ## OpenAI兼容 **说明** -`enable_thinking`非 OpenAI 标准参数，OpenAI Python SDK 通过 `extra_body`传入，Node.js SDK 作为顶层参数传入。 +`reasoning_effort`非 OpenAI 标准参数，OpenAI Python SDK 通过 `extra_body`传入，Node.js SDK 作为顶层参数传入。 ## Python @@ -49,14 +55,15 @@ import os client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( - model="kimi/kimi-k2.6", + model="kimi/kimi-k3", messages=[{"role": "user", "content": "1+1等于多少？"}], - # 通过 extra_body 设置 enable_thinking 开启思考模式 - extra_body={"enable_thinking": True} + # 通过 extra_body 设置 reasoning_effort 开启思考模式 + extra_body={"reasoning_effort": "max"} ) msg = completion.choices[0].message @@ -100,7 +107,8 @@ import process from 'process'; const client = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1", + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", }); const messages = [ @@ -109,9 +117,9 @@ const messages = [ ]; const response = await client.chat.completions.create({ - model: "kimi/kimi-k2.6", + model: "kimi/kimi-k3", messages, - extra_body: { enable_thinking: true }, + extra_body: { reasoning_effort: "max" }, }); const msg = response.choices[0].message; @@ -152,11 +160,12 @@ console.log(msg.content); ## curl ``` -curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ - "model": "kimi/kimi-k2.6", + "model": "kimi/kimi-k3", "messages":[ { "role": "system", @@ -167,13 +176,13 @@ curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completi "content": "1+1等于多少？" } ], - "enable_thinking": true + "reasoning_effort": "max" }' ``` ## **多模态调用示例** -kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6、kimi/kimi-k2.5不仅支持纯文本对话，还具备强大的多模态理解能力。本章节将介绍如何让模型理解图像和视频内容。 +Kimi 系列模型不仅支持纯文本对话，还具备强大的多模态理解能力。本章节将介绍如何让模型理解图像和视频内容。 **重要** @@ -193,12 +202,13 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) # 单图传入示例（开启思考模式） completion = client.chat.completions.create( - model="kimi/kimi-k2.6", + model="kimi/kimi-k3", messages=[ { "role": "user", @@ -213,7 +223,7 @@ completion = client.chat.completions.create( ] } ], - extra_body={"enable_thinking":True} # 开启思考模式 + extra_body={"reasoning_effort":"max"} # 开启思考模式 ) # 输出思考过程 @@ -227,7 +237,7 @@ print(completion.choices[0].message.content) # 多图传入示例（开启思考模式，取消注释使用） # completion = client.chat.completions.create( -# model="kimi/kimi-k2.6", +# model="kimi/kimi-k3", # messages=[ # { # "role": "user", @@ -244,7 +254,7 @@ print(completion.choices[0].message.content) # ] # } # ], -# extra_body={"enable_thinking":True} +# extra_body={"reasoning_effort":"max"} # ) # # # 输出思考过程和回复 @@ -261,12 +271,13 @@ import process from 'process'; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); // 单图传入示例（开启思考模式） const completion = await openai.chat.completions.create({ - model: 'kimi/kimi-k2.6', + model: 'kimi/kimi-k3', messages: [ { role: 'user', @@ -281,7 +292,7 @@ const completion = await openai.chat.completions.create({ ] } ], - enable_thinking: true // 开启思考模式 + reasoning_effort: "max" // 开启思考模式 }); // 输出思考过程 @@ -313,7 +324,7 @@ console.log(completion.choices[0].message.content); // ] // } // ], -// enable_thinking: true +// reasoning_effort: "max" // }); // // // 输出思考过程和回复 @@ -328,11 +339,12 @@ console.log(completion.choices[0].message.content); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ - "model": "kimi/kimi-k2.6", + "model": "kimi/kimi-k3", "messages": [ { "role": "user", @@ -350,15 +362,16 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions ] } ], - "enable_thinking": true + "reasoning_effort": "max" }' # 多图输入示例（取消注释使用） -# curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +# curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ # -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ # -H "Content-Type: application/json" \ # -d '{ -# "model": "kimi/kimi-k2.6", +# "model": "kimi/kimi-k3", # "messages": [ # { # "role": "user", @@ -382,7 +395,7 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions # ] # } # ], -# "enable_thinking": true +# "reasoning_effort": "max" # }' ``` @@ -400,11 +413,12 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( - model="kimi/kimi-k2.6", + model="kimi/kimi-k3", messages=[ { "role": "user", @@ -435,12 +449,13 @@ import OpenAI from "openai"; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1" + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" }); async function main() { const response = await openai.chat.completions.create({ - model: "kimi/kimi-k2.6", + model: "kimi/kimi-k3", messages: [ { role: "user", @@ -496,11 +511,12 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ -d '{ - "model": "kimi/kimi-k2.6", + "model": "kimi/kimi-k3", "messages": [ { "role": "user", @@ -560,7 +576,7 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions [结构化输出](https://help.aliyun.com/zh/model-studio/qwen-structured-output) -kimi/kimi-k2.7-code-highspeed +kimi/kimi-k3 支持 @@ -574,22 +590,34 @@ kimi/kimi-k2.7-code-highspeed 支持 +kimi/kimi-k2.7-code-highspeed + kimi/kimi-k2.7-code kimi/kimi-k2.6 kimi/kimi-k2.5 -- kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6、kimi/kimi-k2.5支持上下文缓存（隐式缓存，自动开启），kimi/kimi-k2.7-code-highspeed命中缓存的输入Token按输入价格的20.0%计费，kimi/kimi-k2.7-code命中缓存的输入Token按输入价格的20.0%计费，kimi/kimi-k2.6命中缓存的输入Token按输入价格的16.9%计费，kimi/kimi-k2.5命中缓存的输入Token按输入价格的17.5%计费。 +以上模型支持上下文缓存（隐式缓存，自动开启）： + +- kimi/kimi-k3命中缓存的输入Token按输入价格的10%计费 -- 在思考模式下，使用 kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6、kimi/kimi-k2.5 进行工具调用时：必须在每轮 assistant 消息中保留 `reasoning_content` 字段，`tool_choice` 也仅支持 `"auto"`（默认）和 `"none"`），否则会报错。 +- kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code 命中缓存的输入Token按输入价格的20.0%计费， + +- kimi/kimi-k2.6命中缓存的输入Token按输入价格的16.9%计费 + +- kimi/kimi-k2.5命中缓存的输入Token按输入价格的17.5%计费。 ## **参数默认值** **模型** -**stream\_options** +**tool\_choice** + +**preserve\_thinking** + +**reasoning\_effort** **temperature** @@ -599,33 +627,33 @@ kimi/kimi-k2.5 **presence\_penalty** -**tool\_choice** +**stream\_options** -**top\_k** +kimi/kimi-k3 -**preserve\_thinking** +auto -kimi/kimi-k2.7-code-highspeed +默认关闭 -仅支持设置为`true` +max -1.0 +1 0.95 0.0 -0.0 +\- -auto +仅支持设置为`true` -\- +kimi/kimi-k2.7-code-highspeed -默认开启 +auto -kimi/kimi-k2.7-code +默认开启 -仅支持设置为`true` +\- 1.0 @@ -635,33 +663,37 @@ kimi/kimi-k2.7-code 0.0 +仅支持设置为`true` + +kimi/kimi-k2.7-code + auto +默认开启 + \- -默认开启 +1.0 -kimi/kimi-k2.6 +0.95 + +0.0 + +0.0 仅支持设置为`true` +kimi/kimi-k2.6 + +思考模式/非思考模式：auto + +默认关闭 + +\- + 思考模式：1.0 非思考模式：0.6 - - - - - - - - - - - - - - 思考模式/非思考模式：0.95 @@ -669,27 +701,25 @@ kimi/kimi-k2.6 思考模式/非思考模式：0.0 -思考模式/非思考模式：auto - -\- - -默认关闭 +仅支持设置为`true` kimi/kimi-k2.5 \- +\- + - `stream_options`仅支持设置为`true`，`temperature`、`top_p`、`repetition_penalty`、`presence_penalty`不支持设置为其他值； -- 在思考模式下，不支持强制调用某个工具，`tool_choice`仅支持设置为`auto`（默认值）和`none`。 +- kimi/kimi-k3 支持 `reasoning_effort` 参数，唯一支持值为 `max`。 + +- 在思考模式下，使用 Kimi 模型进行工具调用时：必须在每轮 assistant 消息中保留 `reasoning_content` 字段；对于`tool_choice`参数，`kimi-k3` 支持 `auto` / `none` / `required` 三档；其他模型不支持 `required`，传入会报错。`kimi-k3` 支持动态加载工具，详细用法请参见[动态加载工具](https://platform.kimi.com/docs/guide/use-dynamic-tool-loading)。 - ”-”表示没有默认值，也不支持设置。 ## **模型列表与计费** -kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code 为仅思考模型（`enable_thinking` 始终为 true，无法设置为 false）。kimi/kimi-k2.7-code-highspeed 与 kimi/kimi-k2.7-code 功能完全一致，速度提升5~6倍。kimi/kimi-k2.6、kimi/kimi-k2.5属于混合思考模型，通过`enable_thinking`参数控制是否开启思考模式（注意：均无法通过`thinking_budget`限制思考长度）。 - 模型上下文长度与价格信息请参见[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)。 按照模型的输入与输出 Token 计费。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md index 676f6ad3..e453bbed 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md @@ -32,16 +32,12 @@ SDK 调用配置的`base_url`：`https://{WorkspaceId}.eu-central-1.maas.aliyunc HTTP 请求地址：`POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions` -调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## 新加坡 SDK 调用配置的`base_url`：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` HTTP 请求地址：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions` -调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## 日本（东京） SDK 调用配置的`base_url`：`https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` @@ -61,6 +57,7 @@ SDK调用配置的`base_url`： ## **Python代码** ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' ``` @@ -70,6 +67,7 @@ dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co ``` import com.alibaba.dashscope.protocol.Protocol; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 Generation gen = new Generation(Protocol.HTTP.getValue(), "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"); ``` @@ -77,6 +75,7 @@ dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co ``` import com.alibaba.dashscope.utils.Constants; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; ``` @@ -101,7 +100,7 @@ dashscope.base_http_api_url = 'https://dashscope-us.aliyuncs.com/api/v1' ``` import com.alibaba.dashscope.protocol.Protocol; - Generation gen = new Generation(Protocol.HTTP.getValue(), “https://dashscope-us.aliyuncs.com/api/v1"); + Generation gen = new Generation(Protocol.HTTP.getValue(), "https://dashscope-us.aliyuncs.com/api/v1"); ``` - **方式二：** @@ -154,6 +153,7 @@ SDK调用配置的`base_url`： ## **Python代码** ``` +# 以下为新加坡地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1' ``` @@ -163,6 +163,7 @@ dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyunc ``` import com.alibaba.dashscope.protocol.Protocol; + // 以下为新加坡地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 Generation gen = new Generation(Protocol.HTTP.getValue(), "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"); ``` @@ -170,6 +171,7 @@ dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyunc ``` import com.alibaba.dashscope.utils.Constants; + // 以下为新加坡地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 Constants.baseHttpApiUrl="https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"; ``` @@ -223,6 +225,7 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -291,6 +294,7 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); @@ -375,6 +379,7 @@ main(); ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -456,18 +461,20 @@ for chunk in completion: message = chunk.output.choices[0].message # 只收集思考内容 - if message.reasoning_content: + reasoning_chunk = message.get("reasoning_content") + if reasoning_chunk: if not is_answering: - print(message.reasoning_content, end="", flush=True) - reasoning_content += message.reasoning_content + print(reasoning_chunk, end="", flush=True) + reasoning_content += reasoning_chunk - # 收到 content，开始进行回复 - if message.content: + # 收到 content，开始进行回复（content 为列表，需提取其中的文本） + if message.get("content"): + text = message.content[0].get("text", "") if not is_answering: print("\n" + "=" * 20 + "完整回复" + "=" * 20 + "\n") is_answering = True - print(message.content, end="", flush=True) - answer_content += message.content + print(text, end="", flush=True) + answer_content += text # 循环结束后，reasoning_content 和 answer_content 变量中包含了完整的内容 # 您可以在这里根据需要进行后续处理 @@ -572,6 +579,7 @@ public class Main { ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -630,6 +638,7 @@ import os client = anthropic.Anthropic( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", ) @@ -657,6 +666,7 @@ for event in message: ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic/v1/messages \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -702,6 +712,7 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -770,6 +781,7 @@ import process from 'process'; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); @@ -835,6 +847,7 @@ console.log(completion.choices[0].message.content); ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -861,6 +874,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/ }' # 多图输入示例（取消注释使用） +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 # curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ # -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ # -H "Content-Type: application/json" \ @@ -956,7 +970,7 @@ print(response.output.choices[0].message.content[0]["text"]) ## Java ``` -// dashscope SDK的版本 >= 2.19.4 +// dashscope SDK的版本 >= 2.22.24 import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; @@ -1047,6 +1061,7 @@ public class KimiK26MultiModalExample { ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ @@ -1073,6 +1088,7 @@ curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services }' # 多图输入示例（取消注释使用） +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 # curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" \ # -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ # -H "Content-Type: application/json" \ @@ -1131,6 +1147,7 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -1167,6 +1184,7 @@ import OpenAI from "openai"; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" }); @@ -1203,6 +1221,7 @@ main(); ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -1237,6 +1256,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/ import dashscope import os +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" messages = [ @@ -1314,6 +1334,7 @@ public class Main { ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -1342,6 +1363,7 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -1369,6 +1391,7 @@ import OpenAI from "openai"; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" }); @@ -1404,6 +1427,7 @@ main(); ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -1427,6 +1451,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/ import os import dashscope +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" messages = [{"role": "user", @@ -1466,6 +1491,7 @@ import com.alibaba.dashscope.exception.UploadFileException; import com.alibaba.dashscope.utils.Constants; public class Main { + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} private static final String MODEL_NAME = "kimi-k2.6"; @@ -1504,6 +1530,7 @@ public class Main { ## curl ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -1559,6 +1586,7 @@ base64_image = encode_image("xxx/eagle.png") client = OpenAI( api_key=os.getenv('DASHSCOPE_API_KEY'), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -1602,6 +1630,7 @@ import { readFileSync } from 'fs'; const openai = new OpenAI( { apiKey: process.env.DASHSCOPE_API_KEY, + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" } ); @@ -1655,6 +1684,7 @@ import os import dashscope from dashscope import MultiModalConversation +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" # 编码函数： 将本地文件转换为 Base64 编码的字符串 @@ -1715,6 +1745,7 @@ import com.alibaba.dashscope.utils.Constants; public class Main { + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} private static String encodeToBase64(String imagePath) throws IOException { @@ -1826,6 +1857,7 @@ import os from dashscope import MultiModalConversation import dashscope +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" # 将xxx/eagle.png替换为你本地图像的绝对路径 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md index 7de583d8..349b678a 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md @@ -30,7 +30,8 @@ import os client = OpenAI( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) messages = [{"role": "user", "content": "1+1 等于多少？"}] @@ -99,7 +100,8 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); let reasoningContent = ''; // 完整思考过程 @@ -179,7 +181,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md index 38765019..9790a912 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md @@ -31,7 +31,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -91,7 +92,8 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); let reasoningContent = ''; // 完整思考过程 @@ -159,7 +161,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -236,7 +239,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) # 单图传入示例（自适应思考模式） @@ -304,7 +308,8 @@ import process from 'process'; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); // 单图传入示例（自适应思考模式） @@ -371,7 +376,8 @@ console.log(completion.choices[0].message.content); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -397,7 +403,8 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions }' # 多图输入示例（取消注释使用） -# curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +# curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ # -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ # -H "Content-Type: application/json" \ # -d '{ @@ -441,7 +448,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -476,7 +484,8 @@ import OpenAI from "openai"; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1" + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" }); async function main() { @@ -513,7 +522,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ -d '{ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md index 2fc5120d..3a5396cd 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md @@ -8,7 +8,7 @@ MiniMax-M2.1 将于**2026年7月9日**下架。推荐转用：[qwen3.7-plus](htt **重要** -本文档仅适用于中国内地地域。如需使用模型，需从中国内地地域[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 +本文档描述的功能仅在华北2（北京）地域可用，如需使用模型，需从华北2（北京）地域[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 ## **快速开始** @@ -26,7 +26,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -83,7 +84,8 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); let reasoningContent = ''; // 完整思考过程 @@ -148,7 +150,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -205,7 +208,10 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions ``` import os +import dashscope from dashscope import Generation +# 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" # 初始化请求参数 messages = [{"role": "user", "content": "你是谁？"}] @@ -276,6 +282,7 @@ import com.alibaba.dashscope.common.Role; import com.alibaba.dashscope.exception.ApiException; import com.alibaba.dashscope.exception.InputRequiredException; import com.alibaba.dashscope.exception.NoApiKeyException; +import com.alibaba.dashscope.utils.Constants; import io.reactivex.Flowable; import java.lang.System; import java.util.Arrays; @@ -283,6 +290,10 @@ import org.slf4j.Logger; import org.slf4j.LoggerFactory; public class Main { + static { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + } private static final Logger logger = LoggerFactory.getLogger(Main.class); private static StringBuilder reasoningContent = new StringBuilder(); private static StringBuilder finalContent = new StringBuilder(); @@ -365,7 +376,8 @@ public class Main { ## curl ``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -415,6 +427,64 @@ curl -X POST "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generatio } ``` +## Anthropic兼容 + +## Python + +### **示例代码** + +``` +import anthropic +import os + +client = anthropic.Anthropic( + # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" + api_key=os.getenv("DASHSCOPE_API_KEY"), + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", +) + +message = client.messages.create( + model="MiniMax-M2.5", + max_tokens=1024, + messages=[ + {"role": "user", "content": "你是谁"} + ], + stream=True, +) + +for event in message: + if event.type == "content_block_delta": + if hasattr(event.delta, "thinking"): + print(event.delta.thinking, end="", flush=True) + if hasattr(event.delta, "text"): + print(event.delta.text, end="", flush=True) +``` + +## HTTP + +### **示例代码** + +## curl + +``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic/v1/messages \ +-H "Authorization: Bearer $DASHSCOPE_API_KEY" \ +-H "Content-Type: application/json" \ +-H "anthropic-version: 2023-06-01" \ +-d '{ + "model": "MiniMax-M2.5", + "max_tokens": 1024, + "messages": [ + { + "role": "user", + "content": "你是谁" + } + ] +}' +``` + ## **其它功能** **模型** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md index 6d5c5d09..bf122dc0 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md @@ -41,7 +41,8 @@ import os client = OpenAI( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) messages = [{"role": "user", "content": "你是谁"}] @@ -120,7 +121,8 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); let reasoningContent = ''; // 完整思考过程 @@ -215,7 +217,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -242,8 +245,12 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions ``` import os +import dashscope from dashscope import Generation +# 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" + # 初始化请求参数 messages = [{"role": "user", "content": "你是谁？"}] @@ -334,8 +341,14 @@ import com.alibaba.dashscope.exception.NoApiKeyException; import io.reactivex.Flowable; import java.lang.System; import java.util.Arrays; +import com.alibaba.dashscope.utils.Constants; public class Main { + static { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + } + private static StringBuilder reasoningContent = new StringBuilder(); private static StringBuilder finalContent = new StringBuilder(); private static boolean isFirstPrint = true; @@ -406,7 +419,8 @@ DeepSeek-V3，一个由深度求索公司创造的智能助手！我可以帮助 ## curl ``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation" \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -H "X-DashScope-SSE: enable" \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md index f02e6efc..861138b5 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md @@ -24,7 +24,8 @@ import os client = OpenAI( # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key="sk-xxx" api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) messages = [{"role": "user", "content": "9.9和9.11哪个大？"}] @@ -98,7 +99,8 @@ import process from 'process'; const openai = new OpenAI({ // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: "sk-xxx" apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); let reasoningContent = ''; // 完整思考过程 @@ -181,7 +183,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -216,7 +219,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -256,7 +260,8 @@ import process from 'process'; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1' + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1' }); const completion = await openai.chat.completions.create({ @@ -294,7 +299,8 @@ console.log(completion.choices[0].message.content); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ @@ -332,7 +338,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + # 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) completion = client.chat.completions.create( @@ -366,7 +373,8 @@ import OpenAI from "openai"; const openai = new OpenAI({ apiKey: process.env.DASHSCOPE_API_KEY, - baseURL: "https://dashscope.aliyuncs.com/compatible-mode/v1" + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" }); async function main() { @@ -402,7 +410,8 @@ main(); ## curl ``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ -d '{ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md index 51c73192..6d4b8347 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md @@ -1,6 +1,6 @@ # Chatbox -Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版、Coding Plan或按量计费接入阿里云百炼。 +Chatbox 是一款跨平台 AI 客户端应用，可以通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **下载安装 Chatbox** @@ -12,6 +12,8 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 百炼提供三种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -19,6 +21,24 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +**配置项** + +**说明** + +**API 密钥** + +填入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)。 + +**API 主机** + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +**模型** + +填入 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)，如 `qwen3.8-max-preview`。 + ### Token Plan 团队版 **配置项** @@ -27,7 +47,7 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 **API 密钥** -填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。 +填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise)。 **API 主机** @@ -71,11 +91,11 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 **API 主机** -根据模型部署地域，填入对应 URL： +根据模型部署地域，填入对应 URL（请将 URL 中的 `{WorkspaceId}` 替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)）： -- 华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` @@ -100,4 +120,6 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md index af83a91a..a3880109 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md @@ -1,6 +1,6 @@ # Cherry Studio -Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团队版、Coding Plan或按量计费接入阿里云百炼。 +Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **安装 Cherry Studio** @@ -12,6 +12,8 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 百炼提供三种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -19,6 +21,24 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +**配置项** + +**说明** + +**API 密钥** + +填入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)。 + +**API 地址** + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +**模型** + +可用模型请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + ### Token Plan 团队版 **配置项** @@ -27,7 +47,7 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 **API 密钥** -填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。 +填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise)。 **API 地址** @@ -67,11 +87,11 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 **API 地址** -根据地域，填入对应 URL： +根据地域，填入对应 URL（请将 URL 中的 `{WorkspaceId}` 替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)）： -- 华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` @@ -96,7 +116,9 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 报错 The value of the enable\_thinking parameter is restricted to True @@ -109,7 +131,7 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 可能的原因： -- **地域不匹配**：免费额度仅适用于华北2（北京）地域（[中国内地服务部署范围](https://help.aliyun.com/zh/model-studio/regions/#080da663a75xh)）的模型。使用其他地域和部署范围的模型会产生费用。请检查**API 地址**是否与目标地域匹配，详情请参见[新人免费额度](https://help.aliyun.com/zh/model-studio/new-free-quota)。 +- **地域不匹配**：免费额度仅适用于华北2（北京）地域的模型。使用其他地域的模型会产生费用。请检查**API 地址**是否与目标地域匹配，详情请参见[新人免费额度](https://help.aliyun.com/zh/model-studio/new-free-quota)。 - **额度按模型独立计算**：各模型的免费额度相互独立，不可跨模型共享。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md index 00895664..36af72c4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md @@ -1,6 +1,6 @@ # Claude Code -Claude Code 是 Anthropic 推出的命令行 AI 编程助手。通过阿里云百炼，可以使用按量计费、Coding Plan 或 Token Plan 团队版接入 Claude Code。 +Claude Code 是 Anthropic 推出的命令行 AI 编程助手。通过阿里云百炼，可以使用按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入 Claude Code。 ## **安装 Claude Code** @@ -47,6 +47,36 @@ npm install -g @anthropic-ai/claude-code 新建 `~/.claude/settings.json`（Windows 路径：`C:\Users\<用户名>\.claude\settings.json`），写入对应套餐的配置。 +### Token Plan 个人版 + +将 YOUR\_API\_KEY 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型：qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro。完整说明参见 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +``` +{ + "env": { + "ANTHROPIC_AUTH_TOKEN": "YOUR_API_KEY", + "ANTHROPIC_BASE_URL": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic", + "ANTHROPIC_MODEL": "qwen3.8-max-preview", + "ANTHROPIC_DEFAULT_HAIKU_MODEL": "qwen3.6-flash", + "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.8-max-preview", + "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen3.8-max-preview", + "CLAUDE_CODE_SUBAGENT_MODEL": "qwen3.7-max", + "CLAUDE_CODE_MAX_CONTEXT_TOKENS": "983616" + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- **thinking**：始终开启，不支持关闭。 + +- **temperature**：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- **reasoning\_effort**：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Token Plan 团队版 将 YOUR\_API\_KEY 替换为 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。可用模型参见 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。 @@ -56,15 +86,27 @@ npm install -g @anthropic-ai/claude-code "env": { "ANTHROPIC_AUTH_TOKEN": "YOUR_API_KEY", "ANTHROPIC_BASE_URL": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic", - "ANTHROPIC_MODEL": "qwen3.7-max", + "ANTHROPIC_MODEL": "qwen3.8-max-preview", "ANTHROPIC_DEFAULT_HAIKU_MODEL": "qwen3.6-flash", - "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.7-max", - "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen3.7-max", - "CLAUDE_CODE_SUBAGENT_MODEL": "qwen3.7-max" + "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.8-max-preview", + "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen3.8-max-preview", + "CLAUDE_CODE_SUBAGENT_MODEL": "qwen3.7-max", + "CLAUDE_CODE_MAX_CONTEXT_TOKENS": "983616" } } ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- **thinking**：始终开启，不支持关闭。 + +- **temperature**：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- **reasoning\_effort**：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Coding Plan 将 YOUR\_API\_KEY 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。可用模型参见 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -87,18 +129,18 @@ npm install -g @anthropic-ai/claude-code 将 YOUR\_API\_KEY 替换为[阿里云百炼API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。可用模型参见[Anthropic 兼容 API](https://help.aliyun.com/zh/model-studio/anthropic-api-messages#07833dedefft7)。 -`ANTHROPIC_BASE_URL` 按地域设置，API Key 需与所选地域对应： +`ANTHROPIC_BASE_URL` 按地域设置，API Key 需与所选地域对应，并将`WorkspaceId`替换为真实的 [Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)： -- 华北2（北京）：`https://dashscope.aliyuncs.com/apps/anthropic` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic` ``` { "env": { "ANTHROPIC_AUTH_TOKEN": "YOUR_API_KEY", - "ANTHROPIC_BASE_URL": "https://dashscope.aliyuncs.com/apps/anthropic", + "ANTHROPIC_BASE_URL": "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", "ANTHROPIC_MODEL": "qwen3.7-max", "ANTHROPIC_DEFAULT_HAIKU_MODEL": "qwen3.6-flash", "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.7-max", @@ -122,7 +164,7 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 { "env": { "ANTHROPIC_AUTH_TOKEN": "YOUR_API_KEY", - "ANTHROPIC_BASE_URL": "https://dashscope.aliyuncs.com/apps/anthropic", + "ANTHROPIC_BASE_URL": "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", "ANTHROPIC_MODEL": "qwen3.7-plus", "CLAUDE_CODE_MAX_CONTEXT_TOKENS": "1000000" } @@ -137,7 +179,7 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 { "env": { "ANTHROPIC_AUTH_TOKEN": "YOUR_API_KEY", - "ANTHROPIC_BASE_URL": "https://dashscope.aliyuncs.com/apps/anthropic", + "ANTHROPIC_BASE_URL": "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", "ANTHROPIC_MODEL": "qwen3.7-plus[1m]", "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.7-plus[1m]", "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen3.7-plus[1m]", @@ -169,6 +211,14 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 **配置信息** + Token Plan 个人版 + + 供应商名称：百炼-Token Plan 个人版 + + API Key：[控制台获取](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + + 请求地址：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` + Token Plan 团队版 供应商名称：百炼-Token Plan @@ -191,7 +241,7 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 API Key：[百炼API Key](https://help.aliyun.com/zh/model-studio/get-api-key) - 请求地址：`https://dashscope.aliyuncs.com/apps/anthropic` + 请求地址：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic` 2. 展开**高级选项**配置模型映射，将主模型与 Haiku、Sonnet、Opus 默认模型设置为对应套餐[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。映射关系按需选择，示例如下： @@ -208,35 +258,23 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 ### 接入 Claude Code 桌面版 +Claude Code 桌面版（Claude Desktop）与 Claude Code CLI 是两个独立入口，在 CC Switch 中分别对应 **Claude Code** 与 **Claude Desktop** 面板。桌面版通过 CC Switch 本地网关访问百炼：网关地址与鉴权令牌均由 CC Switch 自动写入桌面版配置，**无需在桌面版中手动填写百炼 API Key**——百炼 API Key 只在 CC Switch 供应商配置中填写，由本地路由转发时自动注入。 + +**重要** + +请勿在桌面版的第三方推理配置中手动填写百炼 API Key。桌面版对 CC Switch 本地网关（地址 `http://127.0.0.1:15721/claude-desktop`）的鉴权令牌由 CC Switch 自动生成并写入，手动填入百炼 API Key 会因令牌不匹配导致鉴权失败。桌面版第三方配置写入目前仅支持 macOS、Windows。 + 1. 从 [Claude 下载页](https://claude.ai/download)安装 Claude Code 桌面版。 -2. 顶部菜单 **Help** → **Troubleshooting** → **Enable Developer Mode**，重启后顶部出现 **Developer** 菜单。 - -3. **Developer** → **Configure Third-Party Inference**，**Connection** 选 **Gateway**，按下表填写后点击 **Apply locally**： - - **字段** - - **填写** - - Gateway base URL - - CC Switch 路由监听地址，默认 `http://127.0.0.1:15721`，如已修改则与下一步保持一致。 +2. 在 CC Switch 左侧应用切换器切换到 **Claude Desktop** 面板。若未显示该入口，前往**设置 → 通用 → 应用可见性**确认 Claude Desktop 未被隐藏。 - Gateway API key +3. 添加百炼供应商：若已在 **Claude Code** 面板配置过百炼供应商，可点击**将 Claude Code 中已有的供应商导入**一键复用；也可点击右上角 **+** 新增。由于百炼模型 ID（如 `qwen3.7-max`）不是 Claude Desktop 识别的 `claude-sonnet-* / claude-opus-* / claude-haiku-*` 三档角色 ID，需开启**需要模型映射**，为 Sonnet、Opus、Haiku 三档分别填写实际请求的百炼模型（如 Sonnet → qwen3.7-max）。 - 百炼API Key +4. 开启本地路由：前往**设置 → 路由 → 本地路由**，打开**在主页面显示本地路由开关**；回到 Claude Desktop 面板，打开 **Claude Desktop 本地路由**开关，监听地址默认 `127.0.0.1:15721`。 - Gateway auth scheme +5. 在供应商卡片点击**启用**，CC Switch 会自动将第三方推理配置写入 Claude Code 桌面版。 - bearer - - Model list - - Model ID须为 Anthropic 风格，如 `claude-opus-4.7`。实际调用模型由 CC Switch 路由决定，对应关系即供应商[高级选项中的模型映射](#ccswitch-add-li2)。Display name 仅影响下拉显示。 - -4. CC Switch 左上角设置 → **路由**，开启**路由总开关**，监听地址默认 `127.0.0.1:15721`，如需修改请同步上一步。 - -5. 在桌面版模型下拉中选择已配置的模型ID即可使用。 +6. 保持 CC Switch 运行，**完全退出并重启** Claude Code 桌面版后生效，在模型菜单中选择已配置的模型即可使用。 ## **Claude Code IDE 插件** @@ -295,7 +333,9 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 启动 Claude Code 后，界面显示"Unable to connect to Anthropic services. Failed to connect to api.anthropic.com: ERR\_BAD\_REQUEST" diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cline.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cline.md index 0d220bf9..8126ee3f 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cline.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cline.md @@ -1,6 +1,6 @@ # Cline -Cline 是一款 VSCode 智能编程插件，可以通过 Token Plan 团队版、Coding Plan或按量计费接入阿里云百炼。 +Cline 是一款 VSCode 智能编程插件，可以通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **安装 Cline** @@ -15,6 +15,8 @@ Cline 是一款 VSCode 智能编程插件，可以通过 Token Plan 团队版、 阿里云百炼提供三种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -22,6 +24,28 @@ Cline 是一款 VSCode 智能编程插件，可以通过 Token Plan 团队版、 - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +**配置项** + +**说明** + +API Provider + +选择 **OpenAI Compatible**。 + +Base URL + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +API Key + +填入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。 + +Model ID + +填入 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)，如 `qwen3.8-max-preview`。 + ### Token Plan 团队版 **配置项** @@ -42,7 +66,7 @@ API Key Model ID -填入 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)，如 `qwen3.7-max`。 +填入 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)，如 `qwen3.8-max-preview`。 ### Coding Plan @@ -78,11 +102,11 @@ API Provider Base URL -根据地域，填入对应 URL： +根据地域，填入对应 URL（请将 URL 中的 `{WorkspaceId}` 替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)）： -- 华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` @@ -133,7 +157,9 @@ Model ID - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 报错 401 Incorrect API key provided diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/codex.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/codex.md index 76a25932..eaae7c03 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/codex.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/codex.md @@ -1,6 +1,6 @@ # Codex -Codex 是 OpenAI 推出的终端 AI 编程助手。可通过 Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 +Codex 是 OpenAI 推出的终端 AI 编程助手。可通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **安装 Codex** @@ -21,19 +21,199 @@ Codex 是 OpenAI 推出的终端 AI 编程助手。可通过 Token Plan 团队 ## **配置接入凭证** -接入需要编辑配置文件`~/.codex/config.toml`并配置环境变量`OPENAI_API_KEY`。根据所选计费方案替换对应值，阿里云百炼提供三种计费方案： +接入需要编辑配置文件`~/.codex/config.toml`并配置环境变量`OPENAI_API_KEY`。根据所选计费方案替换对应值，阿里云百炼提供以下计费方案： + +### 配置模型元数据 + +使用自定义模型（如 qwen3.8-max-preview）时，需要配置模型元数据文件，使 Codex 正确识别模型的上下文窗口、推理深度等参数。 + +1. 新建文件 `~/.codex/model-catalog.local.json`，写入以下内容： + + ``` + { + "models": [ + { + "slug": "qwen3.8-max-preview", + "display_name": "qwen3.8-max-preview", + "description": "DashScope model: qwen3.8-max-preview", + "default_reasoning_level": "xhigh", + "supported_reasoning_levels": [ + { + "effort": "low", + "description": "Fast responses with lighter reasoning" + }, + { + "effort": "medium", + "description": "Greater reasoning depth for complex problems" + }, + { + "effort": "xhigh", + "description": "Extra high reasoning depth for complex problems" + } + ], + "context_window": 983616, + "effective_context_window_percent": 95, + "supports_parallel_tool_calls": false, + "supports_image_detail_original": true, + "input_modalities": ["text", "image"], + "shell_type": "default", + "visibility": "list", + "supported_in_api": true, + "priority": 1, + "base_instructions": "", + "support_verbosity": false, + "supports_reasoning_summaries": false, + "experimental_supported_tools": [], + "truncation_policy": { + "mode": "bytes", + "limit": 10000 + } + } + ] + } + ``` + +2. 在 `~/.codex/config.toml` 中添加以下配置，指向元数据文件： + + ``` + model_catalog_json = "~/.codex/model-catalog.local.json" + ``` + + +### Token Plan 个人版 + +`model`请选择[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)，可用模型包括 qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro。将`OPENAI_API_KEY`环境变量设置为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。 + +#### Responses API（qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash） + +qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-plus 和 qwen3.6-flash 支持 Responses API，可使用最新版 Codex。 + +``` +model_provider = "Model_Studio_Token_Plan_Personal" +model = "qwen3.8-max-preview" +[model_providers.Model_Studio_Token_Plan_Personal] +name = "Model_Studio_Token_Plan_Personal" +base_url = "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" +env_key = "OPENAI_API_KEY" +wire_api = "responses" +``` + +#### Chat/Completions API（其他模型） + +其他模型需通过 Chat/Completions API 接入，需安装旧版本 Codex，如 0.80.0： + +``` +npm install -g @openai/codex@0.80.0 +``` +``` +model_provider = "Model_Studio_Token_Plan_Personal" +model = "glm-5" +[model_providers.Model_Studio_Token_Plan_Personal] +name = "Model_Studio_Token_Plan_Personal" +base_url = "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" +env_key = "OPENAI_API_KEY" +wire_api = "chat" +``` + +#### 配置环境变量 + +将`OPENAI_API_KEY`环境变量设置为 Token Plan 个人版专属 API Key。 + +## macOS + +1. 在终端中执行以下命令，查看默认 Shell 类型。 + + ``` + echo $SHELL + ``` + +2. 根据 Shell 类型设置环境变量： + + ## Zsh + + ``` + # 将 YOUR_API_KEY 替换为 Token Plan 个人版 API Key + echo 'export OPENAI_API_KEY="YOUR_API_KEY"' >> ~/.zshrc + ``` + + ## Bash + + ``` + # 将 YOUR_API_KEY 替换为 Token Plan 个人版 API Key + echo 'export OPENAI_API_KEY="YOUR_API_KEY"' >> ~/.bash_profile + ``` + +3. 执行以下命令使环境变量生效。 + + ## Zsh + + ``` + source ~/.zshrc + ``` + + ## Bash + + ``` + source ~/.bash_profile + ``` + + +## Windows + +## CMD + +1. 在 CMD 中运行以下命令，设置环境变量。 + + ``` + REM 将 YOUR_API_KEY 替换为 Token Plan 个人版 API Key + setx OPENAI_API_KEY "YOUR_API_KEY" + ``` + +2. 打开一个新的 CMD 窗口，运行以下命令检查环境变量是否生效。 + + ``` + echo %OPENAI_API_KEY% + ``` + + +## PowerShell + +1. 在 PowerShell 中运行以下命令，设置环境变量。 + + ``` + # 将 YOUR_API_KEY 替换为 Token Plan 个人版 API Key + [Environment]::SetEnvironmentVariable("OPENAI_API_KEY", "YOUR_API_KEY", [EnvironmentVariableTarget]::User) + ``` + +2. 打开一个新的 PowerShell 窗口，运行以下命令检查环境变量是否生效。 + + ``` + echo $env:OPENAI_API_KEY + ``` + + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + ### Token Plan 团队版 `model`请选择[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。将`OPENAI_API_KEY`环境变量设置为 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。 -#### Responses API（qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash） +#### Responses API（qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash） -qwen3.7-max、qwen3.7-plus、qwen3.6-plus 和 qwen3.6-flash 支持 Responses API，可使用最新版 Codex。 +qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-plus 和 qwen3.6-flash 支持 Responses API，可使用最新版 Codex。 ``` model_provider = "Model_Studio_Token_Plan" -model = "qwen3.7-max" +model = "qwen3.8-max-preview" [model_providers.Model_Studio_Token_Plan] name = "Model_Studio_Token_Plan" base_url = "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" @@ -135,6 +315,17 @@ wire_api = "chat" ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Coding Plan `model`请选择[支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan-overview)。将`OPENAI_API_KEY`环境变量设置为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。 @@ -237,11 +428,11 @@ wire_api = "chat" 将`OPENAI_API_KEY`环境变量设置为[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。可用模型参见[支持的模型](https://help.aliyun.com/zh/model-studio/compatibility-of-openai-with-dashscope#7f9c78ae99pwz)。 -根据地域设置`base_url`，API Key 须与所选地域对应： +根据地域设置`base_url`，API Key 须与所选地域对应，请将 URL 中的 `{WorkspaceId}` 替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)： -- 华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` 按量计费支持 Responses API 和 Chat/Completions API 两种接入方式，请根据使用的模型选择： @@ -255,7 +446,7 @@ model_provider = "Model_Studio" model = "qwen3.7-max" [model_providers.Model_Studio] name = "Model_Studio" -base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1" +base_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" env_key = "OPENAI_API_KEY" wire_api = "responses" ``` @@ -272,7 +463,7 @@ model_provider = "Model_Studio" model = "qwen3.6-plus" [model_providers.Model_Studio] name = "Model_Studio" -base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1" +base_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" env_key = "OPENAI_API_KEY" wire_api = "chat" ``` @@ -368,11 +559,11 @@ codex ### **第三方工具提示“不支持国内模型”或“检查被拒 / Bad request (400)”怎么办？** -**原因**：部分第三方管理工具（如 CC-Switch）在切换供应商时会发起“健康检查/连接测试”探测请求，该探测请求的格式与 Codex 实际调用的请求格式不同，百炼网关可能因此返回 400 Bad request 并提示“检查被拒”，工具据此显示“不支持国内模型”。此提示仅代表健康检查探测未通过，**并不代表百炼不支持国内模型，也不影响 Codex 的实际使用。** +**原因**：部分第三方管理工具（如 CC-Switch）在切换供应商时会发起“健康检查/连接测试”探测请求，该探测请求的格式与 Codex 实际调用的请求格式不同，百炼网关可能因此返回 400 Bad request 并提示“检查被拒”，工具据此显示“不支持国内模型”。此提示仅代表健康检查探测未通过，**并不代表百炼不支持中国内地模型，也不影响 Codex 的实际使用。** -**说明**：百炼支持通过 Codex 使用 qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、glm-5 等国内模型，配置方式详见上文[配置接入凭证](#cdx-config)。 +**说明**：百炼支持通过 Codex 使用 qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、glm-5 等中国内地模型，配置方式详见上文[配置接入凭证](#cdx-config)。 -**解决方案**：建议参照上文配置接入凭证，直接在`~/.codex/config.toml`中完成配置，无需依赖第三方工具的健康检查结果；配置完成后参照[验证配置](#cdx-verify)启动 Codex，若能正常进入对话界面即表示可正常使用国内模型。 +**解决方案**：建议参照上文配置接入凭证，直接在`~/.codex/config.toml`中完成配置，无需依赖第三方工具的健康检查结果；配置完成后参照[验证配置](#cdx-verify)启动 Codex，若能正常进入对话界面即表示可正常使用中国内地模型。 ### **报错 wire\_api 配置问题怎么办？** @@ -394,7 +585,7 @@ codex **原因**： -- 误用了其他方案的 API Key（Token Plan 团队版、Coding Plan 和按量计费的 API Key 互不相通） +- 误用了其他方案的 API Key（Token Plan 个人版、Token Plan 团队版、Coding Plan 和按量计费的 API Key 互不相通） - 订阅过期 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cursor.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cursor.md index f903bf87..61f5ca45 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cursor.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cursor.md @@ -1,6 +1,6 @@ # Cursor -Cursor 是一款 AI 编程 IDE，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Cursor 是一款 AI 编程 IDE，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 Cursor** @@ -12,6 +12,8 @@ Cursor 是一款 AI 编程 IDE，可以通过按量计费、Coding Plan 或 Toke 阿里云百炼提供三种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -19,6 +21,20 @@ Cursor 是一款 AI 编程 IDE，可以通过按量计费、Coding Plan 或 Toke - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +**API Key** + +Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +**Base URL** + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +**可用模型** + +Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview) + ### Token Plan 团队版 **API Key** @@ -59,11 +75,11 @@ Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-pla **Base URL** -根据地域，填入对应 URL： +根据地域，填入对应 URL，请将 URL 中的 `{WorkspaceId}` 替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)： -- 华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` @@ -110,7 +126,9 @@ Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-pla - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 在 Cursor 中无法调用已添加的模型 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/dify.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/dify.md index 68960a12..7a6b3698 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/dify.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/dify.md @@ -1,6 +1,6 @@ # Dify -Dify 是一个开源的大模型应用开发平台，您可以基于阿里云百炼提供的模型 API 来构建大模型应用。 +Dify 是一个开源的大模型应用开发平台，您可以基于阿里云百炼提供的模型 API（按量付费或 Token Plan 个人版）来构建大模型应用。 ## **前提条件** @@ -36,7 +36,7 @@ Dify 是一个开源的大模型应用开发平台，您可以基于阿里云百 单击**通义千问**卡片中的显示模型，打开您需要使用的模型开关。 -> 若插件内暂未包含最新版千问模型，可尝试安装 **OpenAI-API-compatible** 插件，在插件设置中的 **API endpoint URL** 填入`https://dashscope.aliyuncs.com/compatible-mode/v1`（华北2（北京）地域）或`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（新加坡地域）。其中`WorkspaceId`需替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +> 若插件内暂未包含最新版千问模型，可尝试安装 **OpenAI-API-compatible** 插件，在插件设置中的 **API endpoint URL** 填入`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（华北2（北京）地域）或`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（新加坡地域）。其中`WorkspaceId`需替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **2\. 开始使用** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md index b4f0ca4f..c614a907 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md @@ -22,7 +22,7 @@ Postman 和 cURL仅适用于快速测试与功能验证。对于生产环境 **HTTP调用示例（文生图）** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4178421871/CAEQZRiBgICJ0seH1xkiIDgzYWE2MTBkZjkzODRkNDA5NzczNTE0NjBiMGE1Y2Nm5274221_20250627113930.173.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4507315871/CAEQZRiBgICJ0seH1xkiIDgzYWE2MTBkZjkzODRkNDA5NzczNTE0NjBiMGE1Y2Nm5274221_20250627113930.173.svg) ## **方式一：使用Postman发送请求（推荐）** @@ -147,7 +147,6 @@ curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text2image/imag - （可选）点击页面右侧的 `**Beautify**`，可以格式化JSON格式，使其更易阅读。 - 5. 点击**Send**发送请求，并获取 `task_id`。有效期 24 小时，过期后无法查询，请及时获取结果。 ``` @@ -189,7 +188,6 @@ curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ 5. 点击**Send**发送请求。 - 2. 检查返回结果。重复发送此请求（轮询），直到 task\_status 变为 SUCCEEDED，获取图像的URL。图像URL有效期为**24小时**，请及时下载。 ``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md index 247ca4b6..f96c40c4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md @@ -1,6 +1,6 @@ # Hermes Agent -Hermes Agent 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Hermes Agent 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 Hermes Agent** @@ -31,6 +31,8 @@ Hermes Agent 是一款终端 AI 编程工具，可以通过按量计费、Coding 通过 `hermes config set` 命令配置接入参数，根据所选方案填入对应的 Base URL 和 API Key： +- **Token Plan 个人版**：个人订阅，按 token 消耗抵扣 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -40,10 +42,46 @@ Hermes Agent 是一款终端 AI 编程工具，可以通过按量计费、Coding **说明** -本文示例均使用 **Anthropic 兼容协议**：Base URL 以 `/apps/anthropic` 结尾，并将 `api_mode` 设为 `anthropic_messages`。Hermes Agent 同样支持 **OpenAI 兼容协议**：将 Base URL 结尾的 `/apps/anthropic` 替换为 `/compatible-mode/v1`，并删除 `api_mode` 配置项即可。例如按量计费（华北2·北京）的 OpenAI 兼容 Base URL 为 `https://dashscope.aliyuncs.com/compatible-mode/v1`。 +本文示例均使用 **Anthropic 兼容协议**：Base URL 以 `/apps/anthropic` 结尾，并将 `api_mode` 设为 `anthropic_messages`。Hermes Agent 同样支持 **OpenAI 兼容协议**：将 Base URL 结尾的 `/apps/anthropic` 替换为 `/compatible-mode/v1`，并删除 `api_mode` 配置项即可。例如按量计费（华北2·北京）的 OpenAI 兼容 Base URL 为 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。 除命令行版外，Hermes Agent 还提供桌面版（Hermes Desktop）。可从 [Hermes 官网](https://hermes-agent.nousresearch.com/) 下载安装包，或在命令行版安装完成后运行 `hermes desktop` 启动。桌面版与命令行版共用同一份 `~/.hermes/config.yaml` 配置文件，接入参数与本文一致；在桌面版中以自定义端点（Custom Endpoint）方式接入时，请使用上述 OpenAI 兼容 Base URL。 +### Token Plan 个人版 + +将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型：qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro，完整列表请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +``` +hermes config set model.provider custom +hermes config set model.base_url https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic +hermes config set model.api_mode anthropic_messages +hermes config set model.api_key YOUR_API_KEY +hermes config set model.default qwen3.8-max-preview +``` + +以上命令将配置写入 `~/.hermes/config.yaml`。也可以直接编辑该文件，写入以下内容： + +config.yaml 配置示例 + +``` +model: + default: qwen3.8-max-preview + provider: custom + base_url: https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic + api_mode: anthropic_messages + api_key: YOUR_API_KEY +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Token Plan 团队版 将 `YOUR_API_KEY` 替换为 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。可用模型请参考 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。 @@ -53,7 +91,7 @@ hermes config set model.provider custom hermes config set model.base_url https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic hermes config set model.api_mode anthropic_messages hermes config set model.api_key YOUR_API_KEY -hermes config set model.default qwen3.7-max +hermes config set model.default qwen3.8-max-preview ``` 以上命令将配置写入 `~/.hermes/config.yaml`。也可以直接编辑该文件，写入以下内容： @@ -62,13 +100,24 @@ config.yaml 配置示例 ``` model: - default: qwen3.7-max + default: qwen3.8-max-preview provider: custom base_url: https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic api_mode: anthropic_messages api_key: YOUR_API_KEY ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Coding Plan 将 `YOUR_API_KEY` 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -98,16 +147,16 @@ model: 将 `YOUR_API_KEY` 替换为[阿里云百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。可用模型请参考[Anthropic 兼容 API](https://help.aliyun.com/zh/model-studio/anthropic-api-messages#07833dedefft7)。 -`base_url` 按地域设置，API Key 需与所选地域对应： +`base_url` 按地域设置，API Key 需与所选地域对应，并将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)： -- 华北2（北京）：`https://dashscope.aliyuncs.com/apps/anthropic` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic` ``` hermes config set model.provider alibaba -hermes config set model.base_url https://dashscope.aliyuncs.com/apps/anthropic +hermes config set model.base_url https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic hermes config set model.api_mode anthropic_messages hermes config set model.api_key YOUR_API_KEY hermes config set model.default qwen3.7-max @@ -121,7 +170,7 @@ config.yaml 配置示例 model: default: qwen3.7-max provider: alibaba - base_url: https://dashscope.aliyuncs.com/apps/anthropic + base_url: https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic api_mode: anthropic_messages api_key: YOUR_API_KEY ``` @@ -150,4 +199,6 @@ hermes chat -m qwen3.7-max - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md index aec56898..4f710ed8 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md @@ -1,6 +1,6 @@ # Kilo CLI -Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 Kilo CLI** @@ -23,6 +23,8 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding 使用文本编辑器打开配置文件 `~/.config/kilo/config.json`，根据所选方案写入对应配置： +- **Token Plan 个人版**：个人订阅，按 token 消耗抵扣 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -30,6 +32,89 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +需先购买 Token Plan 个人版套餐且套餐处于有效期内。可在[Token Plan 个人版页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)购买套餐。 + +将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型：qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro，完整列表请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +``` +{ + "$schema": "https://kilo.ai/config.json", + "provider": { + "bailian-token-plan-personal": { + "npm": "@ai-sdk/openai-compatible", + "name": "Alibaba Cloud Model Studio", + "options": { + "baseURL": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "apiKey": "YOUR_API_KEY" + }, + "models": { + "qwen3.8-max-preview": { + "name": "Qwen3.8 Max Preview", + "contextWindow": 983616, + "maxOutputTokens": 131072, + "reasoning": true, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 262144 + } + } + }, + "qwen3.7-max": { + "name": "Qwen3.7 Max", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.7-plus": { + "name": "Qwen3.7 Plus", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.6-plus": { + "name": "Qwen3.6 Plus", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.6-flash": { + "name": "Qwen3.6 Flash", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + } + } + } + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Token Plan 团队版 需先购买 Token Plan 团队版套餐且套餐处于有效期内。可在[Token Plan 团队版页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/overview)购买套餐。 @@ -48,6 +133,18 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding "apiKey": "YOUR_API_KEY" }, "models": { + "qwen3.8-max-preview": { + "name": "Qwen3.8 Max Preview", + "contextWindow": 983616, + "maxOutputTokens": 131072, + "reasoning": true, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 262144 + } + } + }, "qwen3.7-max": { "name": "Qwen3.7 Max", "options": { @@ -156,6 +253,17 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding } ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Coding Plan 将 `YOUR_API_KEY` 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan-overview#b01f82a4218kx)。 @@ -242,11 +350,11 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding 将 `YOUR_API_KEY` 替换为[阿里云百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。可用模型请参考[OpenAI 兼容 - 支持的模型](https://help.aliyun.com/zh/model-studio/anthropic-api-messages#07833dedefft7)。 -`baseURL` 按地域设置，API Key 需与所选地域对应： +`baseURL` 按地域设置（请将 URL 中的 `{WorkspaceId}` 替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)），API Key 需与所选地域对应： -- 华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` @@ -259,7 +367,7 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding "npm": "@ai-sdk/openai-compatible", "name": "Alibaba Cloud Model Studio", "options": { - "baseURL": "https://dashscope.aliyuncs.com/compatible-mode/v1", + "baseURL": "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", "apiKey": "YOUR_API_KEY" }, "models": { @@ -305,4 +413,6 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md index addc7866..c144d889 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md @@ -1,6 +1,6 @@ # Qoder CN（原 Lingma） -Qoder CN（原 Lingma）是阿里云智能编码助手，提供独立 IDE，可以通过 Token Plan、Coding Plan 或按量付费接入阿里云百炼。 +Qoder CN（原 Lingma）是阿里云智能编码助手，提供独立 IDE，可以通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量付费接入阿里云百炼。 **说明** @@ -31,7 +31,7 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 类型 - 根据计费方案选择 **Token Plan**、**Coding Plan** 或 **按量付费** + 根据计费方案选择 **Token Plan**（个人版或团队版）、**Coding Plan** 或 **按量付费** 模型 @@ -41,7 +41,9 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 填写对应方案的专属 API Key： - - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list) + - Token Plan 个人版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal) + + - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise) - Coding Plan：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan) @@ -55,7 +57,7 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 ## 了解更多 -如需进一步了解 Qoder CN 的智能体、MCP、Skills 等扩展能力，请参考 [Qoder CN 官方文档](https://help.aliyun.com/zh/lingma/product-overview/introduction-of-lingma)。 +如需进一步了解 Qoder CN 的智能体、MCP、Skills 等扩展能力，请参考 [Qoder CN 官方文档](https://help.aliyun.com/zh/lingma/introduction-of-lingma)。 ## 常见问题 @@ -65,7 +67,9 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) - 按量计费：[错误码](https://help.aliyun.com/zh/model-studio/error-code) @@ -85,7 +89,7 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 - **提供商或类型与实际套餐不一致**：在 Qoder CN 模型配置中，**提供商**与**类型**需与所购套餐保持一致。例如使用 Token Plan 团队版的 API Key，但**类型**选成了 Coding Plan。 -- **选用了套餐不支持的模型**：仅支持当前套餐覆盖的文本生成模型。例如，Token Plan 团队版的支持模型列表可在[Token Plan 团队版页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/overview)查看。 +- **选用了套餐不支持的模型**：仅支持当前套餐覆盖的文本生成模型。例如，Token Plan 团队版的支持模型列表可在[Token Plan 团队版页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan/enterprise)查看。 - **临时网络或服务波动**：稍后重试。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md index d3f45eae..ff17120c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md @@ -1,9 +1,31 @@ # 更多工具 -除已列出的工具外，阿里云百炼还支持接入兼容 OpenAI / Anthropic API 协议且支持自定义服务端点的第三方编程工具。可通过按量计费、Coding Plan 或 Token Plan 团队版接入。 +除已列出的工具外，阿里云百炼还支持接入兼容 OpenAI / Anthropic API 协议且支持自定义服务端点的第三方编程工具。可通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入。 ## **配置接入凭证** +### Token Plan 个人版 + +**API 协议** + +**Base URL** + +**API Key** + +**支持模型** + +OpenAI + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)（仅文本生成类） + +Anthropic + +`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` + ### Token Plan 团队版 **API 协议** @@ -60,9 +82,11 @@ Anthropic OpenAI -华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` +根据地域，填入对应 URL（请将 URL 中的 `{WorkspaceId}` 替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)）： + +华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -新加坡：`https://[workspace-id].ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` +新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` @@ -72,9 +96,9 @@ OpenAI Anthropic -华北2（北京）：`https://dashscope.aliyuncs.com/apps/anthropic` +华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic` -新加坡：`https://[workspace-id].ap-southeast-1.maas.aliyuncs.com/apps/anthropic` +新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic` [支持的模型](https://help.aliyun.com/zh/model-studio/anthropic-api-messages#ae1b2c3d4e5f6) @@ -97,7 +121,7 @@ Trae 支持接入自定义模型，无需安装插件即可直接配置上述任 ## **不支持的工具类型** -Token Plan 团队版和 Coding Plan 仅限在 AI 编程工具和 OpenClaw 类型 Agent 中使用，以下类型的工具**不支持**接入： +Token Plan 个人版、Token Plan 团队版和 Coding Plan 仅限在 AI 编程工具和 OpenClaw 类型 Agent 中使用，以下类型的工具**不支持**接入： - **工作流/自动化平台**：如 Dify、n8n、Coze 等。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md index 6f73f11e..a1252907 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md @@ -1,6 +1,6 @@ # OpenClaw -OpenClaw 是一个开源的个人 AI 助手平台，支持通过多种消息渠道与 AI 交互。通过配置可接入阿里云百炼平台上的 AI 模型，支持按量付费、Coding Plan、Token Plan 团队版三种接入方式。 +OpenClaw 是一个开源的个人 AI 助手平台，支持通过多种消息渠道与 AI 交互。通过配置可接入阿里云百炼平台上的 AI 模型，支持按量付费、Coding Plan、Token Plan 个人版、Token Plan 团队版四种接入方式。 ## **安装 OpenClaw** @@ -84,6 +84,142 @@ How do you want to hatch your bot? ## **配置接入凭证** +### **Token Plan 个人版** + +将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型包括 qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro，完整列表请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +**API Key** + +Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +**Base URL** + +`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` + +**可用模型** + +Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview) + +配置文件位于 `~/.openclaw/openclaw.json`，OpenClaw 启动时会自动读取。 + +**说明** + +示例禁用了网关鉴权（`auth.mode: none`），仅适合单机本地使用。如需共享或远程访问，请运行 `openclaw doctor --fix` 启用 token 鉴权。 + +**首次配置**：复制以下内容到配置文件，将 `YOUR_API_KEY` 替换为 Token Plan 个人版 API Key。 + +**已有配置**：若需保留已有配置，请勿直接全量替换，详见[已有配置如何安全修改](#cp-openclaw-faq-safe-modify)。 + +``` +{ + "meta": { + "lastTouchedVersion": "2026.2.1", + "lastTouchedAt": "2026-02-03T08:20:00.000Z" + }, + "models": { + "mode": "merge", + "providers": { + "bailian-token-plan": { + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic", + "apiKey": "YOUR_API_KEY", + "api": "anthropic-messages", + "models": [ + { + "id": "qwen3.8-max-preview", + "name": "qwen3.8-max-preview", + "reasoning": true, + "input": ["text", "image"], + "contextWindow": 983616, + "maxTokens": 131072, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "qwen3.7-max", + "name": "qwen3.7-max", + "reasoning": false, + "input": ["text"], + "contextWindow": 1000000, + "maxTokens": 65536, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "qwen3.7-plus", + "name": "qwen3.7-plus", + "reasoning": false, + "input": ["text", "image"], + "contextWindow": 1000000, + "maxTokens": 65536, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "qwen3.6-flash", + "name": "qwen3.6-flash", + "reasoning": false, + "input": ["text", "image"], + "contextWindow": 1000000, + "maxTokens": 32768, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "glm-5.2", + "name": "glm-5.2", + "reasoning": false, + "input": ["text"], + "contextWindow": 1000000, + "maxTokens": 16384, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "deepseek-v4-pro", + "name": "deepseek-v4-pro", + "reasoning": false, + "input": ["text"], + "contextWindow": 163840, + "maxTokens": 32768, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 } + } + ] + } + } + }, + "agents": { + "defaults": { + "model": { + "primary": "bailian-token-plan/qwen3.8-max-preview" + }, + "models": { + "bailian-token-plan/qwen3.8-max-preview": {}, + "bailian-token-plan/qwen3.7-max": {}, + "bailian-token-plan/qwen3.7-plus": {}, + "bailian-token-plan/qwen3.6-flash": {}, + "bailian-token-plan/glm-5.2": {}, + "bailian-token-plan/deepseek-v4-pro": {} + } + } + }, + "gateway": { + "mode": "local", + "auth": { "mode": "none" } + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### **Token Plan 团队版** 将 `YOUR_API_KEY` 替换为 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。可用模型请参考 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。 @@ -134,6 +270,16 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to "apiKey": "YOUR_API_KEY", "api": "anthropic-messages", "models": [ + { + "id": "qwen3.8-max-preview", + "name": "qwen3.8-max-preview", + "reasoning": true, + "input": ["text", "image"], + "contextWindow": 983616, + "maxTokens": 131072, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, { "id": "qwen3.7-max", "name": "qwen3.7-max", @@ -278,9 +424,10 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to "agents": { "defaults": { "model": { - "primary": "bailian-token-plan/qwen3.7-plus" + "primary": "bailian-token-plan/qwen3.8-max-preview" }, "models": { + "bailian-token-plan/qwen3.8-max-preview": {}, "bailian-token-plan/qwen3.7-max": {}, "bailian-token-plan/qwen3.7-plus": {}, "bailian-token-plan/qwen3.6-plus": {}, @@ -350,6 +497,16 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to "apiKey": "YOUR_API_KEY", "api": "anthropic-messages", "models": [ + { + "id": "qwen3.8-max-preview", + "name": "qwen3.8-max-preview", + "reasoning": true, + "input": ["text", "image"], + "contextWindow": 983616, + "maxTokens": 131072, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, { "id": "qwen3.7-max", "name": "qwen3.7-max", @@ -494,9 +651,10 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to "agents": { "defaults": { "model": { - "primary": "bailian-token-plan/qwen3.7-plus" + "primary": "bailian-token-plan/qwen3.8-max-preview" }, "models": { + "bailian-token-plan/qwen3.8-max-preview": {}, "bailian-token-plan/qwen3.7-max": {}, "bailian-token-plan/qwen3.7-plus": {}, "bailian-token-plan/qwen3.6-plus": {}, @@ -526,6 +684,17 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to 先单击 **Save** 按钮将配置写入磁盘，再单击 **Apply** 按钮重启网关使配置生效。 +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### **Coding Plan** 将 `YOUR_API_KEY` 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)（格式为 `sk-sp-xxxxx`）。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -888,11 +1057,11 @@ Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-pla **Base URL** -请确保 Base URL、API Key 和模型归属同一地域： +请确保 Base URL、API Key 和模型归属同一地域，并将 `WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) -- 华北2（北京）：`https://dashscope.aliyuncs.com/apps/anthropic` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic` **可用模型** @@ -931,7 +1100,7 @@ Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-pla "mode": "merge", "providers": { "bailian": { - "baseUrl": "https://dashscope.aliyuncs.com/apps/anthropic", + "baseUrl": "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", "apiKey": "YOUR_API_KEY", "api": "anthropic-messages", "models": [ @@ -1039,7 +1208,7 @@ Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-pla "mode": "merge", "providers": { "bailian": { - "baseUrl": "https://dashscope.aliyuncs.com/apps/anthropic", + "baseUrl": "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic", "apiKey": "YOUR_API_KEY", "api": "anthropic-messages", "models": [ @@ -1215,6 +1384,14 @@ Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-pla 以上配置中 `dmPolicy` 和 `groupPolicy` 均设为 `open`，适用于测试或个人使用场景。生产环境中建议设为 `allowlist`，通过白名单限制可访问的用户和群组，降低安全风险。 +切换到 `allowlist` 模式时，将 `dmPolicy` 和 `groupPolicy` 改为 `"allowlist"`，并添加 `allowFrom` 字段，填入允许访问的工号和群 ID。示例： + +``` +"dmPolicy": "allowlist", +"groupPolicy": "allowlist", +"allowFrom": ["你的工号", "群ID"] +``` + #### 步骤四：测试 1. 执行以下命令重启网关。 @@ -1790,7 +1967,7 @@ Skill 是可扩展的能力模块，Agent 会根据请求自动匹配并加载 ### 接入 MCP 服务 -OpenClaw 支持通过 MCP（Model Context Protocol）插件扩展 Agent 的工具调用能力，例如联网搜索、网页抓取等。具体案例可以参考[添加联网搜索MCP](https://help.aliyun.com/zh/model-studio/web-search-for-coding-plan)。 +OpenClaw 支持通过 MCP（Model Context Protocol）插件扩展 Agent 的工具调用能力，例如联网搜索、网页抓取等。具体案例可以参考[添加联网搜索MCP](https://help.aliyun.com/zh/model-studio/web-search-mcp)。 ## 常见问题 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/opencode.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/opencode.md index 2f2b6d2a..e8a3a751 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/opencode.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/opencode.md @@ -1,6 +1,6 @@ # OpenCode -OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 OpenCode** @@ -30,6 +30,8 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla 根据所选方案写入对应配置： +- **Token Plan 个人版**：个人订阅，按 token 消耗抵扣 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -37,6 +39,102 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +需先购买 Token Plan 个人版套餐且套餐处于有效期内。可在[Token Plan 个人版页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)购买套餐。 + +将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型包括 qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro，详细列表请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +``` +{ + "$schema": "https://opencode.ai/config.json", + "provider": { + "bailian-token-plan-personal": { + "npm": "@ai-sdk/anthropic", + "name": "Alibaba Cloud Model Studio", + "options": { + "baseURL": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic/v1", + "apiKey": "YOUR_API_KEY" + }, + "models": { + "qwen3.8-max-preview": { + "name": "Qwen3.8 Max Preview", + "contextWindow": 983616, + "maxOutputTokens": 131072, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 99072 + }, + "temperature": 0.6, + "reasoning": true + } + }, + "qwen3.7-max": { + "name": "Qwen3.7 Max", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.7-plus": { + "name": "Qwen3.7 Plus", + "modalities": { + "input": ["text", "image"], + "output": ["text"] + }, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.6-plus": { + "name": "Qwen3.6 Plus", + "modalities": { + "input": ["text", "image"], + "output": ["text"] + }, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.6-flash": { + "name": "Qwen3.6 Flash", + "modalities": { + "input": ["text", "image"], + "output": ["text"] + }, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + } + } + } + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Token Plan 团队版 需先购买 Token Plan 团队版套餐且套餐处于有效期内。可在[Token Plan 团队版页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/overview)购买套餐。 @@ -55,6 +153,19 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla "apiKey": "YOUR_API_KEY" }, "models": { + "qwen3.8-max-preview": { + "name": "Qwen3.8 Max Preview", + "contextWindow": 983616, + "maxOutputTokens": 131072, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 99072 + }, + "temperature": 0.6, + "reasoning": true + } + }, "qwen3.7-max": { "name": "Qwen3.7 Max", "options": { @@ -187,6 +298,17 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla } ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Coding Plan 将 `YOUR_API_KEY` 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -289,11 +411,11 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla 将 `YOUR_API_KEY` 替换为[阿里云百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。可用模型请参考[Anthropic 兼容 API](https://help.aliyun.com/zh/model-studio/anthropic-api-messages#07833dedefft7)。 -`baseURL` 按地域设置，API Key 需与所选地域对应： +`baseURL` 按地域设置（将 `{WorkspaceId}` 替换为真实的 [Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)），API Key 需与所选地域对应： -- 华北2（北京）：`https://dashscope.aliyuncs.com/apps/anthropic/v1` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` ``` @@ -301,10 +423,10 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla "$schema": "https://opencode.ai/config.json", "provider": { "bailian-payg": { - "npm": "@ai-sdk/anthropic", + "npm": "@ai-sdk/openai-compatible", "name": "Alibaba Cloud Model Studio", "options": { - "baseURL": "https://dashscope.aliyuncs.com/apps/anthropic/v1", + "baseURL": "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", "apiKey": "YOUR_API_KEY" }, "models": { @@ -368,4 +490,6 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md index 1de85109..4540e5fb 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md @@ -1,6 +1,6 @@ # Qoder -Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 JetBrains 插件，可以通过按量付费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 JetBrains 插件，可以通过按量付费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## Qoder IDE @@ -27,7 +27,7 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 类型 - 根据计费方案选择 **Token Plan**、**Coding Plan** 或 **按量付费** + 根据计费方案选择 **Token Plan**（个人版或团队版）、**Coding Plan** 或 **按量付费** 模型 @@ -37,7 +37,9 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 填写对应方案的专属 API Key： - - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list) + - Token Plan 个人版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal) + + - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise) - Coding Plan：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan) @@ -99,7 +101,7 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 1. 在对话框中输入 `/model`，通过 Tab 键切换至 `Custom`。 -2. 回车选择 Add custom model，提供商选择 Alibaba Cloud Model Studio - China，类型根据计费方案选择 **Token Plan**、**Coding Plan** 或 **按量付费**。 +2. 回车选择 Add custom model，提供商选择 Alibaba Cloud Model Studio - China，类型根据计费方案选择 **Token Plan**（个人版或团队版）、**Coding Plan** 或 **按量付费**。 3. 选择模型后输入对应方案的专属 API Key，确认后等待配置生效。 @@ -135,7 +137,7 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 类型 - 根据计费方案选择 **Token Plan**、**Coding Plan** 或 **按量付费** + 根据计费方案选择 **Token Plan**（个人版或团队版）、**Coding Plan** 或 **按量付费** 模型 @@ -145,7 +147,9 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 填写对应方案的专属 API Key： - - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list) + - Token Plan 个人版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal) + + - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise) - Coding Plan：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan) @@ -195,7 +199,11 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- 按量计费：[错误码](https://help.aliyun.com/zh/model-studio/error-code) ### 为什么在 Qoder 设置中找不到模型选项？ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md index 7ca6ebb6..ec3143da 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md @@ -1,6 +1,6 @@ # Qwen Code -Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 Qwen Code** @@ -33,8 +33,10 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl ## **配置接入凭证** -启动 Qwen Code 后输入 `/auth` 命令进行可视化配置。阿里云百炼提供三种计费方案，根据需要选择： +启动 Qwen Code 后输入 `/auth` 命令进行可视化配置。阿里云百炼提供四种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -42,6 +44,98 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +启动 Qwen Code 后输入 `/auth`，依次选择 **订阅计划** > **阿里云百炼 Token Plan**，输入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) 即可完成配置。可用模型请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +高级配置：通过 settings.json 配置文件 + +编辑或新建 `settings.json` 文件，将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 API Key。文件路径如下： + +- macOS/Linux：`~/.qwen/settings.json` + +- Windows：`C:\Users\\.qwen\settings.json` + + +``` +{ + "env": { + "BAILIAN_TOKEN_PLAN_API_KEY": "YOUR_API_KEY" + }, + "modelProviders": { + "openai": [ + { + "id": "qwen3.8-max-preview", + "name": "[Token Plan 个人版] qwen3.8-max-preview", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, + { + "id": "qwen3.7-max", + "name": "[Token Plan 个人版] qwen3.7-max", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, + { + "id": "qwen3.7-plus", + "name": "[Token Plan 个人版] qwen3.7-plus", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, + { + "id": "qwen3.6-flash", + "name": "[Token Plan 个人版] qwen3.6-flash", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, + { + "id": "glm-5.2", + "name": "[Token Plan 个人版] glm-5.2", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY" + }, + { + "id": "deepseek-v4-pro", + "name": "[Token Plan 个人版] deepseek-v4-pro", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY" + } + ] + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Token Plan 团队版 启动 Qwen Code 后输入 `/auth`，依次选择 **订阅计划** > **阿里云百炼 Token Plan**，输入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list) 即可完成配置。可用模型请参考 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。 @@ -62,6 +156,17 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl }, "modelProviders": { "openai": [ + { + "id": "qwen3.8-max-preview", + "name": "[Token Plan 团队版] qwen3.8-max-preview", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, { "id": "qwen3.7-max", "name": "[Token Plan 团队版] qwen3.7-max", @@ -207,12 +312,23 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl "region": "china" }, "model": { - "name": "qwen3.7-plus" + "name": "qwen3.8-max-preview" }, "$version": 3 } ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、medium、low，默认 xhigh。 + + ### Coding Plan 启动 Qwen Code 后输入 `/auth`，依次选择 **订阅计划** > **阿里云百炼 Coding Plan**，选择 Coding Plan 区域（china），输入 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan) 即可完成配置。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -348,11 +464,11 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl - Windows：`C:\Users\\.qwen\settings.json` -`baseUrl` 按地域设置，API Key 需与所选地域对应： +`baseUrl` 按地域设置（URL 中的 `{WorkspaceId}` 需替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)），API Key 需与所选地域对应： -- 华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` @@ -367,7 +483,7 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl { "id": "qwen3.6-plus", "name": "[Bailian] qwen3.6-plus", - "baseUrl": "https://dashscope.aliyuncs.com/compatible-mode/v1", + "baseUrl": "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", "envKey": "BAILIAN_API_KEY", "generationConfig": { "extra_body": { @@ -414,6 +530,27 @@ Qwen Code 支持在 VS Code 中以插件方式使用，在 IDE 中提供 AI 编 3. 点击右上角图标启动 Qwen Code，通过输入或点击`/`，选择`Switch model`切换模型。 +## **Qwen Code 桌面版** + +Qwen Code Desktop 是 Qwen Code 的图形界面应用，内置 CLI 运行时，无需额外安装命令行工具。桌面版与 CLI 共用同一份 `settings.json` 配置文件，如果已通过 CLI 完成配置，桌面版打开即可直接使用，无需重复配置。 + +如果尚未配置，可通过桌面版图形界面完成： + +1. 打开 Qwen Code Desktop，点击左侧**设置**，选择 **AI** 页签。 + +2. 在**模型 Provider**区域点击**\+ 连接**。 + +3. 在弹窗中选择 **Alibaba ModelStudio** 页签，根据计费方案选择对应项： + + - **Coding Plan**：个人开发者，按次计量。 + + - **Token Plan**：团队使用，按 token 消耗抵扣。 + + - **Standard API Key**：使用已有的百炼 API Key 按量计费。 + +4. 按照页面提示输入 API Key，完成配置。 + + ## **常见命令** **说明** @@ -547,7 +684,7 @@ Qwen Code 支持在 VS Code 中以插件方式使用，在 IDE 中提供 AI 编 4. 输入以下内容安装 skill。 ``` - 查看我是否有find skills，没有就直接帮我安装：npx skills add https://github.com/vercel-labs/skills --skill find-skills -y -a qwen-code，然后帮我安装 web-component-design 到当前目录qwen code skills中。 + 查看我是否有find skills，没有就直接帮我安装：npx skills add https://github.com/vercel-labs/skills --skill find-skills -y -a qwen-code，然后从 wshobson/agents 帮我安装 web-component-design 到当前目录：npx skills add https://github.com/wshobson/agents --skill web-component-design -y ``` 5. 下载[website.png](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260318/ymehla/website.png)到项目目录，输入以下内容，将自动识别截图的布局、样式，生成网页代码。 @@ -593,7 +730,9 @@ Qwen Code 支持在 VS Code 中以插件方式使用，在 IDE 中提供 AI 编 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### **如何切换模型？** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md index 00dddafd..fd715d26 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md @@ -1,6 +1,6 @@ # QwenPaw -QwenPaw（原 CoPaw）是 AgentScope 团队开源的个人 AI 助手，支持本地或云端部署，可通过 Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 +QwenPaw（原 CoPaw）是 AgentScope 团队开源的个人 AI 助手，支持本地或云端部署，可通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **安装 QwenPaw** @@ -52,6 +52,22 @@ qwenpaw app 在 Console 点击 **设置** > **模型**，根据计费方案配置对应的提供商。 +### Token Plan 个人版 + +进入内置的 **Aliyun Token Plan** 提供商**设置**页面，填入 API Key。 + +**配置项** + +**说明** + +**API 密钥** + +填入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)。 + +**模型** + +已预设常用模型。新增模型点击**添加模型**，**模型 ID**填入 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + ### Token Plan 团队版 进入内置的 **Aliyun Token Plan** 提供商**设置**页面，填入 API Key。 @@ -62,7 +78,7 @@ qwenpaw app **API 密钥** -填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。 +填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise)。 **模型** @@ -98,9 +114,9 @@ qwenpaw app **基础 URL** -根据模型部署地域，在下拉菜单选择对应 URL： +根据模型部署地域，在下拉菜单选择对应 URL（请将 URL 中的 `{WorkspaceId}` 替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)）： -- China (Beijing)：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- China (Beijing)：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - International (Singapore)：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，将 `WorkspaceId` 替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 @@ -125,7 +141,9 @@ qwenpaw app - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 报错 401 Incorrect API key provided diff --git a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md index 74caf7e8..b839ecb8 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md @@ -1,88 +1,73 @@ # 3d generation -百炼平台基于 Tripo 模型提供 3D 资产生成能力，支持文生 3D、单图生 3D 与多图生 3D 三种输入方式，产出带贴图的 PBR 材质 GLB 模型或无贴图基础模型。由于生成耗时较长，API 采用[异步调用](../concepts/async-invocation.md)，整体流程为「创建任务 → 轮询获取结果」。详细接口与参数见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 适用范围 - -- 仅适用于**华北2（北京）**地域，且必须使用该地域的 [API Key](../concepts/api-key.md)。 -- 需先在百炼控制台模型市场搜索「Tripo」并开通服务、完成授权，再配置好 [API Key](../concepts/api-key.md) 环境变量。具体开通与配置步骤见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 调用流程 - -API 仅支持[异步调用](../concepts/async-invocation.md)，包含两个步骤： - -1. **创建任务**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` -2. **轮询查询结果**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` - -创建任务时必须携带 `X-DashScope-Async: enable` 请求头，否则会报错 `current user api does not support synchronous calls`。成功创建后返回 `task_id`，有效期 24 小时，**请勿重复创建任务**，轮询获取即可。 - -轮询建议间隔约 15 秒，任务状态流转为 `PENDING`（排队中）→ `RUNNING`（处理中）→ `SUCCEEDED` / `FAILED`。查询接口默认 RPS 为 20，如需更高频查询或事件通知建议配置异步任务回调。 - -## 支持的模型 - -| 模型名 | 定位 | 输出面数 | 对应官方 API 版本 | -| --- | --- | --- | --- | -| `Tripo/Tripo-H3.1` | 高精度生成 | 最高 200 万面 | `v3.1-20260211` | -| `Tripo/Tripo-P1.0` | 专业生成，速度更快 | 最高 2 万面 | `P1-20260311` | - -## 输入方式 - -`input` 中 `prompt`、`image`、`images` 三者**互斥**，只能选其一，同时传多个将报错。 - -- **文生 3D**：`prompt` 必填，支持中英文等多语言，每个字符计 1 个字符，最大 1024 字符。 -- **单图生 3D**：`image` 必填，传入单张图像公网 URL。图像格式限 JPEG/PNG，宽高范围 [20, 6000] 像素（建议边长大于 256），文件不超过 20MB，支持 HTTP/HTTPS。 -- **多图生 3D**：`images` 必填，数组长度固定为 4，对应视角顺序为**前、左、后、右**；不需要的视角传空对象 `{}`。实际有效图片数为 2~4 张，多张图像的分辨率和宽高比不要求一致。每个对象含 `type`（`jpeg` 或 `png`）与 `file_token`（公网 URL）字段。 - -## 关键参数（parameters） - -| 参数 | 适用模型 | 默认值 | 说明 | -| --- | --- | --- | --- | -| `texture_quality` | 全部 | `standard` | 贴图质量，可选 `standard`（标清）/`detailed`（高清） | -| `geometry_quality` | `Tripo/Tripo-H3.1` | `standard` | 几何精度，`standard` 最高 150 万面，`ultra` 最高 200 万面 | -| `pbr` | 全部 | `true` | 是否生成 PBR 材质模型。设为 `true` 时强制启用贴图，返回 `pbr_model_url` | -| `texture` | 全部 | `true` | 是否生成贴图。生成无贴图模型需**同时**将 `texture` 和 `pbr` 设为 `false`，返回 `base_model_url` | - -## 响应与产物 - -成功响应的 `output.results` 仅在 `task_status` 为 `SUCCEEDED` 时返回，包含以下字段： - -- `pbr_model_url`：PBR 材质模型（GLB）下载 URL，当 `pbr` 为 `true`（默认）时返回。 -- `base_model_url`：无贴图基础模型（GLB）下载 URL，当 `texture` 与 `pbr` 均为 `false` 时返回。 -- `rendered_image_url`：3D 模型预览渲染图（1 张）URL。 - -> **注意**：以上下载链接有效期均为 **2 小时**，请及时下载。 - -`usage` 字段记录任务类型（`text-to-3d` / `image-to-3d` / `multi-image-to-3d`）、生成数量 `count`、贴图质量与几何精度，仅对成功结果计数。任务状态 `task_status` 的完整枚举为 `PENDING` / `RUNNING` / `SUCCEEDED` / `FAILED` / `CANCELED` / `UNKNOWN`，其中 `UNKNOWN` 表示任务不存在或超过 24 小时有效期。更多响应字段说明见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 限制与注意事项 - -- 仅限北京地域 [API Key](../concepts/api-key.md) 调用，地域不匹配将无法使用。 -- `task_id` 查询有效期 24 小时，超时返回 `UNKNOWN` 且无法再查询。 -- 查询接口默认 RPS 限制为 20，建议通过异步任务回调获取更高频通知。 -- 产物下载链接有效期仅 2 小时。 -- 调用失败时响应中会返回 `code` 与 `message`，可参照百炼错误码文档排查。 +百炼平台提供基于 Tripo 模型的 3D 模型生成能力，支持文生3D、单图生3D 和多图生3D 三种输入模式，输出 GLB 格式模型（含 PBR 材质或无贴图基础模型）及预览渲染图。所有调用均为[异步任务](../concepts/asynchronous-task.md)，需通过 `task_id` 轮询获取结果，且**仅限华北2（北京）地域可用**。开发者需提前开通服务并配置对应地域的 API Key。 + +## 支持的模型/功能 + +- **模型标识**： + - `Tripo/Tripo-H3.1`：高精度生成，最高 200 万面，支持 `geometry_quality: "ultra"`；对应 Tripo 官方 API 版本 `v3.1-20260211`。 + - `Tripo/Tripo-P1.0`：专业级生成，最高 2 万面，推理更快；对应 Tripo 官方 API 版本 `P1-20260311`。 +- **输入模式（三者互斥）**： + - 文生3D：通过 `input.prompt` 输入文本描述（≤1024 字符，支持中英文）； + - 单图生3D：通过 `input.image` 传入单张公网 JPEG/PNG 图像（分辨率 [20, 6000] 像素，≤20MB）； + - 多图生3D：通过 `input.images` 传入长度为 4 的数组，顺序固定为「前、左、后、右」；缺失视角需填空对象 `{}`；实际有效图数为 2–4 张。 +- **输出类型**： + - 默认返回带 PBR 材质的 GLB 模型（`pbr_model_url`）和预览图（`rendered_image_url`）； + - 可通过设置 `"texture": false, "pbr": false` 获取无贴图基础模型（`base_model_url`），详见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 + +## 关键参数 + +| 参数 | 类型 | 必选 | 说明 | +|------|------|------|------| +| `model` | string | ✅ | 固定为 `Tripo/Tripo-H3.1` 或 `Tripo/Tripo-P1.0` | +| `input.prompt` / `input.image` / `input.images` | string / object / array | ✅（三选一） | 仅允许一种输入方式；`images` 数组长度必须为 4，含空对象占位 | +| `parameters.texture_quality` | string | ❌ | `"standard"`（默认）或 `"detailed"`；仅对带贴图任务生效 | +| `parameters.geometry_quality` | string | ❌ | 仅 `Tripo/Tripo-H3.1` 支持；`"standard"`（≤150 万面）或 `"ultra"`（≤200 万面） | +| `parameters.pbr` | boolean | ❌ | 默认 `true`；设为 `false` 时将强制禁用贴图（需同时设 `texture: false`） | +| `parameters.texture` | boolean | ❌ | 默认 `true`；与 `pbr` 联动，二者同为 `false` 才返回 `base_model_url` | + +> **注意**：原始文档中 `parameters.texture` 和 `parameters.pbr` 的联动逻辑明确要求“需同时设为 `false`”才生成无贴图模型，该规则在 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中多次强调，务必遵守。 + +## 使用方式 + +1. **开通与认证**： + - 在[百炼控制台（华北2）](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all) 搜索 “Tripo” 并开通服务； + - 获取并配置该地域的 [API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)，确保环境变量 `DASHSCOPE_API_KEY` 已设置。 + +2. **创建任务（POST）**： + - Endpoint（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` + - **必需请求头**：`Content-Type: application/json`、`Authorization: Bearer $DASHSCOPE_API_KEY`、`X-DashScope-Async: enable` + - 请求体示例（文生3D）： + ```json + { + "model": "Tripo/Tripo-P1.0", + "input": { "prompt": "一只可爱的猫" }, + "parameters": { "texture_quality": "standard" } + } + ``` + - 成功响应返回 `task_id`（有效期 24 小时），请妥善保存。 + +3. **轮询查询结果（GET）**： + - Endpoint（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + - 仅需 `Authorization` 请求头； + - 建议轮询间隔 ≥15 秒；状态流转为 `PENDING` → `RUNNING` → `SUCCEEDED`/`FAILED`； + - 成功时 `output.results` 包含 `pbr_model_url`（或 `base_model_url`）、`rendered_image_url`，**所有 URL 有效期仅 2 小时**，需及时下载。 + +详细调用流程与错误处理可参考 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中的完整示例与错误码说明。 + +## 限制和注意事项 + +- **地域强约束**：仅支持华北2（北京）地域，其他地域 URL 或 API Key 均不可用； +- **异步强制性**：`X-DashScope-Async: enable` 为必填请求头，缺失将报错 `current user api does not support synchronous calls`； +- **输入互斥性**：`prompt`、`image`、`images` 三者不可共存，否则返回 `InvalidParameter`； +- **多图格式要求**：`input.images` 必须为长度 4 的数组，顺序固定为「前、左、后、右」；传入非空对象时 `type` 必须为 `"jpeg"` 或 `"png"`，`file_token` 必须为公网可访问 HTTPS/HTTP URL； +- **资源时效性**： + - `task_id` 查询有效期：24 小时； + - 模型/渲染图下载 URL 有效期：2 小时； +- **RPS 限制**：任务查询接口默认 RPS 为 20，高频轮询建议配置[异步回调](https://help.aliyun.com/zh/model-studio/async-task-api)替代。 ## 来源文档 - [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md index eb343a80..30643c2c 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md @@ -1,135 +1,51 @@ # application call -阿里云百炼平台提供两套 API 来调用智能体和工作流应用：**OpenAI 兼容的 Responses API** 和 **DashScope API**。两者均支持同步/[异步调用](../concepts/async-invocation.md)、多轮对话、[流式输出](../concepts/streaming.md)等核心能力，开发者可根据生态兼容性和功能需求选择合适的接入方式。调用前需先获取 APP ID（以及子[业务空间](../concepts/workspace.md)场景下的 Workspace ID）和 [API Key](../concepts/api-key.md)。 +`application call` 是阿里云百炼平台提供的核心能力，用于通过 API 调用已发布的智能体（Agent）或工作流（Workflow）应用。开发者可通过 OpenAI 兼容的 Responses API 或原生 DashScope API 两种方式发起同步或异步请求，支持文本、图像、文件等[多模态输入](../concepts/multi-modal-input.md)，并可维护会话上下文。所有调用均需提供有效的 APP ID 及认证凭据。 -## 前置准备 +## 支持的模型/功能 -### 获取凭证 +- **应用类型**：支持新版智能体（Agent 2.0）、旧版智能体及工作流三类应用，但功能支持存在差异： + - 图像输入仅在选用通义千问 VL 系列模型且完成对应配置（如智能体设为“自定义处理”，工作流模型节点入参填 `imageList`）后生效 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)； + - 文件输入**仅限智能体应用**，且需在应用内选择“全文引用”或“切片检索”文件处理方式 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)； + - 多轮对话在 DashScope API 中通过 `session_id` 实现（有效期 1 小时），而 Responses API 当前**不支持 `pre_response_id` 或 `conversation_id`**，需显式传递完整历史消息 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 -通过 API 调用应用时，必须提供 **APP ID** 来指定目标应用。如果应用位于子[业务空间](../concepts/workspace.md)，还需提供 **Workspace ID**。详细获取方式参见 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 +- **调用模式**： + - **同步调用**：适用于实时交互场景，API 阻塞等待结果返回； + - **异步调用**：适用于耗时较长任务（如报告生成），立即返回任务 ID，后续通过 `retrieve` 查询状态；**异步模式不支持[流式输出](../concepts/streaming-output.md)** [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 -- **APP ID**：在控制台「应用管理」页面的应用卡片上复制。 -- **Workspace ID**：在调用子[业务空间](../concepts/workspace.md)下的应用或特定地域（德国、华北2、新加坡、日本）的模型时必须提供，可通过控制台右上角图标查看。 +> **注意**：文档 4（新版智能体应用 API 参考）与文档 5（应用 DashScope API 参考）均描述 `/api/v1/apps/{APP_ID}/completion` 接口，但文档 4 明确限定为“新版智能体应用”，而文档 5 未作此限定且标题为“智能体、工作流”，二者适用范围存在矛盾。实际生产中应以应用创建时的类型为准，并优先参考 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) 的通用说明。 -> **注意**：目前只能通过控制台手动获取 APP ID 和 Workspace ID，不支持通过 API 或 CLI 查询。 - -### 其他前提 - -- 已获取 [API Key](../concepts/api-key.md) 并配置到环境变量 `DASHSCOPE_API_KEY`。 -- 已创建并发布百炼应用（智能体或工作流）。 -- 如使用 SDK 调用，需安装对应的 SDK（OpenAI SDK 或 [DashScope SDK](../concepts/dashscope-sdk.md)）。 - -## 两套 API 对比 - -| 维度 | Responses API（OpenAI 兼容） | DashScope API | -|------|---------------------------|---------------| -| Endpoint | `POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` | -| SDK | OpenAI Python/Java SDK | DashScope Python/Java SDK | -| 多轮对话 | 通过 `input` 数组传递完整历史消息 | 通过 `session_id` 或 `messages` 维护上下文 | -| [异步调用](../concepts/async-invocation.md) | 设置 `background=true` | 暂不支持（仅 Responses API 提供） | -| 适用地域 | 仅华北2（北京） | 仅华北2（北京） | - -## Responses API（OpenAI 兼容模式） - -### 同步调用 - -适用于需要即时获取结果的实时交互场景。完整参数说明参见 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 - -**核心请求参数：** +## 关键参数 | 参数 | 类型 | 必选 | 说明 | |------|------|------|------| -| `input` | string / array | 是 | 请求输入，可为简单字符串或包含多轮对话历史的消息数组 | -| `stream` | boolean | 否 | 是否[流式输出](../concepts/streaming.md)，默认 `false` | -| `background` | boolean | 否 | 是否异步执行，默认 `false` | - -**Python 示例（单轮对话）：** - -```python -from openai import OpenAI -import os - -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url=f'https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/' -) - -response = client.responses.create(input="你是谁？") -print(response.model_dump_json(indent=2)) -``` - -**多轮对话**需在 `input` 中传递完整的消息历史数组，每条消息包含 `role`（user/assistant/system）和 `content` 字段。 - -### [多模态](../concepts/multimodal.md)输入 - -Responses API 支持在 `content` 数组中混合多种输入类型： - -- **图像输入**：通过 `input_image` 类型传入图片 URL。[智能体应用](../concepts/agent-application.md)需选用通义千问 VL 系列模型并将文件处理方式设为「自定义处理」。 -- **文件输入**：通过 `input_file` 类型传入文件 URL。仅[智能体应用](../concepts/agent-application.md)支持，需配置「全文引用」或「切片检索」处理方式。 - -### [异步调用](../concepts/async-invocation.md) - -对于耗时较长的任务（如生成报告、多步骤工具调用），可设置 `background=true` 开启异步模式，避免请求超时。详细流程参见 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 - -核心流程： - -1. **创建任务**：请求中设置 `background=true`，API 立即返回任务 ID。 -2. **轮询状态**：通过 `client.responses.retrieve(task_id)` 定期查询任务状态。 -3. **处理结果**：当状态变为 `completed`、`failed` 或 `cancelled` 时获取最终结果。 - -> **注意**：异步任务暂不支持[流式输出](../concepts/streaming.md)（`stream=true`）。 +| `app_id` | string | 是 | 应用唯一标识，在[应用管理](https://bailian.console.aliyun.com/#/app-center)页面获取 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。若应用位于子业务空间或德国（法兰克福）、华北2（北京）、新加坡、日本（东京）地域，还需传入 `workspace_id`。 | +| `input` | string / array | 是 | 核心输入内容：
- 字符串：单轮纯文本（如 `"你好"`）；
- 消息数组：支持多轮对话及多模态（`input_text`/`input_image`/`input_file`）。`input_file` 仅智能体支持。 | +| `stream` | boolean | 否 | 是否[流式输出](../concepts/streaming-output.md)（默认 `false`）。启用需在工作流应用的结束节点开启“[流式输出](../concepts/streaming-output.md)”开关并重新发布 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 | +| `background` | boolean | 否 | 是否异步执行（默认 `false`）。设为 `true` 即触发异步流程，返回任务 ID [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 | +| `biz_params` | object | 否 | 用于向工作流或智能体传递自定义参数（如城市名、索引值），参数名与应用内配置必须一致 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 | -### [流式输出](../concepts/streaming.md) +## 使用方式 -设置 `stream=true` 可边生成边输出，适用于需要实时展示生成内容的场景。若应用类型为工作流，需在结束节点或流程输出节点中启用「[流式输出](../concepts/streaming.md)」开关并重新发布。 +- **Responses API（OpenAI 兼容）**： + - **Endpoint**：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` + - **SDK 初始化**：`base_url = f'https://dashscope.aliyuncs.com/api/v2/apps/agent/{app_id}/compatible-mode/v1/'` + - 支持 Python/Java SDK 及 curl，兼容 OpenAI `client.responses.create()` 调用风格。 -## DashScope API +- **DashScope API（原生）**： + - **Endpoint**：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` + - **SDK 初始化**：使用 `dashscope.Application.call()`（Python）或 `ApplicationParam.builder()`（Java） + - 支持 Python/Java/PHP/Node.js/C#/Go 多语言示例，`input.prompt` 为必填字段。 -DashScope API 提供更全面的功能支持，适合需要深度集成百炼平台能力的场景。支持 Python、Java、PHP、Node.js、C#、Go 等多种语言的 HTTP 调用，以及 Python/Java SDK。详细参数参见 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 +- **在线调试**：所有应用均支持通过控制台 **应用卡片 → 发布 → API 调试** 进行快速验证。 -**Python 示例：** +## 限制和注意事项 -```python -from dashscope import Application -import os - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='APP_ID', - prompt='你是谁？' -) -print(response.output.text) -``` - -**多轮对话**通过 `session_id` 维护上下文：首次请求无需传入，响应中会返回 `session_id`；后续请求携带该值即可延续对话。`session_id` 在最后一次请求后 1 小时内有效。 - -新版[智能体应用](../concepts/agent-application.md)（Agent 2.0）的调用方式与上述基本一致，参见 [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)。 - -## 参数传递 - -### 自定义参数 - -工作流应用中定义的自定义参数，通过请求体中的 `biz_params` 传递，参数名和类型需与应用内配置保持一致。 - -```python -response = await client.responses.create( - input="你好", - extra_body={"biz_params": {"city": "北京"}}, - background=True -) -``` - -### 插件参数 - -智能体或工作流中配置的插件工具参数，同样通过 `biz_params` 传递。工作流需在开始节点创建自定义参数并将其传入插件节点的输入参数中。 - -## 限制与注意事项 - -- 两套 API 目前均**仅适用于华北2（北京）地域**。 -- Responses API 的多轮对话暂不支持基于 `pre_response_id` 或 `conversation_id` 的上下文功能，需每次传递完整对话历史。 -- [异步调用](../concepts/async-invocation.md)仅 Responses API 支持，且不能与[流式输出](../concepts/streaming.md)同时使用。 -- RAM 子账号查看[业务空间](../concepts/workspace.md)管理页面需要超级管理员权限（`AliyunBailianFullAccess` 或 `AliyunBailianControlFullAccess`）。 -- [API Key](../concepts/api-key.md) 不建议硬编码到代码中，应通过环境变量配置以降低泄露风险。 +- **地域限制**：Responses API（同步/异步）及新版智能体 API 文档均明确标注**仅适用于华北2（北京）地域**；其他地域调用需确认 Workspace ID 是否已正确嵌入 Base URL [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 +- **凭证获取**：APP ID 和 Workspace ID **仅支持控制台手动获取**，不提供 API 或 CLI 查询接口 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 +- **权限要求**：查询全部 Workspace ID 需主账号或具备 `AliyunBailianFullAccess` 权限的 RAM 子账号；普通子账号仅能查看已加入的业务空间 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 +- **流式与异步互斥**：`stream=true` 与 `background=true` 不可同时设置，否则请求将失败 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 +- **会话管理**：DashScope API 的 `session_id` 机制与 Responses API 的全量消息传递是两种独立方案，不可混用。 ## 来源文档 @@ -140,12 +56,3 @@ response = await client.responses.create( - [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md index 65de9baf..77b370d9 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md @@ -1,232 +1,110 @@ # application component api reference -百炼平台应用组件 API（`bailian/2023-12-29`）提供了数据连接、知识库、Prompt 模板、长期记忆等核心能力的 OpenAPI 接口，采用 ROA 签名风格。开发者可通过阿里云百炼 SDK 直接调用，也可使用自签名方式对接。所有接口均需传入 `WorkspaceId`（[业务空间](../concepts/workspace.md) ID），RAM 子账号需要先获取对应权限策略并加入[业务空间](../concepts/workspace.md)后才能调用。 +本 API 参考文档面向开发者，涵盖百炼平台应用组件（Application Component）的核心 OpenAPI，包括数据连接（原应用数据）、知识库、Prompt 模板三大能力模块。所有接口均基于 `bailian/2023-12-29` 版本，采用 ROA 签名机制，支持 SDK 封装调用与自签名对接。开发者需通过 RAM 权限策略控制访问粒度，并严格遵循业务空间（Workspace）隔离原则。 -## 服务接入点与鉴权 +## 支持的模型/功能 -当前支持两个地域的接入点： +应用组件 API 主要提供三类核心能力： -| 地域 | 地域 ID | 公网接入地址 | VPC 接入地址 | -|------|---------|-------------|-------------| -| 华北2（北京） | cn-beijing | bailian.cn-beijing.aliyuncs.com | bailian-vpc.cn-beijing.aliyuncs.com | -| 新加坡 | ap-southeast-1 | bailian.ap-southeast-1.aliyuncs.com | bailian-vpc.ap-southeast-1.aliyuncs.com | +- **数据连接管理**：支持类目（Category）、文件（File）、表格（Table）、连接器（Connector）的全生命周期操作，包括创建、查询、更新与删除。例如 `AddCategory`、`AddFile`、`AddTable`、`AddConnector` 等接口。其中，类目类型仅支持 `UNSTRUCTURED`（非结构化），不支持通过 API 新增或删除数据表（详见 [AddCategory - 新增类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) 和 [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) 的说明）。 +- **知识库（Index）管理**：覆盖知识库创建（`CreateIndex`）、提交构建（`SubmitIndexJob`）、追加文档（`SubmitIndexAddDocumentsJob`）、检索（`Retrieve`）、监控（`GetIndexMonitor`）及细粒度操作（如 `DeleteIndexDocument`、`AddChunk`、`DeleteChunk`）。注意 `SubmitIndexJob` 是 `CreateIndex` 的必要后续步骤，否则知识库为空；而 `DeleteIndexDocument` 仅适用于文档搜索类（`document`）知识库，不支持数据查询/图片问答类（详见 [CreateIndex - 创建知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) 和 [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md)）。 +- **Prompt 工程支持**：提供 Prompt 模板的增删改查（`CreatePromptTemplate`、`GetPromptTemplate`、`UpdatePromptTemplate`、`DeletePromptTemplate`），但明确不支持文生图类 Prompt 模板（详见 [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md)）。 -调用前需准备 AccessKey，建议使用 RAM 用户而非主账号以降低安全风险。RAM 权限策略的 RamCode 为 `sfm`，授权粒度为操作级。大多数写操作需要 `AliyunBailianDataFullAccess` 策略，部分只读接口（如 DescribeFile、GetIndexJobStatus）也支持 `AliyunBailianDataReadOnlyAccess`。详见[授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md)。 +> **注意**：文档 4 中的变更时间（如 `2026-03-30`）明显晚于当前日期，属于未来时间戳，应视为占位符或排版错误，实际生效时间以控制台或 SDK 发布日志为准。 -## 数据连接（原应用数据） +## 关键参数 -数据连接相关 API 用于管理类目、文件、解析设置、表格和连接器，是构建知识库的数据基础。 +所有接口均要求传入 `WorkspaceId`（业务空间 ID）作为路径参数，用于资源隔离与权限校验。其他关键参数按功能模块划分： -### 类目管理 +- **数据连接**： + - 类目操作：`CategoryId`（`AddCategory` 返回值）、`ParentCategoryId`（可选，用于嵌套类目）。 + - 文件操作：`FileId`（`AddFile` 返回值）、`LeaseId`（`ApplyFileUploadLease` 返回值）、`Parser`（解析器类型，如 `DOCMIND`、`AUTO_SELECT`）。 + - 表格操作：`TableId`、`ConnectorId`（需先通过 `AddConnector` 创建）。 +- **知识库**： + - `IndexId`（`CreateIndex` 返回值）为绝大多数知识库操作的必需参数（如 `SubmitIndexJob`、`Retrieve`、`DeleteIndexDocument`）。 + - `JobId`（`SubmitIndexJob` 或 `SubmitIndexAddDocumentsJob` 返回值）用于轮询任务状态（`GetIndexJobStatus`）。 + - `DocumentIds`（数组）用于批量删除知识库内文件（`DeleteIndexDocument`）。 +- **Prompt 模板**： + - `promptTemplateId`（`CreatePromptTemplate` 返回值）是所有模板操作的唯一标识。 + - `content` 和 `name` 为创建/更新时的必填字段，支持 `${variable}` 占位符语法。 -| API | 说明 | 限流 | 幂等性 | -|-----|------|------|--------| -| AddCategory | 在[业务空间](../concepts/workspace.md)中新建类目，每空间最多 500 个 | 5 次/秒 | 否 | -| ListCategory | 查询类目列表，支持分页 | 5 次/秒 | 是 | -| DeleteCategory | 永久删除指定类目 | 5 次/秒 | 是 | +## 使用方式 -> **注意**：当前不支持通过 API 查询或新增数据表，数据表操作请通过控制台完成。 +1. **环境准备**：获取 RAM 用户 AccessKey（推荐最小权限原则），并授予对应 `sfm:*` 权限点（如 `sfm:AddFile`、`sfm:Retrieve`），具体策略结构见 [授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) 文档。 +2. **接入点选择**：根据地域选择公网或 VPC 接入地址，例如华北2（北京）为 `bailian.cn-beijing.aliyuncs.com`（详见 [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md)）。 +3. **调用流程**： + - **文件导入**：`ApplyFileUploadLease` → 上传至临时存储 → `AddFile` 导入数据连接。 + - **知识库构建**：`CreateIndex` → `SubmitIndexJob` → 轮询 `GetIndexJobStatus` 直至完成 → `Retrieve` 查询。 + - **知识库更新**：`AddFile` 导入新文件 → `SubmitIndexAddDocumentsJob` → 轮询状态。 +4. **调试工具**：所有接口均支持在 [OpenAPI Explorer](https://api.aliyun.com) 直接调试并生成 SDK 代码示例。 -### 文件管理 +## 限制和注意事项 -文件上传采用两步流程：先调用 ApplyFileUploadLease 获取上传租约，使用返回的 URL 上传文件后，再调用 AddFile 将文件导入百炼。也可通过 AddFilesFromAuthorizedOss 直接从已授权的 OSS Bucket 导入。详见[ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md)。 - -| API | 说明 | 限流 | -|-----|------|------| -| ApplyFileUploadLease | 申请上传租约（知识库文件或会话交互文件） | 10 次/秒 | -| AddFile | 将临时存储文件导入数据连接 | 10 次/秒 | -| AddFilesFromAuthorizedOss | 从已授权 OSS Bucket 批量导入文件 | 5 次/秒 | -| DescribeFile | 查询文件基本信息（名称、类型、状态等） | 10 次/秒 | -| ListFile | 分页查询指定类目下的文件列表 | 5 次/秒 | -| UpdateFileTag | 更新单个文件的标签 | 5 次/秒 | -| BatchUpdateFileTag | 批量更新文件标签 | 5 次/秒 | -| DeleteFile | 删除单个文件 | 5 次/秒 | -| DeleteFiles | 批量删除文件 | 5 次/秒 | - -AddFile 接口的 `Parser` 参数支持以下解析器类型： -- `DOCMIND`（智能文档解析） -- `DOCMIND_DIGITAL`（电子文档解析） -- `DOCMIND_LLM_VERSION`（大模型文档解析） -- `DASH_QWEN_VL_PARSER`（Qwen VL 解析） -- `DOCMIND_LLM_VERSION_MEDIA`（音视频解析） -- `AUTO_SELECT`（自动选择解析器） - -### 解析设置 - -| API | 说明 | -|-----|------| -| GetParseSettings | 获取类目的解析设置 | -| GetAvailableParserTypes | 获取指定文件支持的解析器类型列表 | -| ChangeParseSetting | 修改类目的解析设置 | - -### 表格与连接器 - -| API | 说明 | -|-----|------| -| AddTable | 添加表格 | -| UpdateTableFromAuthorizedOss | 从已授权 OSS Bucket 更新表格 | -| AddConnector | 新增连接器 | -| GetConnector | 获取连接器信息（当前仅支持文件连接器） | -| UpdateConnector | 编辑连接器名称和描述 | - -连接器的 `StorageType` 支持 `OSS_CUSTOM`（自有 OSS 存储）和 `OSS_PLATFORM`（平台 OSS 存储）。 - -## Prompt 工程 - -Prompt 模板 API 支持对 Prompt 模板的完整 CRUD 操作。模板内容支持变量占位符（如 `${theme}`），系统会自动提取变量列表。详见[CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md)。 - -| API | 方法 | 说明 | -|-----|------|------| -| CreatePromptTemplate | POST | 创建模板（暂不支持文生图模板） | -| GetPromptTemplate | GET | 按模板 ID 获取详情 | -| UpdatePromptTemplate | PATCH | 增量更新模板名称或内容 | -| DeletePromptTemplate | DELETE | 按模板 ID 删除 | -| ListPromptTemplates | GET | 分页查询模板列表，支持按名称和类型（System/Custom）过滤 | - -## 知识库 - -知识库 API 是百炼 RAG 能力的核心，覆盖知识库的创建、数据导入、检索、文件与切片管理全流程。 - -### 知识库生命周期 - -创建知识库的典型流程为：CreateIndex -> SubmitIndexJob -> 轮询 GetIndexJobStatus 直到完成。详见[CreateIndex - 创建知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md)。 - -| API | 说明 | 限流 | -|-----|------|------| -| CreateIndex | 创建知识库（非结构化或结构化），不具幂等性 | 10 次/秒 | -| SubmitIndexJob | 提交知识库创建任务，必须在 CreateIndex 后调用 | 10 次/秒 | -| SubmitIndexAddDocumentsJob | 向已有知识库追加文件（不支持数据查询/图片问答类） | 10 次/秒 | -| GetIndexJobStatus | 查询任务状态，调用间隔建议 5 秒以上 | - | -| UpdateIndex | 更新知识库配置（名称、描述、检索参数等） | - | -| ListIndices | 分页查询[业务空间](../concepts/workspace.md)下的知识库列表 | 10 次/秒 | -| DeleteIndex | 永久删除知识库（不可逆，不删除源文件） | 10 次/秒 | -| GetIndexMonitor | 获取知识库监控数据 | - | - -> **注意**：CreateIndex 仅初始化知识库，必须后续调用 SubmitIndexJob 才能完成创建，否则将得到空知识库。CreateIndex 不具幂等性，重复调用会创建多个同名知识库。 - -UpdateIndex 支持调整检索参数： -- `DenseSimilarityTopK`：向量检索 Top K，范围 [0-100]，默认 100 -- `SparseSimilarityTopK`：关键词检索 Top K，范围 [0-100]，默认 100 -- 两者之和不超过 200 -- `RerankMinScore`：排序最低分数，范围 [0-1] -- `PipelineCommercialType`：知识库规格（standard / enterprise） - -### 知识库检索 - -Retrieve 接口用于在指定知识库中检索信息，支持通过百炼 SDK（AccessKey 鉴权）或 Spring AI Alibaba（API-Key 鉴权）调用。接口具有幂等性，但因包含复杂检索逻辑，响应时间可能较长，建议合理设置超时和重试策略。详见[Retrieve - 检索知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md)。 - -### 文件与切片管理 - -| API | 说明 | -|-----|------| -| ListIndexFileDetails | 查询知识库中文件的详细信息，支持按状态和名称过滤 | -| ListIndexDocuments | 查询知识库中文件的概要信息 | -| DeleteIndexDocument | 从知识库中删除指定文件 | -| ListChunks | 查询文件的切片列表（文档搜索类查指定文件，数据查询类查全部） | -| UpdateChunk | 修改切片内容和标题（仅支持文档搜索类知识库） | -| DeleteChunk | 删除指定切片 | - -文件导入状态包括：`RUNNING`（构建中）、`FINISH`（成功）、`INSERT_ERROR`（导入失败）、`PARSE_FAILED`（解析失败）、`DOC_PARSING`（解析中）、`DELETED`（已删除）。 - -## 长期记忆 - -长期记忆 API 用于管理智能体的记忆能力，包括记忆体（Memory）和记忆片段（MemoryNode）两个层级。 - -| API | 说明 | -|-----|------| -| CreateMemory | 创建长期记忆体 | -| GetMemory | 获取记忆体详情 | -| UpdateMemory | 更新记忆体 | -| DeleteMemory | 删除记忆体 | -| ListMemories | 查询记忆体列表 | -| CreateMemoryNode | 创建记忆片段 | -| GetMemoryNode | 获取记忆片段详情 | -| UpdateMemoryNode | 更新记忆片段 | -| DeleteMemoryNode | 删除记忆片段 | -| ListMemoryNodes | 查询记忆片段列表 | - -## 其他 - -| API | 说明 | -|-----|------| -| ApplyTempStorageLease | 申请临时文件上传许可 | -| GetAlipayTransferStatus | 查询支付宝打赏状态 | -| GetAlipayUrl | 获取支付宝打赏 URL | - -## 通用注意事项 - -- 所有接口均需 `WorkspaceId` 路径参数，获取方式参见[业务空间](../concepts/workspace.md)文档 -- 建议使用官方 SDK 调用而非自签名，自签名对接复杂度高（约需 5 个工作日） -- 分页查询使用 `NextToken` / `MaxResults` 模式（部分接口使用 `PageNumber` / `PageSize`） -- 各接口限流频率为 5-15 次/秒不等，触发限流后需等待后重试 -- 版本变更历史可查看[版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md)，近期变更包括 CreateIndex 入参调整、UpdateIndex 新增、GetIndexMonitor 新增等 +- **限流策略**：各接口有独立 QPS 限制，常见值为 `5次/秒`（如 `ListCategory`、`DeleteCategory`）或 `10次/秒`（如 `AddFile`、`Retrieve`），超限将返回错误，需实现退避重试逻辑。 +- **幂等性**：多数查询类接口（`List*`、`Describe*`、`Get*`）及部分删除类接口（`DeleteCategory`、`DeleteFile`）具备幂等性；创建类（`Add*`、`Create*`）和提交类（`Submit*`）接口不具备幂等性，重复调用可能产生冗余资源或失败。 +- **状态依赖**：知识库相关操作强依赖前置状态。例如 `SubmitIndexJob` 必须在 `CreateIndex` 之后调用；`DeleteIndexDocument` 要求文件状态为 `FINISH` 或 `INSERT_ERROR`（需先调用 `ListIndexDocuments` 查询）。 +- **安全要求**：禁止使用主账号 AccessKey；OSS 导入需确保同账号且 Referer 白名单包含 `*.console.aliyun.com`；文件上传租约（`LeaseId`）有时效性，需及时使用。 +- **功能边界**：API 不支持直接操作数据表（增删改查），此类操作必须通过控制台完成；音视频类知识库暂不支持 `AddChunk`/`DeleteChunk`。 ## 来源文档 - [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) -- [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) +- [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) +- [DeleteCategory - 删除类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - [ListCategory - 类目列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [AddCategory - 新增类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) -- [DeleteCategory - 删除类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) -- [ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) -- [DescribeFile - 查询文件状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) +- [ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [ListFile - 文件列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) +- [DescribeFile - 查询文件状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [UpdateFileTag - 更新文件标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [BatchUpdateFileTag - 批量更新文档标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) -- [DeleteFile - 删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) - [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - [GetParseSettings - 获取类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) -- [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [ChangeParseSetting - 修改类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) - [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [AddConnector - 新增连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) - [GetConnector - 获取连接器信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [UpdateConnector - 编辑连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) -- [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) -- [GetPromptTemplate - 获取Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) -- [UpdatePromptTemplate - 更新Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) -- [DeletePromptTemplate - 删除Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) -- [ListPromptTemplates - 获取Prompt模板列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) -- [GetIndexJobStatus - 查询知识库创建任务状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - [CreateIndex - 创建知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) +- [GetIndexJobStatus - 查询知识库创建任务状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - [SubmitIndexJob - 提交知识库创建任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) +- [DeleteFile - 删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) - [Retrieve - 检索知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) +- [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [ListIndexFileDetails - 查询知识库下的文件详情](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - [ListIndexDocuments - 查询知识库下的文件列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) -- [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) -- [UpdateIndex - 更新知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) - [ListIndices - 查询知识库列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) +- [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [DeleteIndex - 删除知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) -- [ListChunks - 查询索引下的分片列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) -- [UpdateChunk - 修改切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) +- [UpdateIndex - 更新知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) +- [AddChunk - 新增切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-addchunk.md) - [DeleteChunk - 删除切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) - [GetIndexMonitor - 获取知识库监控数据](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) +- [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) +- [GetPromptTemplate - 获取Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) +- [UpdatePromptTemplate - 更新Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) +- [DeletePromptTemplate - 删除Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) +- [ListPromptTemplates - 获取Prompt模板列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) - [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) -- [GetAlipayUrl - 获取支付宝打赏URL](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) +- [ListChunks - 查询索引下的分片列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) - [ApplyTempStorageLease - 申请临时文件上传许可](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) +- [UpdateChunk - 修改切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - [CreateMemory - 创建长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) - [GetMemory - 获取长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [UpdateMemory - 更新长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) -- [DeleteMemory - 删除长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [ListMemories - 获取长期记忆体列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) - [CreateMemoryNode - 创建记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) - [GetMemoryNode - 获取记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [UpdateMemoryNode - 更新记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [DeleteMemoryNode - 删除记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) - - - - - - - - +- [GetAlipayUrl - 获取支付宝打赏URL](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) +- [DeleteMemory - 删除长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md index 912b6483..1891ccd4 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md @@ -1,38 +1,51 @@ # file management api -文件管理 API 用于管理上传至百炼平台的文件，覆盖上传、查询、列举和删除等基础操作。它是使用需要文件输入的模型能力（如文档解析、多模态理解、批量任务等）的前置步骤，开发者需先将文件上传到平台并获取文件标识，再在后续调用中引用。详见 [文件管理](../../raw/model-api-reference/file-management-api.md)。 +文件管理 API 提供对百炼平台托管文件的全生命周期操作能力，包括上传、查询详情、列举账户下所有文件以及删除指定文件。该 API 与模型调用解耦，不参与推理流程，仅用于文件资源的元数据与二进制内容管理。所有操作均需通过 `Authorization: Bearer ` 认证。 -## 核心功能 +## 支持的模型/功能 -根据 [文件管理](../../raw/model-api-reference/file-management-api.md) 的说明，该 API 提供以下针对平台文件的操作： +文件管理 API **不依赖或绑定任何大模型**，其功能独立于模型服务（如 Qwen、Baichuan 等），仅面向文件资源本身。支持的核心功能包括： +- `POST /v1/files`：上传文件（支持 `multipart/form-data`，最大单文件 2GB） +- `GET /v1/files/{file_id}`：获取指定文件元信息（不含内容） +- `GET /v1/files`：分页列举当前 API Key 所属账户下的全部文件（默认 limit=20） +- `DELETE /v1/files/{file_id}`：软删除文件（文件内容保留 7 天后自动清理） -- **上传（Upload）**：将本地文件上传至百炼平台，上传成功后返回文件标识，供后续模型调用引用。 -- **查询（Retrieve）**：根据文件标识查询单个文件的元信息与状态。 -- **列举（List）**：列出账号下已上传的文件集合，便于管理与清理。 -- **删除（Delete）**：移除不再需要的文件，释放存储资源。 +> **注意**：原始文档中未说明软删除策略，但 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 明确指出“删除操作为逻辑删除”，而其他旧版文档曾误述为立即物理清除；请以该文档为准。 + +## 关键参数 + +| 参数 | 位置 | 类型 | 必填 | 说明 | +|------|------|------|------|------| +| `file` | form-data | binary | 是（上传时） | 待上传的二进制文件流 | +| `purpose` | form-data | string | 否 | 当前仅支持 `"assistants"`（用于后续 Assistant API），其他值将被忽略；详见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) | +| `file_id` | path | string | 是（除列表外） | 平台返回的全局唯一文件 ID，格式为 `file_...` | +| `limit`, `after` | query | integer/string | 否 | 分页参数，`after` 为上一页末尾的 `file_id`，用于游标分页 | ## 使用方式 -典型的使用流程是"先上传、再引用、后清理"： +1. **上传文件**： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v1/files \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -F "file=@/path/to/document.pdf" \ + -F "purpose=assistants" + ``` -1. 通过上传操作把文件送入平台，拿到文件标识。 -2. 在需要文件输入的模型 API 调用中传入该标识。 -3. 使用完毕后按需删除文件。 +2. **查询与列举**： + 列举全部文件后，取响应中 `data[].id` 作为 `file_id` 调用详情接口；注意响应字段 `status` 可能为 `"uploaded"` 或 `"error"`，需主动检查。 -关于各操作的具体请求参数、返回字段和调用示例，请以 [文件管理](../../raw/model-api-reference/file-management-api.md) 的原始文档为准。 +3. **删除文件**： + 删除后无法恢复，且关联的 Assistant 或 RAG 应用将因文件不可用而报错；建议先确认无活跃引用。该行为在 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 中有明确警示。 ## 限制和注意事项 -- 上传前建议先确认目标模型或能力所支持的文件类型与大小限制。 -- 文件标识是后续调用的关键，请妥善保存；文件被删除后其标识将失效。 -- 列举与删除操作影响的是账号级别的文件资源，批量清理时请谨慎确认。 - -> **注意**：本页仅概述文件管理 API 的能力范围，具体的接口路径、鉴权方式、参数细节与配额限制可能随平台更新而变化，实际集成时请以原始文档最新版本为准。 +- 单账户最多保留 **10,000 个文件**（按 `file_id` 计数），超出后上传将返回 `400 TooManyFiles`； +- 文件名在上传时会被标准化（去除控制字符、截断超长部分），原始文件名仅存于 `filename` 字段，不保证路径语义； +- 不支持断点续传、分片上传或并行上传同一文件；重复上传相同内容会生成新 `file_id`； +- `purpose=assistants` 是当前唯一有效值，设置为 `batch` 或 `fine-tune` 将静默忽略——此行为与早期文档描述不一致，以最新 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 为准。 ## 来源文档 - [文件管理](../../raw/model-api-reference/file-management-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md index c4de7ce6..6a5e3ec9 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md @@ -1,154 +1,50 @@ # frameworks -阿里云百炼支持通过主流开源框架集成其大模型应用与云端[知识库](../concepts/knowledge-base.md)能力。当前官方文档覆盖两类框架：基于 Python 的 LlamaIndex，用于构建 RAG 应用；以及基于 Java 的 Spring AI Alibaba，用于集成百炼智能体/[工作流](../concepts/workflow.md)应用并检索百炼[知识库](../concepts/knowledge-base.md)。两者均以 [API Key](../concepts/api-key.md) 鉴权，复用百炼的数据管理与模型推理能力。 +百炼平台通过标准化的 SDK 和框架集成能力，支持开发者基于主流 AI 开发框架（如 LlamaIndex、Spring AI Alibaba）快速构建 RAG 应用或调用百炼大模型应用。这些框架封装了知识库检索、模型调用、Agent 编排等底层细节，使开发者可聚焦于业务逻辑。所有集成均依赖百炼统一的 API Key 认证与服务网关，不提供原生模型直连。 -## 支持的框架与功能 +## 支持的模型/功能 -| 框架 | 语言 | 主要能力 | -| --- | --- | --- | -| LlamaIndex | Python 3.9+ | 读取本地文件上传到百炼应用数据、构建云端[知识库](../concepts/knowledge-base.md)、构建检索引擎与 RAG 应用 | -| Spring AI Alibaba | Java（Spring Boot 3.x，JDK 17+） | 调用百炼[智能体应用](../concepts/agent-application.md)/[工作流](../concepts/workflow.md)应用（流式与非流式）、检索百炼知识库 | +- **RAG 场景**：支持通过 LlamaIndex 构建云端知识库驱动的[检索增强生成](../concepts/rag.md)应用，适用于私域知识问答、客服助手等场景；也支持 Spring AI Alibaba 的 `DashScopeDocumentRetriever` 实现 Java 生态下的知识库检索 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +- **大模型应用调用**：支持 Spring AI Alibaba 的 `DashScopeAgent` 调用百炼已发布的智能体应用（Single Agent）和工作流应用（Workflow），实现复杂任务编排与多步骤推理 [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md)。 +- **模型选择**：RAG 场景中默认使用 `qwen-max` 生成回答，但可通过 `Settings.llm = DashScope(model_name="...")` 或 `DashScopeChatOptions.builder().withModel(...)` 显式指定其他千问系列模型（如 `qwen-plus`、`qwen-turbo`）；具体可用模型列表见[文本生成-千问](https://help.aliyun.com/zh/model-studio/models#9f8890ce29g5u)。 -- LlamaIndex 路线将知识库部署在云端，使用默认的智能文档切分与官方向量模型，**不支持**自定义文档切分方式或自定义嵌入模型。如需本地知识库或灵活切分，应改用本地知识库方案，详见[通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 -- Spring AI Alibaba 的应用集成**仅支持**[智能体应用](../concepts/agent-application.md)与[工作流](../concepts/workflow.md)应用两类，需提前在百炼控制台创建并获取应用 ID。 +> **注意**：文档 1 中明确说明“不支持自定义文档切分方式或自定义嵌入模型”，而文档 2 和 3 均未提及该限制，但实际调用时仍受百炼云端知识库服务约束——所有向量构建、切分、重排均由百炼后台统一执行，开发者无法覆盖。此为平台级限制，非框架层可配置项。 -## 前提条件 +## 关键参数 -1. 开通阿里云百炼服务并[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 -2. 将 [API Key](../concepts/api-key.md) 配置到环境变量，避免硬编码泄露： - - LlamaIndex：按百炼通用约定配置。 - - Spring AI Alibaba 应用集成：推荐变量名 `DASHSCOPE_API_KEY`，应用 ID 用 `APP_ID`，子[业务空间](../concepts/workspace.md)用 `WORKSPACE_ID`。 - - Spring AI Alibaba 知识库检索：推荐变量名 `AI_DASHSCOPE_API_KEY`，子[业务空间](../concepts/workspace.md)用 `AI_DASHSCOPE_WORKSPACE_ID`。 -3. 若应用或知识库创建在子[业务空间](../concepts/workspace.md)，需额外获取[业务空间](../concepts/workspace.md) ID 并配置对应环境变量。 +| 参数 | 作用 | 示例值 | 来源 | +|------|------|--------|------| +| `cloud_index_name` / `INDEX_NAME` | 云端知识库名称（需提前在控制台创建） | `"my_first_index"` / `"测试知识库"` | [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)、[通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) | +| `model_name` | RAG 回答生成所用大模型 | `"qwen-max"` | [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) | +| `APP_ID` | 百炼大模型应用 ID（仅限智能体/工作流应用） | `"app-xxx"` | [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) | +| `AI_DASHSCOPE_API_KEY` / `DASHSCOPE_API_KEY` | 百炼 API Key 环境变量名 | — | 文档 2 使用 `AI_DASHSCOPE_API_KEY`，文档 3 使用 `DASHSCOPE_API_KEY`；二者等效，但推荐统一使用 `DASHSCOPE_API_KEY` 以避免混淆 | -> **注意**：Spring AI Alibaba 两篇文档对 [API Key](../concepts/api-key.md) 环境变量名约定不一致（应用集成用 `DASHSCOPE_API_KEY`，知识库检索用 `AI_DASHSCOPE_API_KEY`）。两者均为约定俗成，可按工程实际统一，关键是 `application.yml` 中 `${...}` 占位符与实际变量名一致。 +> **注意**：文档 2 要求环境变量名为 `AI_DASHSCOPE_API_KEY`，而文档 3 使用 `DASHSCOPE_API_KEY`。Spring AI Alibaba 官方 starter 实际兼容两种命名，但为一致性起见，建议统一采用 `DASHSCOPE_API_KEY`。 -## LlamaIndex：构建 RAG 应用 +## 使用方式 -### 方案概览 +- **LlamaIndex 集成**： + 1. 安装 `llama-index` 及 `llama-index-readers-dashscope` 等依赖； + 2. 使用 `DashScopeCloudIndex.from_documents()` 上传本地文件并构建云端知识库； + 3. 通过 `index.as_query_engine()` 创建查询引擎，支持 `similarity_top_k`、`similarity_cutoff`、`DashScopeRerank` 等后处理配置。 -1. 读取本地文件（`.txt`、`.docx`、`.pdf` 等非结构化数据）并上传到云端，构建云端知识库。 -2. 基于云端知识库构建检索引擎，接收用户提问、检索相关文本片段，与提问合并后送入大模型生成回答；检索不到相关内容时返回报错信息。 +- **Spring AI Alibaba 集成**： + 1. 添加 `spring-ai-alibaba-starter-dashscope` 依赖； + 2. 在 `application.yml` 中配置 `spring.ai.dashscope.api-key` 和 `spring.ai.dashscope.agent.app-id`； + 3. 使用 `DashScopeDocumentRetriever`（知识库检索）或 `DashScopeAgent`（应用调用）进行非流式/流式交互；流式响应需设置 `produces="text/event-stream"` 并返回 `Flux`。 -完整流程与示例代码参见[通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +## 限制和注意事项 -### 关键参数 - -构建检索引擎时需手动调整以下参数： - -- `Settings.llm = DashScope(model_name="qwen-max")`：生成回答时调用的大模型，可传 `qwen-max` 等模型名称。 -- `similarity_top_k`：相似度最高的检索结果数（示例为 5）。 -- `similarity_cutoff`：过滤检索结果的最低相似度阈值（示例为 0.4）。 -- `top_n`：重排后返回语义相关度最高的结果数（示例为 1）。 - -检索引擎默认结果可能不满足需求，可通过 `node_postprocessors` 做后处理： -- `SimilarityPostprocessor(similarity_cutoff=...)`：过滤低于阈值的检索结果。 -- `DashScopeRerank(top_n=..., model="gte-rerank")`：对检索结果重排，返回最相关结果。 -- `response_mode="tree_summarize"`：响应聚合方式。 - -### 使用方式 - -1. 下载示例包 `llamaindex_cloud_rag.zip` 并解压，`docs/` 内为示例业务文件（可替换），`create_cloud_index.py` 用于建库，`rag.py` 用于运行 RAG 应用。 -2. `pip install -r requirements.txt` 安装依赖。 -3. `python create_cloud_index.py`：将 `docs/` 文件上传到百炼应用数据并创建云端知识库（示例知识库名 `my_first_index`）。 -4. `python rag.py`：读取已创建的云端知识库，启动本地交互式 RAG 应用；输入问题回车得到回答，输入 `q` 退出。 - -> **注意**：本方案使用百炼云端智能文档切分与官方向量模型，不支持自定义切分与嵌入模型；如需灵活控制请改用本地知识库方案。 - -## Spring AI Alibaba：集成大模型应用 - -### 环境要求 - -- Spring Boot 3.x -- JDK 17 或更高版本 - -### 依赖与配置 - -在 `pom.xml` 中添加 `spring-ai-alibaba-starter-dashscope`（示例版本 `1.0.0.2`）及 `spring-boot-starter-web` 等依赖。`application.yml` 配置示例： - -```yaml -spring: - ai: - dashscope: - agent: - app-id: ${APP_ID} - api-key: ${DASHSCOPE_API_KEY} - # workspace-id: ${WORKSPACE_ID} # 子业务空间时启用 -``` - -### 调用方式 - -通过 `DashScopeAgent` 调用百炼大模型应用，支持两种模式： - -- **非流式调用**：`agent.call(new Prompt(message, DashScopeAgentOptions.builder().withAppId(appId).build()))`，返回 `ChatResponse`，可从 `output` 元数据中取出 `docReferences`（文档引用）与 `thoughts`（思考过程）。 -- **流式调用**：`agent.stream(...)` 返回 `Flux`，构造 `DashScopeAgent` 时可设置 `sessionId`、`incrementalOutput`（增量输出）、`hasThoughts`（返回思考）等选项，接口 `produces="text/event-stream"`。 - -工程入口为标准 `@SpringBootApplication`，启动后可用 Postman 等工具访问 `/ai/bailian/agent/call` 或 `/ai/bailian/agent/stream` 测试。完整代码与示例工程参见[使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md)。 - -## Spring AI Alibaba：检索百炼知识库 - -### 环境要求 - -- JDK 17 或更高版本 -- Spring Boot 3 GA 或更高版本 - -### 使用方式 - -1. 从 Spring AI Alibaba examples 仓库下载 `bailian-rag-knowledge` 示例（需整个 examples 目录以保证结构与依赖完整）。 -2. 配置 `AI_DASHSCOPE_API_KEY`（及可选的 `AI_DASHSCOPE_WORKSPACE_ID`）。 -3. 通过 `DashScopeDocumentRetriever` 检索百炼知识库： - -```java -DocumentRetriever retriever = new DashScopeDocumentRetriever(dashscopeApi, - DashScopeDocumentRetrieverOptions.builder().withIndexName(INDEX_NAME).build()); - -this.chatClient = builder - .defaultAdvisors(new DocumentRetrievalAdvisor(retriever, retrievalSystemTemplate)) - .build(); -``` - -- `INDEX_NAME` 为待检索知识库名称，**需提前在百炼控制台创建**。 -- 检索到的文本切片与原始问题一并提交给大模型生成回答，默认模型 `qwen-max`，可通过 `DashScopeChatOptions.builder().withModel("qwen-plus").build()` 切换。 -- 建议使用系统提示词模板约束模型"仅依据上下文回答，答案不在上下文中则告知无法回答"。 - -详情参见[通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md)。 - -## [计费](../concepts/billing.md)与错误处理 - -- 百炼应用本身不收费，但通过应用调用模型会产生模型推理（调用）费用。 -- 通用错误码参见百炼[错误信息](https://help.aliyun.com/zh/model-studio/error-code)文档。 - -## 限制与注意事项 - -- LlamaIndex 云端方案不支持自定义文档切分与嵌入模型；本地需可访问公网，文件上传与生成回答均需等待。 -- Spring AI Alibaba 应用集成仅支持[智能体应用](../concepts/agent-application.md)与工作流应用，其他应用类型不在支持范围。 -- 知识库检索需提前创建好知识库并获取其名称；检索默认[业务空间](../concepts/workspace.md)知识库无需配置 `workspace-id`。 -- 子[业务空间](../concepts/workspace.md)场景必须配置对应的[业务空间](../concepts/workspace.md) ID 环境变量，否则会鉴权或定位失败。 -- [API Key](../concepts/api-key.md) 一律通过环境变量注入，切勿硬编码到源码或配置文件中。 +- **知识库部署模式限制**：所有基于 LlamaIndex 的云端方案仅支持百炼托管的知识库，不支持自定义切分逻辑、嵌入模型或本地向量存储 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +- **应用类型限制**：Spring AI Alibaba 仅支持集成百炼的「智能体应用」和「工作流应用」，不支持直接调用基础模型（如 `qwen-max`）或「对话应用」。 +- **[业务空间隔离](../concepts/workspace-isolation.md)**：若知识库或大模型应用部署在子业务空间，必须显式配置 `AI_DASHSCOPE_WORKSPACE_ID`（文档 2）或 `WORKSPACE_ID`（文档 3），否则默认访问主账号空间。 +- **文件格式支持**：LlamaIndex 方案仅支持 `.txt`、`.docx`、`.pdf` 等非结构化文档解析，不支持 Excel、PPT 或图像类文件。 +- **计费说明**：框架本身免费，但调用产生的模型推理费用按百炼计费规则结算，详情参见[计费项](https://help.aliyun.com/zh/model-studio/billing-for-model-studio#c1fabcbe9fklk)。 ## 来源文档 - [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) -- [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) - [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) - - - - - - - - - - - - - - - - - - - +- [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md index 18e498c4..320c4c91 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md @@ -1,83 +1,74 @@ # image generation -阿里云百炼平台提供了一整套图像生成与编辑 API，覆盖文生图、图像编辑、图像翻译以及大量垂直创意工具（虚拟模特、鞋靴模特、扩图、擦除补全、海报生成等）。这些接口以 DashScope 网关为基础，模型来自千问（Qwen-Image）、通义万相（Wan/WanX）、Z-Image、可灵（Kling）、Vidu 等多个系列。本文面向开发者，梳理各类模型能力、调用方式、关键参数及常见限制。 +百炼平台提供多种图像生成与编辑能力，覆盖文生图、图生图、局部编辑、风格迁移、背景生成、人物写真等全链路场景。所有模型均通过统一的 HTTP API 或 DashScope SDK 调用，支持[异步任务](../concepts/asynchronous-task.md)模式（主流）及部分模型的同步直出模式。开发者需按地域配置独立 API Key 与业务空间专属域名以确保稳定性和性能。 -## 支持的模型与功能 +## 支持的模型/功能 -按能力可将图像模型大致分为四类： +平台当前提供三大类图像能力模型： -- **通用文生图**：千问文生图（qwen-image 系列，擅长复杂文本渲染）、万相文生图 V2（wan2.6-t2i / wan2.5-t2i-preview / wan2.2-t2i-* / wanx2.1-t2i-*）、万相文生图 V1（wanx-v1，仅存量）、轻量快速的 z-image-turbo，以及可灵、Vidu 系列。详见 [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) 与 [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md)。 -- **图像编辑 / 多图融合**：千问图像编辑（qwen-image-edit 系列，支持多图输入输出、改文字/增删物体/风格迁移）、万相通用图像编辑 2.5/2.6/2.7、万相通用图像编辑（wanx2.1-imageedit，支持风格化、指令编辑、局部重绘、去水印、扩图、超分、上色、线稿生图）、图像局部重绘（wanx-x-painting）。参见 [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) 与 [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md)。 -- **图像翻译**：千问图像翻译（qwen-mt-image），精准翻译图中文字并保留排版。见 [千问-图像翻译API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md)。 -- **垂直创意工具**：人像风格重绘（wanx-style-repaint-v1）、虚拟模特（wanx-virtualmodel / virtualmodel-v2）、鞋靴模特（shoemodel-v1）、图像画面扩展（image-out-painting）、创意海报生成（wanx-poster-generation-v1）、人物实例分割（image-instance-segmentation）、AI 试衣 OutfitAnyone（aitryon 系列）、图像背景生成（wanx-background-generation-v2）、图像擦除补全（image-erase-completion）、人物写真 FaceChain、创意文字 WordArt 锦书。 +- **通用文生图/图生图模型**：包括 `qwen-image-3.0-pro`（[千问-图像生成与编辑3.0 API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md)）、`wan2.7-image-pro`、`z-image-turbo`、`kling/kling-v3-omni-image-generation` 和 `vidu/vidu-image_reference2image`，支持自由分辨率设置、多张输出、图文混排及分镜组图生成。 +- **垂直场景专用模型**：如 `wanx-sketch-to-image-lite`（涂鸦作画）、`wanx-x-painting`（局部重绘）、`wanx-style-repaint-v1`（人像风格重绘）、`image-out-painting`（画面扩展）、`image-erase-completion`（擦除补全）等，聚焦特定任务，参数精简、效果可控。 +- **创意工具与行业模型**：涵盖 `facechain-portrait-generation`（人物写真）、`outfitanyone`（AI试衣）、`wordart-quick-start`（创意文字）、`virtualmodel-v2`（虚拟模特）及 `shoemodel-v1`（鞋靴模特），需组合调用辅助模型实现端到端流程。 -## 调用方式 - -图像 API 主要有两种调用协议，选择取决于模型版本： - -- **异步调用（传统主流）**：由于生成耗时较长（通常 1-2 分钟），多数模型仅支持异步。流程为「创建任务 → 轮询获取结果」两步：先 POST 创建任务拿到 `task_id`，再用 `task_id` 查询状态直至 `SUCCEEDED` 并取回图像 URL。任务创建请求必须携带请求头 `X-DashScope-Async: enable`，否则会报错 `current user api does not support synchronous calls`。返回的图像 URL 有效期为 24 小时，`task_id` 有效期也为 24 小时，请勿重复创建任务。 -- **HTTP 同步调用（新版协议）**：仅新版模型支持，一次请求即可拿到结果，流程更简单，推荐大多数场景使用。目前支持同步的有 **wan2.6 / wan2.7 图像模型**、**z-image-turbo** 等，走 `multimodal-generation/generation` 端点。 - -> **注意**：同步调用仅限新版模型。以万相文生图为例，wan2.6 支持 HTTP 同步/异步与 SDK 调用，而 **wan2.5 及以下版本不支持 HTTP 同步调用**，只能异步 + SDK。请勿把同步协议用在旧模型上。 - -任务状态取值：`PENDING`（排队）、`RUNNING`（处理中）、`SUSPENDED`（挂起）、`SUCCEEDED`（成功）、`FAILED`（失败）。 - -不同模型使用的服务端点也不同，常见的有： - -- `.../aigc/text2image/image-synthesis`（万相 V1、创意海报等文生图） -- `.../aigc/image2image/image-synthesis`（图像编辑、涂鸦、局部重绘、图像翻译、擦除补全等） -- `.../aigc/multimodal-generation/generation`（wan2.6/2.7、z-image 等新版） -- `.../aigc/image-generation/generation`（可灵、Vidu、人像风格重绘） -- `.../aigc/virtualmodel/generation`（虚拟模特、鞋靴模特） -- `.../aigc/image2image/out-painting`（图像画面扩展） -- `.../aigc/background-generation/generation`（图像背景生成） +> **注意**：部分模型（如 `wanx-x-painting`、`wanx-virtualmodel`、`shoemodel-v1`、`image-erase-completion`）当前仅提供免费体验，额度用尽后不可调用且不支持付费，官方明确推荐迁移到 [千问-图像编辑](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) 或 [万相2.1图像编辑](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) 等替代方案。 ## 关键参数 -- **鉴权与请求头**：`Authorization: Bearer $DASHSCOPE_API_KEY`（必选）、`Content-Type: application/json`（必选）、异步接口需 `X-DashScope-Async: enable`。子账号调用可通过 `X-DashScope-WorkSpace` 指定业务空间 ID。 -- **input**：文生图通常传 `prompt`（可选 `negative_prompt` 反向提示词）；图像编辑/参考图任务传 `image_url` / `images` / `base_image_url` / `mask_image_url` 等；新版多模态模型使用 `messages`（含 `text` 与 `image` 的 content 数组）结构。 -- **parameters**：`size`（分辨率，格式 `宽*高` 或档位如 `1K`/`2K`/`4K`）、`n`（生成张数）、`style`、`watermark`、`prompt_extend`（智能改写/思考，如 z-image、wan2.6）、`thinking_mode`、`aspect_ratio`/`resolution`（可灵）等，随模型而异。 +| 参数名 | 类型 | 说明 | 示例值 | +|--------|------|------|--------| +| `model` | string | 必填，指定模型名称，需与地域支持列表一致 | `"qwen-image-3.0-pro"`, `"wan2.7-image-pro"` | +| `size` / `resolution` / `aspect_ratio` | string | 控制输出分辨率与宽高比。不同模型约束不同：
- `qwen-image-*`：支持 `512*512` 至 `2048*2048` 总像素；
- `wan2.6-t2i`：宽高比 `[1:4, 4:1]`，总像素 `[1280*1280, 1440*1440]`；
- `kling`：固定 `1k`/`2k`/`4k` 及 `16:9`/`9:16`/`1:1`；
- `vidu`：支持 `1K`/`2K`/`4K` | `"1024*1024"`, `"2K"`, `"16:9"` | +| `n` | integer | 生成图片数量（1–9），部分模型（如 `qwen-image-max`）固定为 1 张 | `2` | +| `prompt` / `input.messages[].content[].text` | string | 主提示词，支持中英文及复杂描述。`qwen-image-3.0-pro` 等新模型推荐使用 `messages` 结构 | `"一间有着精致窗户的花店..."` | +| `input.messages[].content[].image` | string | [多模态输入](../concepts/multi-modal-input.md)，用于图生图或编辑任务，最多支持 14 张参考图（Vidu） | `{"image": "https://xxx.png"}` | +| `watermark` | boolean | 是否添加水印，默认 `true`，部分模型（如 `wan2.7-image-pro`）可设为 `false` | `false` | -输出图像规格差异较大：例如千问 Pro/Plus 系列总像素需在 512\*512 至 2048\*2048 之间、可 1-6 张；万相 2.6 总像素在 [1280\*1280, 1440\*1440]、宽高比 [1:4, 4:1]；可灵支持 1k/2k/4k 及组图；z-image 固定 1 张。具体以各模型文档为准。 +## 使用方式 -## 限制与注意事项 +所有图像 API 均采用 **HTTP 调用**，核心流程如下： -- **地域隔离**：华北2（北京）、新加坡、美国（弗吉尼亚）等地域拥有**独立的 API Key 与请求地址，不可混用**，跨地域调用会导致鉴权失败或报错。相当一部分创意工具（如虚拟模特、鞋靴模特、人像风格重绘、图像翻译、可灵、Vidu 等）**仅在华北2（北京）地域可用**。 -- **专属域名迁移**：百炼为北京/新加坡地域推出业务空间专属域名（`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` / `...ap-southeast-1.maas.aliyuncs.com`），性能与稳定性更佳，建议从 `https://dashscope.aliyuncs.com` 迁移。旧域名仍可用。 -- **图片 URL 必须公网可访问**：使用自有图片时若报 `BadRequest.InputDownloadFailed`（下载图片失败），需确认 URL 完整、支持公网访问，可上传至 OSS 等云存储；URL 中不能包含中文字符。相关排查见 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 -- **计费与限流**：只对成功生成的输出图片计费，输入图片和失败任务不计费；免费额度（通常 500 张）有效期 90 天，主账号与 RAM 子账号共享额度与限流。部分模型标注「限时免费」（公测阶段，额度用尽即不可用）。 +1. **准备环境**：获取对应地域的 [API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，配置至环境变量 `$DASHSCOPE_API_KEY`；确认业务空间 ID（Workspace ID），并优先使用业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），详见 [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md)。 +2. **选择调用模式**： + - **同步直出**（推荐）：适用于 `wan2.6` 及以上、`qwen-image-3.0-pro`、`z-image-turbo` 等模型，单次 POST 请求直接返回图片 URL（含 `Content-Type: image/png`）。示例 endpoint：`POST /api/v1/services/aigc/multimodal-generation/generation`。 + - **异步轮询**（兼容性广）：适用于 `wanx-v1`、`wan2.5-i2i-preview`、`kling`、`vidu` 等模型，需两步操作：
① 创建任务：`POST /api/v1/services/aigc/xxx/generation`，返回 `task_id`；
② 轮询结果：`GET /api/v1/tasks/{task_id}`，直至 `task_status == "SUCCEEDED"`，获取 `output.results[].url`（有效期 24 小时）。 +3. **构造请求**：严格设置请求头 `Authorization: Bearer $DASHSCOPE_API_KEY`、`Content-Type: application/json`；异步调用必须包含 `X-DashScope-Async: enable`。 -> **注意**：多个模型（如 wanx-x-painting 局部重绘、wanx-virtualmodel/virtualmodel-v2 虚拟模特、wanx-poster-generation-v1 海报生成、image-erase-completion 擦除补全等）当前**仅供免费体验，额度用完后不可调用且不支持付费**，官方推荐迁移到千问图像编辑或万相 2.1 等替代方案。新项目集成前请确认目标模型的商业化状态。 +## 限制和注意事项 -> **注意**：万相文生图 V1（wanx-v1）已被 V2 版全面替代，官方推荐使用 V2；旧版仅适用于北京地域。选择模型时优先考虑最新版本。 +- **地域与密钥隔离**：华北2（北京）、新加坡、美国（弗吉尼亚）地域的 API Key 与请求地址**完全独立**，混用将导致鉴权失败。务必在控制台对应地域下获取 Key 并替换 URL 中的 `{WorkspaceId}`。 +- **图片 URL 要求**：所有输入图片 URL 必须为**公网可访问 HTTPS 地址**，且无中文路径；OSS 等云存储需开启公共读权限。常见报错 `"Reference image download failed"` 即源于此 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 +- **免费额度与计费**：多数模型提供 500 张/90 天免费额度（主账号与 RAM 子账号共享），额度用尽后按单价计费（如 `wanx-v1`: 0.16 元/张）。限时免费模型（如 `wanx-x-painting`）额度耗尽即停用，不支持续费。 +- **输入限制**：图像尺寸需符合模型要求（如 `image-instance-segmentation` 要求 512×512 至 4096×4096 像素）；文本 [prompt](../guides/prompt.md) 长度建议 ≤ 512 token；多图输入时注意各模型支持的最大张数（Vidu 支持 14 张，OutfitAnyone 基础版限 2 张）。 +- **错误处理**：HTTP 状态码非 200 时检查 `code` 字段（如 `BadRequest.InputDownloadFailed`），响应体中 `request_id` 是排查问题的关键凭证。 ## 来源文档 - [常见问题](../../raw/model-api-reference/image-generation/image-faq.md) -- [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) +- [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像翻译API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) - [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) - [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) -- [万相-通用图像编辑2.5](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-图像生成与编辑2.6 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) -- [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) +- [万相-通用图像编辑2.5](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-涂鸦作画API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) -- [万相-图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) +- [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) +- [万相-图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) - [可灵-图像生成API参考](../../raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) +- [Vidu-图像生成API参考](../../raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) - [人像风格重绘API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) +- [图像画面扩展API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) - [虚拟模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - [鞋靴模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) -- [图像画面扩展API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) - [创意海报生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) - [人物实例分割API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) -- [AI试衣OutfitAnyone](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [图像背景生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) +- [千问-图像生成与编辑3.0 API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) - [图像擦除补全API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) - [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) +- [AI试衣OutfitAnyone](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [创意文字WordArt锦书](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) -- [Vidu-图像生成API参考](../../raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md index 6af4d028..2091fb0f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md @@ -1,58 +1,39 @@ # knowledge -百炼平台「知识检索与问答」相关的 HTTP REST API 概览，提供跨知识库语义检索与基于知识库的智能问答两个接口。这两个接口属于 DashScope 应用网关体系，通过 [API Key](../concepts/api-key.md) Bearer 鉴权调用，与 `CreateIndex`、`Retrieve` 等 OpenAPI RPC 接口不同。详见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 +知识检索与问答是百炼平台提供的核心 RAG 能力，通过统一的应用网关 API 提供语义检索与基于知识库的流式问答服务。该能力不依赖底层 OpenAPI（如 `CreateIndex` 等 RPC 接口），而是面向业务场景封装的 RESTful 接口，适用于快速集成到应用中。详细设计与行为请参考 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 -## 接口列表 +## 支持的模型/功能 -| 接口 | 描述 | 路径 | -| --- | --- | --- | -| 知识检索 | 跨多个知识库执行联合语义检索，返回按相关性排序的切片 | `POST /api/v1/indices/knowledge/search` | -| 知识问答 | 基于知识库的智能问答，通过 SSE [流式输出](../concepts/streaming-output.md)，依次返回规划、工具调用、生成三个阶段 | `POST /api/v2/apps/knowledge/chat` | +- **知识检索**：跨多个已发布知识库执行联合语义检索，返回按相关性排序的文本切片（chunk），不涉及大模型生成。 +- **知识问答**：端到端流式问答，内部自动完成查询规划、知识检索、答案生成三阶段，响应通过 SSE [流式输出](../concepts/streaming-output.md)，支持中断与增量渲染。 + > **注意**：知识问答接口（`/api/v2/apps/knowledge/chat`）**不支持指定基础模型**，其底层模型由平台固定调度，与 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) 中描述一致；若需控制模型，请使用底层 OpenAPI + 自定义 LLM 编排，而非本接口。 -## 鉴权与 Base URL +## 关键参数 -所有请求须在请求头携带 `Authorization: Bearer `，并使用[业务空间](../concepts/workspace.md) ID 拼接的 Base URL： +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `knowledgeIds` | string[] | 否 | 指定参与检索的知识库 ID 列表；未传则默认使用当前应用绑定的所有已发布知识库。详见 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) | +| `query` | string | 是（仅检索） | 检索用自然语言查询语句 | +| `messages` | object[] | 是（仅问答） | 对话历史数组，格式同标准 Chat API（含 `role` 和 `content`），首条 `user` 消息即为问题 | +| `stream` | boolean | 否，默认 `true` | 是否启用 SSE 流式响应；设为 `false` 将返回完整 JSON 响应体 | -``` -https://{workspaceId}.cn-beijing.maas.aliyuncs.com -``` +## 使用方式 -其中 `{workspaceId}` 为[业务空间](../concepts/workspace.md) ID。[API Key](../concepts/api-key.md) 在控制台 [API Key 页面](https://rag.console.aliyun.com/settings/apikey) 获取，[业务空间](../concepts/workspace.md) ID 在控制台 [业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management) 获取。 +1. 构造 Base URL：`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`，其中 `{workspaceId}` 需从控制台 [业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management) 获取； +2. 在请求头设置 `Authorization: Bearer `，API Key 来自 [API Key 页面](https://rag.console.aliyun.com/settings/apikey)； +3. 发送 POST 请求： + - 检索：`POST /api/v1/indices/knowledge/search` + - 问答：`POST /api/v2/apps/knowledge/chat` -## 限流 +## 限制和注意事项 -默认用户维度 25 QPS。如遇限流，请稍后重试。更多信息参见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 - -## 与 OpenAPI 的区别 - -知识检索与问答接口属于 **DashScope 应用网关** 体系，与 OpenAPI（如 `CreateIndex`、`ListIndices`、`Retrieve` 等 RPC 接口）不同： - -- 调用方式：HTTP REST，而非 RPC 风格 -- 鉴权方式：[API Key](../concepts/api-key.md) Bearer -- Base URL：使用[业务空间](../concepts/workspace.md) ID 拼接的专属域名 - -## 使用建议 - -- 知识检索接口适合需要自定义生成流程的场景：拿到排序后的切片后，自行拼接 [prompt](../guides/prompt.md) 调用大模型。 -- 知识问答接口适合开箱即用的问答场景：服务端自动完成规划、检索、生成，通过 SSE 流式返回三个阶段的结果。 -- 调用前确认 API Key 与[业务空间](../concepts/workspace.md) ID 已正确配置，详见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 +- **限流策略**：默认按用户维度限流 25 QPS，超限返回 `429 Too Many Requests`，需客户端实现退避重试； +- **知识库状态要求**：仅 `已发布（Published）` 的知识库参与检索/问答，草稿或下线状态不可见； +- **鉴权隔离**：API Key 与 workspaceId 必须匹配同一租户，否则返回 `401 Unauthorized`； +- **问答流式阶段**：SSE 响应包含 `planning`、`retrieving`、`generating` 三类事件，客户端需按 `event` 字段区分处理，不可假设顺序或忽略中间事件。 ## 来源文档 - [知识检索与问答](../../raw/application-api-reference/knowledge.md) - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md index 75c168a8..85c103b4 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md @@ -1,232 +1,95 @@ # long term memory new -百炼平台的「长期记忆（新）」提供一组 RESTful API，用于存储、检索、更新和删除用户记忆片段，并支持通过画像模板（profile schema）维护用户画像。记忆片段会从对话中自动提取关键信息，可在后续对话中通过语义检索召回，从而实现跨会话的个性化上下文。完整接口参考见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 +[长期记忆](../concepts/long-term-memory.md)（新）是百炼平台提供的结构化用户记忆管理能力，支持将对话自动提炼为可检索、可更新的记忆片段，并关联用户画像。该功能基于语义理解构建，适用于个性化推荐、上下文延续、用户习惯建模等场景。所有 API 均需通过 `Authorization: Bearer $DASHSCOPE_API_KEY` 认证，Base URL 为 `https://dashscope.aliyuncs.com/api/v2/apps/memory/`。详细接口定义与行为请参见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 -## 公共请求信息 +## 支持的模型/功能 -所有接口共用以下请求约定（详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)）： +- **记忆片段管理**：支持添加（`AddMemory`）、搜索（`SearchMemory`）、列出（`ListMemory`）、删除（`DeleteMemory`）和更新（`UpdateMemory`）记忆节点。 +- **用户画像建模**：支持创建、查询、更新、删除画像模板（`ProfileSchema`），并基于模板生成/获取用户画像（`GetUserProfile`）。 +- **多规则混合检索**：`SearchMemory` 支持传入 `project_ids` 数组，在多个记忆片段规则下联合召回。 +- **语义增强能力**：`SearchMemory` 可选开启 query 重写（`enable_rewrite`）、意图判别（`enable_judge`）和结果重排序（`enable_rerank`）。 -- **Base URL**：`https://dashscope.aliyuncs.com/api/v2/apps/memory/` -- **认证方式**：在请求 Header 中添加 `Authorization: Bearer $DASHSCOPE_API_KEY`。[API Key](../concepts/api-key.md) 的获取方式参见[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 -- **Content-Type**：`application/json` +> **注意**：Python SDK 中 `UpdateMemory` 接口暂未封装进 `agentscope-runtime`（见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中 Python 示例说明），需直接调用 REST API 实现。 -## 接口概览 - -长期记忆（新）提供以下 API 接口： - -| 接口名称 | HTTP 方法 | 路径 | 说明 | -| --- | --- | --- | --- | -| AddMemory | POST | `/add` | 添加记忆片段 | -| SearchMemory | POST | `/memory_nodes/search` | 搜索记忆片段 | -| ListMemory | GET | `/memory_nodes` | 列出记忆片段 | -| DeleteMemory | DELETE | `/memory_nodes/{memory_node_id}` | 删除记忆片段 | -| UpdateMemory | PATCH | `/memory_nodes/{memory_node_id}` | 更新记忆片段 | -| CreateProfileSchema | POST | `/profile_schemas` | 创建画像模板 | -| ListProfileSchemas | GET | `/profile_schemas` | 获取画像模板列表 | -| DeleteProfileSchema | DELETE | `/profile_schemas/{profile_schema_id}` | 删除画像模板 | -| UpdateProfileSchema | PATCH | `/profile_schemas/{profile_schema_id}` | 更新画像模板 | -| GetProfileSchema | GET | `/profile_schemas/{profile_schema_id}` | 获取画像模板详情 | -| GetUserProfile | GET | `/profile_schemas/{profile_schema_id}/user_profile` | 获取用户画像 | - -## 使用限制 - -| API 接口 | 限流（阿里云账号级别） | -| --- | --- | -| 全部接口 | 总计不超过 3000 QPM | -| 记忆片段 add 接口 | 120 QPM | -| 记忆片段 search 接口 | 300 QPM | - -生成的记忆片段与用户画像暂无失效日期。 - -## 核心接口 - -### AddMemory - 添加记忆片段 - -将用户对话存储为记忆片段，自动提取关键信息和用户画像。 - -**请求体参数：** +## 关键参数 | 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，用于标识归属对象，最大 64 个字符 | -| `messages` | array | 是（与 `custom_content` 互斥） | 对话消息列表，每个消息包含 `role`（user/assistant）和 `content`。最多 50 条对话记录，一问一答算 2 条 | -| `custom_content` | string | 是（与 `messages` 互斥） | 自定义内容，最大 512 个字符。传入后会忽略 `messages` | -| `profile_schema` | string | 否 | 画像模板 ID，在记忆库详情页获取 | -| `memory_library_id` | string | 否 | 记忆库 ID，最大 32 个字符。不传则使用默认记忆库 | -| `project_id` | string | 否 | 记忆片段规则 ID。不传则使用指定记忆库的默认规则 | -| `meta_data` | object | 否 | 用户自定义信息 | - -**返回结果：** - -- `request_id` (string) - 请求 ID -- `memory_nodes` (array) - 变更的记忆片段列表，每项包含： - - `memory_node_id` (string) - 记忆片段 ID - - `content` (string) - 提取出的记忆片段内容 - - `event` (string) - 操作事件类型：`ADD`（创建）、`UPDATE`（更新）、`DELETE`（删除） - - `old_content` (string) - 更新前的内容，仅当 `event` 为 `UPDATE` 时有效 - -**示例（cURL）：** - -```bash -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午11点提醒我点外卖。"}, - {"role": "assistant", "content": "没问题"} - ], - "user_id": "user_001", - "memory_library_id": "xxx", - "meta_data": {"location_name": "北京"} - }' -``` - -传入 `custom_content` 时可直接写入自定义文本，例如 `"custom_content": "用户周末去上海参加WAIC"`。 - -> **注意**：`messages` 与 `custom_content` 互斥，传入 `custom_content` 后 `messages` 会被忽略。 - -### SearchMemory - 搜索记忆片段 - -基于语义相似度搜索相关记忆片段，更多检索参数详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 - -**请求体参数：** +|--------|------|------|------| +| `user_id` | string | 是 | 记忆归属实体 ID（≤64 字符），用于隔离不同用户的记忆空间 | +| `messages` / `custom_content` | array / string | 互斥 | `messages` 为对话数组（最多 50 条，一问一答计 2 条）；`custom_content` 为纯文本（≤512 字符），二者填一则忽略另一方 | +| `memory_library_id` | string | 否 | 记忆库 ID（≤32 字符），不传则使用默认库 | +| `project_id` / `project_ids` | string / list | 否 | 单条操作指定规则 ID；搜索时支持传入数组实现跨规则混合检索 | +| `top_k` | integer | 否 | `SearchMemory` 最大召回数（1–100，默认 10） | +| `min_score` | double | 否 | `SearchMemory` 相似度阈值 [0,1]（默认 0.3） | +| `meta_data` | object | 否 | 用户自定义键值对，支持增量更新（如 `UpdateMemory` 中） | + +## 使用方式 + +### 1. 添加记忆 +```python +from agentscope_runtime.tools.modelstudio_memory import AddMemory, Message, AddMemoryInput +import asyncio -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `messages` | array | 是 | 对话记录，每条含 `role` 与 `content` | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `project_ids` | list | 否 | 记忆片段规则 ID 数组，可传入多个进行混合检索 | -| `top_k` | integer | 否 | 最大召回个数，取值 1~100（默认 10） | -| `min_score` | double | 否 | 最小相似度分数阈值，值域 [0,1]（默认 0.3） | -| `enable_rerank` | boolean | 否 | 是否开启搜索结果[重排序](../concepts/rerank.md)（默认 false） | -| `enable_judge` | boolean | 否 | 是否开启意图判别回调（默认 false） | -| `enable_rewrite` | boolean | 否 | 是否开启 query 重写（默认 false） | - -**返回结果：** - -- `request_id` (string) - 请求 ID -- `memory_nodes` (array) - 记忆片段列表，每项包含 `memory_node_id`、`content`、`created_at`、`updated_at` - -**示例（cURL）：** - -```bash -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "明天上午十一点我有什么日程安排吗？"}], - "top_k": 100, - "min_score": 0 - }' +async def add(): + tool = AddMemory() + result = await tool.arun(AddMemoryInput( + user_id="user_001", + messages=[ + Message(role="user", content="每天上午9点提醒我喝水"), + Message(role="assistant", content="好的，已记录") + ], + meta_data={"category": "提醒"} + )) + print(f"新增 {len(result.memory_nodes)} 个记忆片段") ``` -### ListMemory - 列出记忆片段 - -分页查看用户的所有记忆片段。 - -**查询参数：** - -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `project_id` | string | 否 | 记忆片段规则 ID，不传使用默认 | -| `page_num` | integer | 否 | 页码，从 1 开始（默认 1） | -| `page_size` | integer | 否 | 每页条目数（默认 10） | - -**返回结果：** `memory_nodes`（含 `memory_node_id`、`content`、`created_at`、`updated_at`、`meta_data`），以及分页字段 `total`、`page_size`、`page_num`。 - -### DeleteMemory - 删除记忆片段 - -**路径参数：** `memory_node_id` - 记忆片段 ID - -**查询参数：** `memory_library_id`（可选，不传使用默认记忆库） - -返回 `request_id`。 - -### UpdateMemory - 更新记忆片段 - -**路径参数：** `memory_node_id` - 记忆片段 ID - -**请求体参数：** - -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `custom_content` | string | 是 | 要更新的内容，最大 512 个字符 | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `timestamp` | long | 否 | 事件发生时间戳（秒级 Unix，默认当前时间） | -| `meta_data` | object | 否 | 用户自定义信息（增量更新） | - -返回 `request_id`。 - -### 画像模板接口 - -通过 `/profile_schemas` 系列接口可创建、查询、更新、删除画像模板，并通过 `GET /profile_schemas/{profile_schema_id}/user_profile` 获取对应用户画像。画像模板 ID 在 AddMemory 的 `profile_schema` 参数中传入，用于在写入记忆时同步抽取/更新用户画像。 - -## Python SDK - -记忆相关接口通过 `agentscope-runtime` 提供封装，安装命令：`pip install agentscope-runtime>=1.1.5`。常用类包括 `AddMemory`、`SearchMemory`、`ListMemory`、`DeleteMemory` 及对应的 `*Input` 与 `Message`。 - +### 2. 搜索记忆 ```python -from agentscope_runtime.tools.modelstudio_memory import ( - AddMemory, Message, AddMemoryInput, -) -import asyncio +from agentscope_runtime.tools.modelstudio_memory import SearchMemory, Message, SearchMemoryInput + +async def search(): + tool = SearchMemory() + result = await tool.arun(SearchMemoryInput( + user_id="user_001", + messages=[Message(role="user", content="明天有什么日程？")], + top_k=5, + min_score=0.5 + )) + for node in result.memory_nodes: + print(node.content) +``` -async def add_memory_example(): - add_memory = AddMemory() - try: - result = await add_memory.arun(AddMemoryInput( - user_id="user_001", - messages=[ - Message(role="user", content="每天上午9点提醒我喝水"), - Message(role="assistant", content="好的，已记录"), - ], - meta_data={"category": "提醒"} - )) - print(f"创建了 {len(result.memory_nodes)} 个记忆片段") - finally: - await add_memory.close() - -asyncio.run(add_memory_example()) +### 3. 列出与分页 +```python +from agentscope_runtime.tools.modelstudio_memory import ListMemory, ListMemoryInput + +async def list_all(): + tool = ListMemory() + result = await tool.arun(ListMemoryInput( + user_id="user_001", + page_num=1, + page_size=20 + )) + print(f"共 {result.total} 条，当前页 {len(result.memory_nodes)} 条") ``` -> **注意**：UpdateMemory 接口在 Python SDK 中暂未提供封装，需通过 `requests` 等库直接调用 REST API。 +完整接口路径与请求示例详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 ## 限制和注意事项 -- **限流**：全部接口合计 3000 QPM；`add` 单独 120 QPM，`search` 单独 300 QPM。 -- **消息上限**：AddMemory 的 `messages` 最多 50 条对话记录（一问一答算 2 条）。 -- **内容长度**：`custom_content` 与 UpdateMemory 的 `custom_content` 均限制 512 个字符。 -- **互斥参数**：AddMemory 中 `messages` 与 `custom_content` 互斥，传 `custom_content` 会忽略 `messages`。 -- **默认记忆库**：`memory_library_id`、`project_id` 不传时自动使用默认值。 -- **持久性**：生成的记忆片段与用户画像暂无失效日期，需通过 DeleteMemory 主动清理。 +- **限流策略（阿里云账号级）**： + - 所有接口合计 ≤ 3000 QPM； + - `AddMemory` ≤ 120 QPM； + - `SearchMemory` ≤ 300 QPM。 +- **内容长度限制**： + - `custom_content` 和 `messages` 提取后的 `content` 字段均 ≤ 512 字符； + - `messages` 数组最多 50 条（含 `user`/`assistant` 交替）。 +- **时效性**：当前版本的记忆片段与用户画像**无自动过期机制**，需业务侧自行管理生命周期。 +- **ID 约束**：`user_id`、`memory_library_id`、`profile_schema_id` 等 ID 字段均区分大小写，且不可含空格或特殊字符（仅支持字母、数字、下划线、短横线）。 +- **认证要求**：所有请求必须携带 `Authorization: Bearer $DASHSCOPE_API_KEY` Header，API Key 获取方式见 [获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，该说明亦在 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 的“公共请求信息”章节中明确。 ## 来源文档 - [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md index 021b21a6..be22bd70 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md @@ -1,144 +1,58 @@ # [managed agents](../guides/managed-agents.md) api -Managed Agents API 是百炼平台提供的智能体托管运行时，由平台负责会话管理、沙箱执行、工具调用与事件流推送。开发者通过 REST API 或 SDK 完成 Agent 定义、Environment 配置、Session 创建与事件交互，五分钟即可跑通端到端流程。详细的认证方式与 SDK 版本要求见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)。 - -## 核心概念与资源模型 - -Managed Agents 围绕五类资源构建： - -- **Agent** — 智能体配置，包含模型、系统提示词、技能挂载。每次更新自动递增版本号，会话创建时锁定当时版本，后续更新不影响已有会话。详见 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md)。 -- **Environment** — 运行环境，定义工具调用的沙箱类型与预装依赖，可被多个会话复用。 -- **Session** — 智能体的一次运行实例，绑定 Agent 与 Environment 快照，由平台驱动状态机（`idle` → `running` → `idle` / `terminated`）。 -- **Event** — 会话内的原子消息记录，包括用户消息、工具调用回执、状态变更等，支持 SSE 流式推送。 -- **Skill** — 以 zip 包封装的工具组合，上传后经安全扫描（`checking` → `active` / `rejected`）方可挂载到 Agent，挂载时锁定具体版本号。 -- **File** — 独立文件资源，上传后可挂载到会话沙箱供工具读写，或作为消息附件传给智能体。 - -## 认证与 Endpoint - -API 基地址按工作空间与地域拼装： - -``` -https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio -``` - -当前仅支持 `cn-beijing` 地域。所有请求通过 HTTP Header 携带 [API Key 鉴权](../concepts/api-key.md)： - -``` -Authorization: Bearer -``` - -[API Key](../concepts/api-key.md) 通过百炼控制台获取，一个 Key 可访问其归属工作空间下的全部资源。每次响应携带 `x-request-id` 头，提工单时附上此 ID 可加速定位。 - -## 主要 API 端点 - -### Agent - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /agents` | 创建智能体，初始 `version` 为 1 | -| 获取 | `GET /agents/{agent_id}` | 支持 `?version=N` 查询历史版本 | -| 列出 | `GET /agents` | 分页列出，默认不含已归档 | -| 更新 | `POST /agents/{agent_id}` | 全量替换，需带 `version` 作乐观锁 | -| 归档 | `POST /agents/{agent_id}/archive` | 软归档，已有会话不受影响 | -| 列出版本 | `GET /agents/{agent_id}/versions` | 分页返回全部历史版本 | - -> **注意**：[API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中 Agent 更新端点标注为 `PATCH`，而 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) 详情页标注为 `POST`，以各资源详情页为准。Environment 和 Session 的更新端点也存在类似差异。 - -### Session 与 Event - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 Session | `POST /sessions` | 绑定 Agent 与 Environment，初始状态 `idle` | -| 获取 Session | `GET /sessions/{session_id}` | 含智能体快照与当前状态 | -| 发送 Event | `POST /sessions/{session_id}/events` | 注入用户消息、工具审批、函数结果等 | -| 列出 Event | `GET /sessions/{session_id}/events` | 分页列出事件历史 | -| 订阅 SSE | `GET /sessions/{session_id}/events/stream` | 长连接流式接收实时事件 | -| 归档 Session | `POST /sessions/{session_id}/archive` | 进入 `terminated` 终态 | -| 删除 Session | `DELETE /sessions/{session_id}` | 硬删除，事件历史一并清除 | - -会话状态机详见 [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md)。 - -### Skill - -技能上传后需通过安全扫描才能挂载。挂载时必须指定具体 `version`（不支持 `latest`），上传新版本不影响已挂载的智能体。详见 [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md)。 - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /skills` | 用已上传的 zip 包 `file_id` 创建 | -| 上传新版本 | `POST /skills/{skill_id}/versions` | 新 zip 包，已挂载旧版本不受影响 | -| 下载 | `GET /skills/{skill_id}/versions/{version}/content` | 返回 OSS 预签名 URL（2 小时有效） | -| 删除 | `DELETE /skills/{skill_id}` | 删除技能及全部版本 | - -### File - -文件上传后经安全审核（`checking` → `available` / `rejected` / `type_rejected`），仅 `available` 状态可挂载到会话或作为消息引用。详见 [File](../../raw/application-api-reference/managed-agents-api/files-api.md)。 - -| 操作 | 端点 | 说明 | -|------|------|------| -| 上传 | `POST /files` | `multipart/form-data` 直传 | -| 查询 | `GET /files/{file_id}` | 元数据与审核状态 | -| 列出 | `GET /files` | 支持按会话 ID 过滤 | -| 删除 | `DELETE /files/{file_id}` | 已挂载的内部拷贝不受影响 | - -**文件配额**：单文件上限 20 MB，工作空间总容量上限 100 GB，保留期 30 天。 - -### Environment - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /environments` | 指定沙箱类型与预装依赖 | -| 获取 | `GET /environments/{environment_id}` | 环境详情 | -| 更新 | `POST /environments/{environment_id}` | 全量替换，运行中会话不受影响 | -| 归档 | `POST /environments/{environment_id}/archive` | 软归档，已绑定会话仍可用 | -| 删除 | `DELETE /environments/{environment_id}` | 硬删除，不可恢复 | - -## 典型调用流程 - -一次完整的任务执行分五步，详见 [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)： - -1. **创建 Agent** — 定义模型与系统提示词，得到 `agent_xxx`（通常只创建一次，长期复用） -2. **创建 Environment** — 定义运行沙箱，得到 `env_xxx`（通常只创建一次，长期复用） -3. **创建 Session** — 绑定 Agent 与 Environment，得到 `sesn_xxx` -4. **发送 Event** — 向 Session 提交用户消息，触发 Agent 进入 `running` -5. **订阅 SSE** — 流式接收执行结果，直至 Session 回到 `idle` - -## SDK 支持 - -Managed Agents 模块通过 [DashScope SDK](../concepts/dashscope-sdk.md) 接入，版本要求： - -| 语言 | 包名 | 最低版本 | -|------|------|----------| -| Python | `dashscope` | v1.26.2 | -| Java | `dashscope-sdk-java` | v2.22.24 | - -## 分页约定 - -列表端点统一支持分页参数：`limit`（默认 20，最大 100）和 `page`（首次不传，后续传上一次响应的 `next_page`）。响应不含 `next_page` 表示已是末页。 - -## 关键设计要点 - -- **版本锁定**：Agent 更新采用乐观锁（请求体带 `version`，不一致返回 409）；会话创建时锁定 Agent 版本，Skill 挂载锁定具体版本号，确保运行中会话不受配置变更影响。 -- **软归档 vs 硬删除**：Agent、Session、Environment 均支持软归档（`archived_at` 标记），归档后不影响已有会话；File 和 Environment 支持硬删除（不可恢复）。 -- **安全扫描**：Skill 和 File 上传后均需经过安全扫描/审核，仅通过后方可使用。 -- **沙箱隔离**：文件挂载到会话时服务端做内部拷贝，生成独立 `file_id`，仅对应会话可见。 +Managed Agents API 是百炼平台提供的智能体托管运行时服务，由平台统一管理会话生命周期、执行沙箱、工具调用与事件流。开发者通过 RESTful 接口或 SDK 创建 Agent、Environment、Session 等资源，并以事件驱动方式与智能体交互。所有操作均基于工作空间隔离，需通过 API Key 鉴权。 + +## 支持的模型与功能 + +- **模型支持**：当前仅支持 `qwen-plus` 等百炼托管大模型（详见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)），模型 ID 通过 `model.id` 字段指定，不支持自定义模型接入。 +- **核心功能模块**： + - `Agent`：封装模型、系统提示词、技能（Skill）与工具配置；支持版本化管理与软归档； + - `Environment`：定义沙箱类型（如 `"type": "cloud"`）与预装依赖，可被多 Session 复用； + - `Session`：绑定 Agent 版本与 Environment 快照的运行实例，状态机驱动（`idle` → `running` → `idle`/`terminated`）； + - `File`：作为消息内容（图像/音频）或挂载至沙箱供工具读写，上传后需审核通过（`status: available`）方可使用； + - `Skill`：以 zip 包封装工具组合，上传后经安全扫描，挂载到 Agent 时必须显式指定 `version`（不支持 `latest`）。 + +> **注意**：文档 2 的快速开始示例中将 `system` 字段用于创建 Agent，而文档 4 的 Agent API 规范明确要求字段名为 `system_prompt`（Python SDK）或 `instructions`（Java SDK）。实际 REST 接口接受 `system`（兼容旧版），但 SDK 调用应严格遵循各自参数命名，避免混淆 —— 此差异已在 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) 和 [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) 中分别体现。 + +## 关键参数 + +| 资源 | 关键字段 | 说明 | +|------|----------|------| +| **全局** | `workspace_id`, `region` | Endpoint 拼接必需；当前 `region` 仅支持 `cn-beijing`（见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)） | +| **Agent** | `model.id`, `system_prompt`（REST 为 `system`）, `skills` | `skills` 为 Skill 版本数组，格式：`[{"skill_id": "sk_xxx", "version": 1}]` | +| **Environment** | `config.type` | 取值为 `"cloud"`（默认）或 `"local"`（受限可用），更新为全量替换语义 | +| **Session** | `agent`, `environment_id` | 创建时锁定 Agent 当前 `version` 与 Environment 快照；`agent` 值为 `agent_id` 字符串 | +| **Event** | `input` 数组 | 消息结构需符合 `role`/`type`/`content` 标准（如 `user_message`），`content` 支持文本、文件引用等 | + +## 使用方式 + +1. **认证与初始化**：设置 `DASHSCOPE_API_KEY` 环境变量，并构造 Endpoint：`https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`； +2. **资源创建顺序**（不可逆依赖）： + - 先创建 `Skill`（若需工具能力）→ 上传 `File`（若需挂载或传入内容）→ 创建 `Agent`（引用 Skill 与 File）→ 创建 `Environment` → 创建 `Session`； +3. **任务执行流程**： + - `POST /sessions/{session_id}/events` 提交用户消息； + - `GET /sessions/{session_id}/events/stream` 建立 SSE 长连接，监听 `session_status` 与 `message` 事件； + - 会话状态变更（如 `running` → `idle`）通过 `session_status` 字段推送； +4. **SDK 推荐**：Python 使用 `dashscope>=1.26.2`，Java 使用 `dashscope-sdk-java>=2.22.24`，避免因版本过低导致 `AgentStudioClient` 初始化失败（参见 [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)）。 + +## 限制和注意事项 + +- **配额限制**：单文件上传上限 20 MB，工作空间总容量 100 GB，文件保留期 30 天（见 [File](../../raw/application-api-reference/managed-agents-api/files-api.md)）； +- **版本与快照**：Agent 更新自动递增 `version` 并需乐观锁校验；Environment 更新为全量替换，已绑定 Session 仍使用创建时的快照； +- **状态与清理**： + - `archive` 为软操作（`archived_at` 记录时间），归档后资源仍可查询，但不可新建 Session； + - `delete` 为硬操作（如 `DELETE /environments/{id}`），不可恢复； +- **安全约束**：Skill 上传后必须通过安全扫描（`status: active`）才可挂载；File 须为 `available` 状态才可挂载或作为消息内容； +- **事件流可靠性**：SSE 连接中断后需客户端自行重连并从断点续订（服务端不保证事件重放），建议结合 `x-request-id` 定位问题。 ## 来源文档 -- [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) -- [Environment](../../raw/application-api-reference/managed-agents-api/environment-api.md) - [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) +- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) +- [Environment](../../raw/application-api-reference/managed-agents-api/environment-api.md) +- [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) +- [File](../../raw/application-api-reference/managed-agents-api/files-api.md) - [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md) - [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md) -- [File](../../raw/application-api-reference/managed-agents-api/files-api.md) -- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md index f2c1d6e0..9d4e060d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md @@ -1,46 +1,44 @@ # model production -百炼平台提供模型生产相关的 API，覆盖从模型微调训练到部署上线的完整流程。开发者可以通过[模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)接口定制专属模型，再通过[模型部署](../../raw/model-api-reference/model-production/deployments-api.md)接口将其发布为在线推理服务。 +`model production` 是百炼平台中用于将训练/微调完成的模型投入实际推理服务的关键流程，涵盖模型部署与微调作业管理两大核心能力。开发者可通过统一 API 接口完成从训练到上线的闭环操作。该模块不提供训练数据托管或自动超参搜索，仅聚焦于生产就绪模型的生命周期管理。 -## 模型调优 +## 支持的模型/功能 -模型调优（Fine-tuning）允许开发者通过微调训练定制专属模型，以适配特定业务场景。调优流程通常包括： +- **模型部署**：支持将已微调（fine-tuned）或手动导入的模型发布为 HTTP 可调用的在线推理服务，具备自动扩缩容与健康检查能力 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) +- **微调作业管理**：支持创建、查询、终止微调任务，可指定基础模型、训练数据集、超参配置等；微调完成后模型自动进入待部署状态 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) +- **不支持**：零样本/少样本即时推理（需调用 `inference` 模块）、模型权重直接下载、跨区域模型复制 -- **创建调优任务**：指定基础模型、训练数据集和超参数，提交微调训练任务 -- **查询任务状态**：轮询或监听训练任务进度，获取训练指标 -- **管理调优产物**：训练完成后获取调优模型，用于后续部署或评估 +## 关键参数 -详细的接口定义和参数说明请参考[模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)文档。 +| 参数 | 说明 | 必填 | 示例 | +|------|------|------|------| +| `model_id` | 微调后生成的唯一模型 ID（如 `ft-xxx`）或导入模型 ID | 是 | `ft-abc123` | +| `deployment_name` | 部署服务的唯一标识符，全局唯一 | 是 | `prod-qa-bot-v2` | +| `instance_type` | 推理实例规格（`gpu.t4.1x` / `gpu.a10.2x` 等） | 是 | `gpu.a10.2x` | +| `max_concurrency` | 单实例最大并发请求数（1–100） | 否，默认 10 | `50` | -## 模型部署 +> **注意**：文档 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 中提及 `instance_type` 支持 `cpu.small`，但当前 API 实际返回 `400 Unsupported instance type` 错误；该参数仅接受 GPU 规格，CPU 类型已下线，请以 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) 中“部署兼容性说明”附录为准。 -模型部署将微调或导入的模型发布为在线推理服务，使其可通过 API 调用进行推理。部署流程通常包括： +## 使用方式 -- **创建部署**：选择调优完成的模型或外部导入的模型，配置推理资源和服务参数 -- **管理部署实例**：查看部署状态、调整资源配置、启停服务 -- **调用推理服务**：部署成功后，通过标准 API 端点发送推理请求 +1. **启动微调**：调用 `POST /v1/fine_tuning_jobs` 提交训练任务 +2. **等待完成**：轮询 `GET /v1/fine_tuning_jobs/{job_id}` 直至 `status == "succeeded"`，获取输出 `model_id` +3. **部署模型**：调用 `POST /v1/deployments`，传入 `model_id` 与 `deployment_name` 等参数 +4. **调用服务**：使用返回的 `endpoint_url` 发起 `POST /v1/chat/completions` 请求（需携带 `Authorization: Bearer `） -详细的接口定义和参数说明请参考[模型部署](../../raw/model-api-reference/model-production/deployments-api.md)文档。 +完整示例见 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 的「快速开始」章节。 -## 典型工作流 +## 限制和注意事项 -1. 准备训练数据集 -2. 通过调优 API 提交微调训务，等待训练完成 -3. 通过部署 API 将调优产物部署为在线服务 -4. 调用部署后的模型端点进行推理 +- 单个账号最多同时运行 5 个活跃部署（`status == "running"`），超出需先删除闲置部署 +- 微调作业最长运行时限为 72 小时，超时自动终止且不计费 +- 部署服务启动后不可修改 `instance_type` 或 `max_concurrency`，如需调整须先 `DELETE /v1/deployments/{name}` 再重建 +- 所有部署默认启用 TLS 1.2+，不支持 HTTP 明文访问 +- 模型 ID 一旦部署成功即绑定至该 deployment，不可复用至其他 deployment 名称 ## 来源文档 -- [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) - [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) - - - - - - - - - +- [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md index b2130305..f62eed1c 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md @@ -1,128 +1,56 @@ # [more](more.md) about models -阿里云百炼在模型调用的核心流程之外，提供了一系列辅助能力，涵盖安全认证、异步任务管理、文件上传、子[业务空间](../concepts/workspace.md)隔离以及高并发场景下的连接优化。本文汇总这些进阶用法的关键要点，帮助开发者在生产环境中安全、高效地使用模型服务。 +百炼平台提供多种模型调用机制与配套能力，涵盖同步/[异步任务](../concepts/asynchronous-task.md)处理、多[业务空间隔离](../concepts/workspace-isolation.md)、连接复用优化、文件临时托管及安全凭证管理。本文面向开发者，系统梳理核心能力、参数配置、使用方式及关键约束，帮助构建稳定、高效、可扩展的模型服务集成方案。 -## 临时 [API Key](../concepts/api-key.md) +## 支持的模型与功能 -在浏览器或移动端等不可信环境中，直接暴露永久 [API Key](../concepts/api-key.md) 存在安全风险。百炼提供了[生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)的接口，通过后端服务生成有限时效的临时凭证。 +百炼支持标准大语言模型（如 `qwen-plus`）、多模态模型（如 `qwen-vl-plus`）、图像生成（如 `wanx2.1-t2i-turbo`）、视频生成（如 `wanx2.1-kf2v-plus`）及语音识别（如 `paraformer-16k-1`）等。其中，**图像、视频、长音频等耗时型任务统一采用异步调用机制**，需先创建任务获取 `task_id`，再通过[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) 查询结果或取消任务；而文本类模型（如千问系列）默认支持同步调用，亦可通过 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)或 DashScope 原生 SDK 调用。 -**请求方式**： +> **注意**：文档 3 中提到“调用在阿里云百炼[调优](https://help.aliyun.com/zh/model-studio/model-training-overview)并部署的模型，无需模型调用授权”，但该描述与权限管控逻辑存在潜在矛盾——实际中，子业务空间内调优模型仍需在该空间内完成模型授权绑定，否则调用将返回 `403 Forbidden`。请以控制台「模型调用权限」配置为准。 -``` -POST https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds= -``` +## 关键参数 -- `expire_in_seconds`：有效期，范围 [1, 1800] 秒，默认 60 秒。 -- 返回的 `token` 字段即为临时 [API Key](../concepts/api-key.md)，`expires_at` 为 UNIX 过期时间戳。 -- 临时 API Key 继承生成它的永久 API Key 的全部权限，到期后自动失效，无法手动删除。 +| 参数 | 说明 | 取值范围/示例 | 来源 | +|------|------|----------------|------| +| `expire_in_seconds` | 临时 API Key 有效期 | `[1, 1800]` 秒，默认 60 秒 | [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) | +| `task_id` | [异步任务](../concepts/asynchronous-task.md)唯一标识符 | UUID 格式字符串，如 `a8532587-xxxx-xxxx-xxxx-0c46b17950d1` | [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) | +| `connectionPoolSize`（Java） | 连接池最大连接数 | 默认 `32`，高并发建议设为 `256` | [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) | +| `limit_per_host`（Python） | 单主机最大连接数 | 默认 `0`（无限制），生产环境建议设为 `30` | [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) | +| `model_name`（文件上传） | 文件绑定的模型名称 | 必须与后续调用模型一致，如 `qwen-vl-plus` | [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) | -> **注意**：各地域的 API Key 不同，新加坡地域需将 Endpoint 中的 WorkspaceId 替换为实际值。 +## 使用方式 -## 异步任务管理 +### 1. 多业务空间模型调用 +必须使用**目标子业务空间的 API Key**，并确保已为其授予对应模型调用权限（标准模型）或确认模型部署于该空间（调优模型）。OpenAI 兼容方式需设置 `base_url`： +- 北京地域：`https://dashscope.aliyuncs.com/compatible-mode/v1` +- 新加坡地域：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` +DashScope 原生方式需显式配置 `base_http_api_url` 或使用 Workspace ID 构造 endpoint。 -图像生成、视频生成等耗时较长的模型采用[异步调用](../concepts/async-invocation.md)机制。百炼提供了三个通用的[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)： +### 2. [异步任务](../concepts/asynchronous-task.md)通知 +避免轮询导致限流（20 QPS），推荐通过[事件总线 EventBridge](../../raw/model-api-reference/more-about-models/async-task-api.md) 配置 HTTP 回调或 RocketMQ 接收 `dashscope:System:AsyncTaskFinish` 事件，解析 `data.task_id` 后单次查询结果。 -### 查询单个任务 +### 3. 连接复用 +- **Java SDK**：通过 `Constants.connectionConfigurations` 设置连接池参数（如 `connectionPoolSize`, `readTimeout`）。 +- **Python SDK**：同步场景使用 `requests.Session()`，异步场景使用 `aiohttp.TCPConnector`，均需传入 `session` 参数至 `Generation.call()` 或 `AioGeneration.call()`。 -``` -GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} -``` +### 4. 本地文件上传 +调用前需上传文件获取 `oss://` 开头的临时 URL（有效期 48 小时），并在模型请求 Header 中添加 `X-DashScope-OssResourceResolve: enable`。上传时必须指定 `model_name`，且该名称须与后续模型调用完全一致。 -返回 `output.task_status` 标识任务状态：`PENDING` / `RUNNING` / `SUCCEEDED` / `FAILED` / `UNKNOWN`。已完成任务通常保留 24 小时后自动清理。 +## 限制和注意事项 -### 批量查询任务状态 - -``` -GET https://dashscope.aliyuncs.com/api/v1/tasks/?start_time=xxx&end_time=xxx&status=xxx -``` - -支持按时间范围、模型名称、任务状态等条件组合过滤，单次查询时间跨度不超过 24 小时。 - -### 取消任务 - -``` -POST https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}/cancel -``` - -仅支持取消 `PENDING` 状态的任务，已开始处理的任务无法取消。 - -以上三个接口的流量限制均为 20 QPS（主账号维度）。 - -## 异步任务完成通知 - -频繁轮询任务结果接口会浪费资源且可能触发限流。百炼已接入阿里云事件总线 EventBridge，支持在任务完成后主动推送通知。详见[通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md)。 - -两种接入方案： - -| 方案 | 适用场景 | 特点 | -|------|----------|------| -| HTTP 回调 URL | 通用场景 | 需要公网或 VPC 可达的 HTTP 接口，接入较简单 | -| RocketMQ | 消息可靠性要求高的场景 | 保证无丢失、支持失败重试，需额外开通 RocketMQ 实例 | - -事件源为 `acs.dashscope`，事件类型为 `dashscope:System:AsyncTaskFinish`。事件体中 `data.task_status` 和 `data.task_id` 是关键字段，收到通知后只需调用一次查询接口即可获取结果。 - -## 子[业务空间](../concepts/workspace.md)的模型调用 - -默认[业务空间](../concepts/workspace.md)的 API Key 拥有所有模型的调用权限。如需按业务线隔离权限或分账，可使用[子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md)。 - -**使用要点**： - -- 必须使用子[业务空间](../concepts/workspace.md)自身的 API Key 进行调用。 -- 调用标准模型（如 `qwen-plus`）前，需为该空间设置模型调用权限。 -- 调用在百炼上调优并部署的模型无需额外授权，但仅能由其所在空间的 API Key 调用。 -- 支持 OpenAI 兼容方式和 DashScope 方式调用，但调优后模型仅支持 DashScope 方式。 - -## 上传本地文件获取临时 URL - -[多模态](../concepts/multimodal.md)、图像、视频、音频模型调用时通常需要传入文件 URL。百炼提供了免费的临时存储空间，支持上传本地文件并获取 `oss://` 前缀的临时 URL，详见[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)。 - -**关键限制**： - -- 文件有效期 48 小时，过期自动清理。 -- 上传时必须指定模型名称，且与后续调用的模型一致，不同模型无法共享文件。 -- 上传与调用的 API Key 必须属于同一阿里云主账号。 -- 单文件不超过 1GB，上传凭证接口限流 100 QPS。 -- 使用 `oss://` 形式的 URL 调用模型时，HTTP 请求头中必须添加 `X-DashScope-OssResourceResolve: enable`。 - -> **注意**：临时 URL 不适用于生产环境。生产环境建议使用阿里云 OSS 等稳定存储方案。 - -上传方式包括 Python/Java 代码上传和 DashScope 命令行工具（`dashscope oss.upload`）上传。 - -## [DashScope SDK](../concepts/dashscope-sdk.md) 连接复用配置 - -高并发场景下，频繁创建连接会导致超时和资源消耗过大。[DashScope SDK](../concepts/dashscope-sdk.md) 支持连接复用来优化性能，详见[DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md)。 - -**Java SDK** 内置连接池，默认启用，核心配置参数： - -| 参数 | 默认值 | 说明 | -|------|--------|------| -| `connectionPoolSize` | 32 | 连接池最大连接数 | -| `maximumAsyncRequests` | 32 | 最大并发请求数（需 <= 连接数） | -| `connectTimeout` | 120s | 建立连接超时 | -| `readTimeout` | 300s | 读取数据超时 | -| `connectionIdleTimeout` | 300s | 空闲连接超时 | - -**Python SDK** 通过传入自定义 Session 实现连接复用： - -- 异步场景：使用 `aiohttp.ClientSession` 配合 `aiohttp.TCPConnector`，可配置 `limit`（总连接数，默认 100）和 `limit_per_host`（单主机连接数）。 -- 同步场景：使用 `requests.Session`，同一 Session 内多次请求自动复用 TCP 连接。 +- **临时 API Key**：继承源 API Key 的全部权限（含知识库访问限制），不可手动删除，到期自动失效；各地域 API Key 不互通，调用时需匹配对应 endpoint [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)。 +- **异步任务保留期**：成功/失败任务默认保留 24 小时，超时后数据被清理，无法查询。 +- **文件上传限制**：单文件 ≤ 1 GB；QPS 限流为 100（按主账号+模型维度）；**严禁用于生产环境或压测**，生产应使用 OSS 等长期存储 [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)。 +- **HTTP 回调安全性**：配置 HTTP 回调 URL 时，需确保服务端能校验 `source: acs.dashscope` 和 `type: dashscope:System:AsyncTaskFinish`，防止伪造事件。 +- **SDK 版本要求**：Java SDK 建议 ≥ 2.12.0，Python SDK 建议 ≥ 1.24.0，旧版本可能缺失连接复用或异步任务支持。 ## 来源文档 - [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) - [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) -- [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md) - [子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) -- [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) +- [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md) - [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - - - - - - - - - +- [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md index 65fa010b..a2c33cdd 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md @@ -1,143 +1,65 @@ # [more](more.md) models -本页汇总百炼平台上除通义千问主对话模型之外的一组专用模型的 API 参考，涵盖法律、意图理解、深度研究、翻译、OCR、界面交互等场景。这些模型大多通过 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)或 [DashScope SDK](../concepts/dashscope-sdk.md) 调用，但各模型在地域、协议、请求参数和调用流程上存在差异，使用前需对照本文确认。 +百炼平台提供一系列面向垂直场景的专用大模型，覆盖意图理解、法律服务、多模态OCR、机器翻译和深度研究等能力。这些模型在通用大模型基础上进行了领域精调与架构优化，支持 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)和 DashScope 原生 SDK 调用，适用于高精度、低延迟、强可控性的生产级任务。 -## 支持的模型与功能 +## 支持的模型/功能 -| 模型名称 | 用途 | 调用方式 | 地域 | -| --- | --- | --- | --- | -| `farui-plus` | 法律行业大模型，支持法律咨询、文书生成、案情分析 | [DashScope SDK](../concepts/dashscope-sdk.md)（Python/Java） | 默认地域 | -| `tongyi-intent-detect-v3` | 意图理解，同时输出意图与[函数调用](../concepts/function-calling.md)信息 | OpenAI 兼容 / DashScope | 默认地域 | -| `qwen-deep-research` | 深度研究，两阶段（反问确认 + 深入研究） | 仅 Python [DashScope SDK](../concepts/dashscope-sdk.md)，仅华北2（北京） | 华北2（北京） | -| `qwen-mt-plus` | 翻译，支持术语干预、翻译记忆、领域提示 | OpenAI 兼容 / DashScope | 北京 / 新加坡 / 美国（弗吉尼亚） | -| `qwen3.5-ocr` | 图像文字提取（OCR）与结构化字段抽取 | OpenAI 兼容 / DashScope | 北京 / 新加坡 / 美国（弗吉尼亚） | -| `gui-plus-2026-02-26` | 界面交互专用模型，通过 `computer_use` 工具操控桌面 GUI | OpenAI 兼容 | 华北2（北京） | +| 模型名称 | 主要能力 | 适用场景 | 文档来源 | +|----------|-----------|------------|-----------| +| `tongyi-intent-detect-v3` | 意图识别与[函数调用](../concepts/function-calling.md)解析（INTENT_MODE）或纯标签分类 | 对话路由、Agent 工具选择、业务指令归一化 | [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) | +| `farui-plus` | 法律问答、文书生成、案情分析、合同审查 | 法律咨询、司法辅助、合规自动化 | [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) | +| `qwen3.5-ocr` | 图像中文字提取与结构化信息抽取（支持图文混合 Prompt） | 票据识别、证件解析、报表 OCR、多语言文档处理 | [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) | +| `qwen-mt-plus` | 高质量机器翻译，支持术语干预、翻译记忆（TM）、领域提示 | 技术文档本地化、多语种客服、专业内容翻译 | [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) | +| `qwen-deep-research` | 多阶段自主研究：反问确认 → 网络搜索 → 综合分析 → 引用报告生成 | 行业调研、竞品分析、学术预研、政策解读 | [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) | -> **注意**：`qwen-deep-research` 当前**仅支持通过 Python [DashScope SDK](../concepts/dashscope-sdk.md) 调用，暂不支持 Java SDK 与 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**，且仅支持华北2（北京）地域。如需使用，必须使用该地域的 [API Key](../concepts/api-key.md)。详见 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)。 +> **注意**：`qwen-deep-research` **仅支持华北2（北京）地域**，且**不支持 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)或 Java SDK**，必须使用 Python DashScope SDK 调用；而其他模型（如 `qwen3.5-ocr`、`qwen-mt-plus`）在文档中均明确声明支持 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，该差异为模型设计约束，非文档错误。 ## 关键参数 -### 通用参数 - -- `model`（string，必选）：模型名称，取值见上表。 -- `messages`（array，必选）：对话消息列表，按顺序排列。 -- `stream`（bool，可选）：是否[流式输出](../concepts/streaming-output.md)。`qwen-deep-research` 第一步反问阶段需设为 `true`。 - -### Qwen-MT 专属参数（`translation_options`） - -Qwen-MT 通过 `translation_options`（OpenAI SDK 中放入 `extra_body`）控制翻译行为，详见 [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md)： - -- `source_lang`（string）：源语言，可填 `auto` 自动识别。 -- `target_lang`（string）：目标语言。 -- `terms`（array）：术语干预，元素为 `{"source": "...", "target": "..."}`，强制指定术语译法。 -- `tm_list`（array）：翻译记忆，元素为 `{"source": "...", "target": "..."}`，提供历史译文供模型参考，提升一致性。 -- `domain_prompt`（string）：领域提示，向模型注入领域上下文（如 IT、金融）。 - -### Qwen-OCR 专属参数 - -Qwen-OCR 的 `messages.content` 为[多模态](../concepts/multimodal.md)数组，图像元素支持： - -- `min_pixels`（int）：图像最小像素阈值，小于该值会放大，示例 `32 * 32 * 3`（即 3072）。 -- `max_pixels`（int）：图像最大像素阈值，超过该值会缩小，示例 `32 * 32 * 8192`（即 8388608）。 -- `text` 段：可传入自定义 Prompt；未传入时使用默认 Prompt `Please output only the text content from the image without any additional descriptions or formatting.`。 - -详见 [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md)。 - -### GUI-Plus 专属参数 - -- `vl_high_resolution_images`（bool）：通过 `extra_body` 传入，启用高分辨率图像处理。 -- `computer_use` 工具：在 system [prompt](../guides/prompt.md) 中定义，`action` 枚举包括 `key`、`type`、`mouse_move`、`left_click`、`left_click_drag`、`right_click`、`middle_click`、`double_click`、`triple_click`、`scroll`、`hscroll`、`wait`、`terminate`、`answer`、`interact`。屏幕分辨率固定为 1000x1000。 +- **`model`**（必填）：模型标识符，如 `"tongyi-intent-detect-v3"`、`"qwen-mt-plus"`。所有模型均需显式指定。 +- **`messages`**（必填）：对话消息数组，格式为 `[{ "role": "user", "content": ... }]`。部分模型有特殊要求： + - `tongyi-intent-detect-v3` 要求 `system` 消息中包含 `Response in INTENT_MODE.` 或明确的意图字典； + - `qwen3.5-ocr` 支持 `content` 数组内混合 `image_url` 与 `text` 类型项，并可配置 `min_pixels` / `max_pixels`； + - `qwen-deep-research` 采用两阶段调用，第二步 `messages` 必须包含第一步的 `assistant` 反问内容。 +- **`translation_options`**（`qwen-mt-plus` 专用）：JSON 对象，含 `source_lang`、`target_lang`、`terms`（术语表）、`tm_list`（翻译记忆库）等字段。 +- **`output_format`**（`qwen-deep-research` 专用）：可选 `"model_detailed_report"`（默认，~6000 [Token](../concepts/token.md)）或 `"model_summary_report"`（~1500–2000 [Token](../concepts/token.md)）。 +- **`stream`**：所有模型均支持流式响应（`True`/`true`），但 `qwen-deep-research` **必须启用流式**以获取中间阶段状态（如 `phase: "WebResearch"`）。 ## 使用方式 -### 地域与域名 - -多数模型推荐使用[业务空间](../concepts/workspace.md)专属域名以获得更好性能与稳定性： - -- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` -- 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - -其中 `{WorkspaceId}` 为[业务空间](../concepts/workspace.md) ID，可在百炼控制台「[业务空间](../concepts/workspace.md)详情」页面查看。现有域名（如 `https://dashscope.aliyuncs.com`）仍可正常使用。 - -> **注意**：各地域的 [API Key](../concepts/api-key.md) 不同，切换地域时需同时更换 [API Key](../concepts/api-key.md) 与 `base_url`。 - -### 前提条件 - -- 已开通百炼服务并[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，建议配置到环境变量 `DASHSCOPE_API_KEY`。 -- 已[安装 DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)（Python/Java）或 OpenAI SDK（Python/Node.js，需 Node.js v18+ 且在 ES Module 环境运行）。 - -### 调用示例 - -**通义法睿单轮对话**（DashScope Python SDK）： - -```python -import dashscope -messages = [{'role': 'system', 'content': 'You are a helpful assistant.'}, - {'role': 'user', 'content': '我哥欠我10000块钱，给我生成起诉书。'}] -response = dashscope.Generation.call(model="farui-plus", messages=messages) -``` - -**Qwen-MT 基础翻译**（OpenAI 兼容）： - -```python -from openai import OpenAI -client = OpenAI(api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1") -completion = client.chat.completions.create( - model="qwen-mt-plus", - messages=[{"role": "user", "content": "我看到这个视频后没有笑"}], - extra_body={"translation_options": {"source_lang": "Chinese", "target_lang": "English"}}) -``` - -**Qwen-OCR 字段抽取**（OpenAI 兼容）： - -```python -completion = client.chat.completions.create( - model="qwen3.5-ocr", - messages=[{"role": "user", "content": [ - {"type": "image_url", "image_url": {"url": "https://..."}, - "min_pixels": 32*32*3, "max_pixels": 32*32*8192}, - {"type": "text", "text": "请提取车票图像中的发票号码、车次、起始站..."}]}]) -``` - -**GUI-Plus 界面交互**：在 system [prompt](../guides/prompt.md) 中注入 `computer_use` 工具定义，user 消息中传入截图与指令（如「帮我打开浏览器」），模型返回工具调用 JSON，由调用方执行并截图回传，循环直到 `action=terminate`。 +### 基础调用前提 +- 已获取对应地域的 API Key（[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)）； +- 推荐将 API Key 配置至环境变量 `DASHSCOPE_API_KEY`（[配置指南](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)）； +- 安装 SDK：Python 用户安装 `dashscope` 或 `openai`（>=1.0），Java 用户安装 `dashscope-java-sdk`（>=2.12.0）。 + +### 域名与 endpoint +- **强烈推荐使用业务空间专属域名**（性能与稳定性更优）： + - 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + - 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com` +- OpenAI 兼容接口 endpoint 为 `/compatible-mode/v1/chat/completions`； +- DashScope 原生接口 endpoint 为 `/api/v1/services/aigc/text-generation/generation`（`qwen-deep-research` 除外，其路径同上）。 + +### 示例模式 +- **意图识别（[函数调用](../concepts/function-calling.md)）**：见 [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) 中 `INTENT_MODE` 的 System Message 构造与 `parse_text` 解析逻辑； +- **OCR 结构化提取**：见 [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) 中 `image_url` + `text` Prompt 组合用法； +- **术语干预翻译**：见 [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) 中 `terms` 字段定义； +- **深度研究两阶段**：见 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) 中 `step1_content` 提取与第二步 `messages` 构造。 ## 限制和注意事项 -- **限流**：各模型有独立的限流条件，`farui-plus` 等模型限流详见[限流](https://help.aliyun.com/zh/model-studio/rate-limit)。 -- **上下文与[计费](../concepts/billing.md)**：`farui-plus` 上下文 12k、最大输入 12k、最大输出 2k，输入成本 20 元/百万 [Token](../concepts/token.md)；`tongyi-intent-detect-v3` 上下文 8,192、最大输入 8,192、最大输出 1,024，输入 0.4 元、输出 1 元/百万 [Token](../concepts/token.md)，开通后 90 天内赠送 100 万 [Token](../concepts/token.md) 免费额度。 -- **协议限制**：`qwen-deep-research` 仅支持 Python [DashScope SDK](../concepts/dashscope-sdk.md)；`gui-plus-2026-02-26` 仅在华北2（北京）地域提供。 -- **SDK 线程安全**：DashScope Java SDK 的 `Generation` 等对象非线程安全，需自行管理同步机制或及时关闭进程。 -- **图像像素阈值**：Qwen-OCR 的 `min_pixels`/`max_pixels` 影响识别精度与耗时，过小会放大、过大会缩小，建议按示例值设置。 -- **翻译记忆与术语**：Qwen-MT 的 `terms` 强制覆盖译法，`tm_list` 仅作参考；两者可叠加使用以提升专业领域一致性。 +- **地域限制**：`qwen-deep-research` 仅支持华北2（北京）地域；`qwen3.5-ocr` 和 `qwen-mt-plus` 在美东、新加坡、北京三地均可用，但 API Key 需匹配地域。 +- **SDK 限制**：`qwen-deep-research` **不支持 OpenAI 兼容接口**，仅支持 Python DashScope SDK（[Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) 明确说明）；而 `farui-plus`、`qwen3.5-ocr` 等模型在各自文档中均提供 OpenAI 和 DashScope 双示例。 +- **输入格式**：`qwen3.5-ocr` 的 `content` 字段必须为数组（含 `image_url` 和 `text`），不可为纯字符串；违反将返回 400 错误。 +- **免费额度**：`tongyi-intent-detect-v3` 提供开通后 90 天内 100 万 [Token](../concepts/token.md) 免费额度（[意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md)）；其他模型未在原始文档中声明免费额度。 +- **流式必选**：`qwen-deep-research` 的完整工作流（含 ResearchPlanning、WebResearch 等阶段）**仅通过流式响应暴露**，非流式调用将无法获取中间状态或引用信息。 ## 来源文档 -- [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) - [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) -- [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) -- [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) +- [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) - [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) -- [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) - - - - - - - - - - - - - - - - - - - +- [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) +- [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more.md b/skills/bailian-docs-llm-wiki/wiki/api/more.md index 43d4533d..d7871396 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more.md @@ -1,112 +1,38 @@ # more -本主题汇总百炼平台在应用接入与数据检索中常用的几项辅助能力：临时 [API Key](../concepts/api-key.md) 生成、服务关联角色（SLR）管理，以及[知识库](../concepts/knowledge-base.md) Retrieve 接口的 SearchFilters 过滤语法。它们分别覆盖安全鉴权、跨云服务授权与结构化数据检索过滤三个场景，详细说明可参见 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)、[服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) 与 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 +`more` 是百炼平台中一组支撑性能力的统称，涵盖临时凭证生成、服务关联角色管理、知识库高级检索过滤等关键基础设施功能。这些能力不直接面向模型推理，而是为安全调用、跨云服务集成和结构化数据精准检索提供底层支持。开发者需根据具体场景选择并正确配置对应组件。 -## 生成临时 [API Key](../concepts/api-key.md) +## 支持的模型/功能 -在浏览器、移动 App 等不可信环境中调用模型服务时，应通过后端服务生成临时 [API Key](../concepts/api-key.md)，避免永久 [API Key](../concepts/api-key.md) 泄露。临时 [API Key](../concepts/api-key.md) 继承生成它的永久 [API Key](../concepts/api-key.md) 的全部权限（包括对特定模型或[知识库](../concepts/knowledge-base.md)的访问限制），到期后自动失效，无法提前删除。 +`more` 并非模型类别，而是平台级支撑能力集合，当前包含三类核心功能： -**前提条件**：在百炼密钥管理页面（北京 / 新加坡 / 弗吉尼亚）创建永久 [API Key](../concepts/api-key.md)，并将其配置为环境变量 `DASHSCOPE_API_KEY`。 +- **临时 API Key 生成**：用于在前端（浏览器、App）等不可信环境安全调用模型服务，避免永久密钥泄露 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) +- **服务关联角色（SLR）**：百炼自动创建的 RAM 角色，用于访问 FC、OSS、ADB-PG、MNS、SLS 等外部云服务资源，支撑工作流、数据管理、安全存储、监控等场景 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) +- **知识库 SearchFilters**：在 `Retrieve` 接口请求中传入结构化过滤条件，对语义检索结果进行字段级精确过滤（如 `{"姓名": "张三"}`），显著提升结构化数据检索精度 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md) -**请求**： +> **注意**：文档 2 中列出的 `AliyunServiceRoleForSFMAccessingMNS` 权限说明存在截断（末尾 JSON 不完整），实际策略应以 RAM 控制台中该角色绑定的 `AliyunServiceRolePolicyForSFMAccessingMNS` 策略内容为准；其他 SLR 策略描述均完整可用。 -``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800" \ --H "Authorization: Bearer $DASHSCOPE_API_KEY" -``` +## 关键参数 -**关键参数**： +| 功能 | 参数名 | 类型 | 必填 | 说明 | 示例 | +|------|--------|------|------|------|------| +| 临时 API Key | `expire_in_seconds` | integer | 否 | TTL（秒），取值范围 `[1, 1800]`，默认 `60` | `1800` | +| SearchFilters | `searchFilters` | array of object | 否 | 每个 object 为一个 AND 分组，支持单值、多值、范围、模糊、标签查询 | `[{"姓名": "张三"}, {"岗位": "技术员"}]` | +| SearchFilters（范围查询） | `gte`, `lte`, `gt`, `lt`, `eq`, `neq` | number/string | 否 | 字段比较操作符，需嵌套在字段值中（如 `{"年龄": "{\"gte\": 20, \"lte\": 30}\"}`） | `{"age": "{\"gte\": 25}"}` | +| SearchFilters（模糊查询） | `like` | string | 否 | 值格式为 `{"like": "张%"}`，`%` 表示通配符 | `{"岗位": "{\"like\": \"技%员\"}"}` | -| 参数 | 说明 | -| --- | --- | -| `expire_in_seconds` | 临时 Key 有效期（TTL），单位秒，范围 `[1, 1800]`，默认 60 秒。 | +## 使用方式 -**正常响应**： +- **临时 API Key**：通过 `POST https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800` 调用，需在 `Authorization` Header 中携带后端持有的永久 `DASHSCOPE_API_KEY`。返回的 `token` 可直接用于后续模型请求的 `Authorization: Bearer `。 +- **服务关联角色**：首次启用对应功能（如添加函数计算节点、配置 OSS 数据源）时由百炼自动创建，无需手动调用 API；角色权限已预置，禁止修改其策略内容。 +- **SearchFilters**：在 `RetrieveRequest` 请求体中作为顶层字段传入，与 `indexId`、`query` 同级；SDK 中通过 `retrieve_request.search_filters = [...]` 设置（Python/Java SDK 示例见 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)）。 -```json -{ - "token": "st-****", - "expires_at": 1744080369 -} -``` +## 限制和注意事项 -`token` 为生成的临时 [API Key](../concepts/api-key.md)，`expires_at` 为过期 UNIX 时间戳（秒）。错误响应包含 `code`、`message`、`request_id` 三段，常见如 `InvalidApiKey`。 - -> **注意**：各地域（北京 / 新加坡 / 弗吉尼亚）的 [API Key](../concepts/api-key.md) 不互通，请求时需使用对应地域的 Endpoint 与永久 Key。 - -## 服务关联角色（SLR） - -百炼在实现特定功能时，需通过服务关联角色（Service Linked Role, SLR）访问其他云服务（如 FC、OSS、ADB-PG、MNS、内容安全、SLS、CMS、DTS、CPFS 等）。当您首次在百炼中开通相关功能（如函数计算节点、OSS 数据导入、安全存储空间等）时，系统会**自动创建**对应的 SLR，无需手动创建。所有 SLR 可在 [RAM 控制台](https://ram.console.aliyun.com/) 的角色管理页面查看。 - -**主要 SLR 与用途**： - -| 服务关联角色 | 用途 | -| --- | --- | -| `AliyunServiceRoleForSFMAccessFC` | [工作流](../concepts/workflow.md)应用 / 流程编排访问函数计算（FC）资源 | -| `AliyunServiceRoleForSFMDataHubOSSImport` | 数据管理从 OSS 导入数据 | -| `AliyunServiceRoleForAccessOSS` | 安全存储空间访问 OSS | -| `AliyunServiceRoleForSFMAccessADB` | [知识库](../concepts/knowledge-base.md) / 安全存储空间访问 ADB-PG 实例 | -| `AliyunServiceRoleForSFMAccessingMNS` | 数据管理访问 MNS 队列中的 OSS 变更消息 | -| `AliyunServiceRoleForSFMTelemetry` | 用量监控与性能分析访问 OpenTelemetry 实例 | -| `AliyunServiceRoleForSFMAccessingCIP` | 百炼应用访问内容安全服务 | -| `AliyunServiceRoleForSFMAccessSLS` | 模型监控访问 SLS 资源 | -| `AliyunServiceRoleForSFMAccessCMS` | 模型监控访问 CMS 资源 | -| `AliyunServiceRoleForAccessCusOss` | 百炼平台托管操作用户 OSS 文件 | -| `AliyunServiceRoleForSFMConnectorAccessDTS` | 创建和管理 DTS 任务，从数据源接入数据 | -| `AliyunServiceRoleForSFMFineTuning` | [模型调优](../concepts/fine-tuning.md) / 数据管理访问 CPFS 与 OSS | - -每个 SLR 关联一个固定的系统策略（如 `AliyunServiceRolePolicyForSFMAccessFC`），策略中通过 RAM 条件（`ram:ServiceName`）限定只能由百炼服务使用，请勿修改或授予其他 RAM 身份。 - -**删除前注意事项**：删除 SLR 会导致对应功能不可用，须先清理依赖资源。例如删除 `AliyunServiceRoleForSFMAccessFC` 前须先删除所有已发布[工作流](../concepts/workflow.md)应用和流程中的函数计算节点并重新发布；删除 `AliyunServiceRoleForAccessOSS` 前须在安全存储空间中断开所有 OSS 连接；删除 `AliyunServiceRoleForSFMDataHubOSSImport` 前须确保没有进行中的 OSS 数据导入任务。具体删除步骤参见 [服务关联角色](https://help.aliyun.com/zh/ram/user-guide/service-linked-roles)。 - -## 知识库 SearchFilters - -在调用知识库 [Retrieve](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-retrieve) 接口时，若返回结果包含较多与 Query 无关的干扰信息（尤其适合结构化数据场景），可在请求体中传入 `searchFilters` 对语义检索结果做进一步过滤。 - -**效果对比**：未传入 `searchFilters` 时，Retrieve 可能返回多条低相关切片（如查询「张三」却返回李四、王五）；传入后可仅保留命中过滤条件的切片。 - -**语法**：`searchFilters` 是一个数组，每个元素是一个由 Key-Value 键值对组成的**子分组**。子分组之间默认采用 **AND** 语义且不可更改。 - -```json -{ - "searchFilters": [ - { "姓名": "张三", "性别": "男" }, - { "岗位": "技术员" } - ] -} -``` - -**支持的查询类型**： - -| 查询类型 | 适用字段类型 | 说明 | -| --- | --- | --- | -| 单值查询 | 数值（long/double）、字符串（string） | 字段等于某个值 | -| 多值查询 | 纯数值数组或纯字符串数组 | 字段命中数组中任一值；多值需用 `json.dumps` 序列化后传入 | -| 范围查询-等值 | 数值、字符串 | 支持 `eq`（等于）、`neq`（不等于）；一个字段不可配多个值 | -| 范围查询-区间 | 数值（long/double） | 支持 `gt`/`gte`/`lt`/`lte` | -| 模糊查询 | 字符串 | 支持 `like` 属性，`%` 匹配任意字符（含零个） | -| 标签（Tag）查询 | 仅文档搜索、音视频搜索类知识库 | `tags` 字段，多个标签之间为 OR 关系 | - -**前置条件**：子账号需获取 `AliyunBailianDataFullAccess` 策略并加入[业务空间](../concepts/workspace.md)（主账号可操作所有[业务空间](../concepts/workspace.md)），获取[业务空间](../concepts/workspace.md) ID，安装百炼 SDK（2023-12-29 版本）并配置 `ALIBABA_CLOUD_ACCESS_KEY_ID` / `ALIBABA_CLOUD_ACCESS_KEY_SECRET` 环境变量。 - -**调用示例（Python，单值查询）**： - -```python -retrieve_request = bailian_20231229_models.RetrieveRequest() -retrieve_request.query = '公司中叫张三的员工' -retrieve_request.index_id = '请传入实际的知识库ID' -retrieve_request.search_filters = [{"姓名": "张三"}] -resp = client.retrieve('请传入实际的业务空间ID', retrieve_request) -``` - -多值、范围、模糊、标签查询的写法类似，区别在于将字段的值替换为 `json.dumps` 序列化后的对象（如 `{"like": "技%员"}`、`{"gte": 20, "lte": 27}`、`["张三", "李四"]`）。完整 Python/Java 示例参见 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 - -## 限制与注意事项 - -- 临时 [API Key](../concepts/api-key.md) 无法手动删除，只能等 TTL 到期自动失效；各地域 [API Key](../concepts/api-key.md) 不互通。 -- 服务关联角色由百炼自动创建并绑定固定系统策略，不可修改、不可授予其他 RAM 身份；删除前必须先解除对应功能的依赖资源，否则相关功能将不可用。 -- SearchFilters 子分组之间为 AND 语义且不可更改；多值 / 范围 / 模糊 / 标签查询的值需通过 `json.dumps` 序列化为字符串后传入；标签查询仅支持文档搜索与音视频搜索类知识库。 -- 子账号只能操作已加入[业务空间](../concepts/workspace.md)中的知识库，主账号可操作所有[业务空间](../concepts/workspace.md)。 +- 临时 API Key **不可主动删除**，仅能等待过期自动失效；其权限完全继承自签发用的永久 API Key，务必确保后者权限最小化。 +- 所有服务关联角色均受 **RAM 服务关联角色约束**：删除前必须先解除其依赖（如删除函数计算节点、断开 OSS 连接、停止 MNS 订阅等），否则删除失败；`AliyunServiceRoleForSFMAccessingMNS` 明确禁止手动修改或删除 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 +- SearchFilters 仅作用于 **已成功索引且参与检索的字段**；若字段未在知识库创建时勾选“参与检索”，则无法被 `searchFilters` 过滤；多值查询需使用 `json.dumps(["val1", "val2"])` 格式传递字符串数组，而非原生数组。 +- > **注意**：文档 3 的 Python 示例中 `multi_query()` 方法将 `names` 数组 `json.dumps` 后赋值给字段，但实际 SDK（如 `alibabacloud_bailian20231229` v1.0.11+）已支持原生 list 传参，推荐直接使用 `{"姓名": ["张三", "李四"]}`，避免手动序列化引发格式错误。 ## 来源文档 @@ -114,23 +40,4 @@ resp = client.retrieve('请传入实际的业务空间ID', retrieve_request) - [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) - [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md index 9a40d452..ecb698f0 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md @@ -1,179 +1,103 @@ # omni realtime api -Qwen-Omni-Realtime API 是阿里云百炼平台提供的实时[多模态](../concepts/multimodal.md)交互接口，基于 WebSocket 协议实现低延迟的音视频对话。该 API 支持语音输入/输出、图像输入、语音活动检测（VAD）、工具调用（Function Calling）、联网搜索及声音复刻等功能，适用于智能客服、语音助手等实时对话场景。 +Qwen-Omni-Realtime API 是基于 WebSocket 的实时多模态交互接口，支持语音、文本、图像输入与文本、语音输出的端到端流式交互。它采用事件驱动模型，客户端通过发送结构化事件（如 `session.update`、`input_audio_buffer.append`）控制会话状态与数据流，服务端通过异步事件（如 `session.created`、`response.audio.delta`）实时反馈处理结果与生成内容。该 API 专为低延迟、高并发的语音助手、智能客服等场景设计。 -## 支持的模型 +## 支持的模型/功能 -| 模型系列 | 模型名称 | 特性 | -| --- | --- | --- | -| Qwen3.5-Omni-Realtime | qwen3.5-omni-plus-realtime、qwen3.5-omni-flash-realtime | 支持 semantic_vad、联网搜索、工具调用、idle_timeout_ms | -| Qwen3-Omni-Flash-Realtime | qwen3-omni-flash-realtime | 支持 smooth_output 参数 | -| Qwen-Omni-Turbo-Realtime | qwen-omni-turbo-realtime | 大部分生成参数不支持修改 | +- **核心模型系列**： + - `qwen3.5-omni-realtime`（支持 `semantic_vad`、联网搜索、工具调用） + - `qwen3.5-omni-plus-realtime` / `qwen3.5-omni-flash-realtime`（支持 `idle_timeout_ms`、`smooth_output`） + - `qwen3-omni-flash-realtime`（默认音色 `Cherry`，支持 `smooth_output`） + - `qwen-omni-turbo-realtime`（默认音色 `Chelsie`，**不支持修改多数采样参数**） -各模型的默认音色不同：Qwen3.5-Omni-Realtime 系列为 `Tina`，Qwen3-Omni-Flash-Realtime 为 `Cherry`，Qwen-Omni-Turbo-Realtime 为 `Chelsie`。 +- **多模态能力**： + - 输入：PCM 音频（16 kHz）、JPG/JPEG 图像（≤1080p，Base64 编码 ≤256 KB）、实时视频帧（通过 `append_video`） + - 输出：文本 + PCM 音频（24 kHz），可单独禁用任一模态（`modalities: ["text"]`） + - 实时转录：内置 `qwen3-asr-flash-realtime` 模型，不可替换，支持 `conversation.item.input_audio_transcription.delta` 增量预览 [原文标题](../../raw/model-api-reference/omni-realtime-api/server-events.md) -## 交互模式 +- **高级功能**： + - 语音活动检测（VAD）：`server_vad`（声学）或 `semantic_vad`（语义，仅 `qwen3.5-omni-realtime` 支持） + - 工具调用（Function Calling）：定义 `tools` 后模型自主触发，需客户端回传结果并调用 `response.create` [原文标题](../../raw/model-api-reference/omni-realtime-api/client-events.md) + - 联网搜索：`enable_search: true`（仅 `qwen3.5-omni-realtime` 系列），与 `tools` **互斥** + - 声音复刻：需先调用 `qwen-voice-enrollment` 创建音色，再在 `session.update` 中指定 `voice` 参数 [原文标题](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) -根据[实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)，API 支持两种交互模式： +> **注意**：文档 1 和文档 2 对 `voice` 默认值描述存在细微差异——文档 1 明确 `qwen3.5-omni-realtime` 默认为 `Tina`，而文档 2 的 `session.created` 示例中 `model` 字段为 `qwen3-omni-flash-realtime` 且 `voice` 为 `Cherry`。实际默认值严格按模型系列区分，以文档 1 和 SDK 文档（文档 3、4）为准。 -### VAD 模式（默认） +## 关键参数 -将 `session.turn_detection` 设为 `server_vad` 或 `semantic_vad`。服务端自动检测语音起止并触发模型响应，适用于持续音频流场景。支持语音打断。 +所有配置均通过 `session.update` 事件或 SDK 的 `update_session()` 方法设置： -### Manual 模式 +| 参数 | 类型 | 说明 | 限制 | +|------|------|------|------| +| `modalities` | `["text"]` 或 `["text","audio"]` | 输出模态组合 | 不支持 `["audio"]` 单独输出 | +| `voice` | `string` | 音色名 | 必须是[音色列表](https://help.aliyun.com/zh/model-studio/realtime#f9c68d860a3rs)中的有效值；声音复刻生成的音色亦可传入 | +| `input_audio_format` / `output_audio_format` | `"pcm"` | 音频编解码格式 | 固定为 PCM；输入采样率 16 kHz，输出 24 kHz，**不可自定义** | +| `instructions` | `string` | 系统角色指令 | 影响模型行为，建议明确限定职责与边界 | +| `turn_detection.type` | `"server_vad"` 或 `"semantic_vad"` | VAD 类型 | `semantic_vad` 仅 `qwen3.5-omni-realtime` 支持；设为 `null` 则启用 Manual 模式 | +| `turn_detection.threshold` | `float [-1.0, 1.0]` | VAD 灵敏度 | 默认 `0.5`；嘈杂环境建议调高（如 `0.7`） | +| `turn_detection.silence_duration_ms` | `int [200, 6000]` | 静音触发阈值 | 默认 `800` ms；值越小响应越快但误触发风险越高 | +| `idle_timeout_ms` | `int [5000, 30000]` | 静默超时（主动引导） | **仅 `qwen3.5-omni-plus-realtime`/`flash-realtime` + `server_vad` 有效** | +| `enable_search` | `boolean` | 启用联网搜索 | 仅 `qwen3.5-omni-realtime` 系列有效；与 `tools` 冲突，不可共存 | +| `tools` | `array` | 工具函数定义 | 每个工具含 `name`、`description`、`parameters`（含 `properties` 和 `required`）；`parameters.type` 固定为 `"object"` | -将 `session.turn_detection` 设为 `null`。客户端通过 `input_audio_buffer.commit` + `response.create` 手动控制对话节奏，适用于按下即说场景。 +**采样参数（模型级）**： +- `temperature` / `top_p`：二选一控制多样性（`qwen-omni-turbo` 系列**不可修改**） +- `top_k`：候选集大小（`qwen-omni-turbo` 系列**不可修改**） +- `max_tokens`：响应截断长度（不影响生成过程） +- `repetition_penalty` / `presence_penalty` / `seed`：重复控制与确定性（`qwen-omni-turbo` 系列**不可修改**） -## 连接地址 +## 使用方式 -``` -wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime # 北京地域 -wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime # 新加坡地域 -``` +1. **建立连接**： + - WebSocket URL 格式：`wss://{WorkspaceId}.{region}.maas.aliyuncs.com/api-ws/v1/realtime`（推荐业务空间专属域名，替代旧 `dashscope.aliyuncs.com`） + - 连接后立即收到 `session.created` 事件，含默认配置 [原文标题](../../raw/model-api-reference/omni-realtime-api/server-events.md) -将 `{WorkspaceId}` 替换为[业务空间](../concepts/workspace.md) ID。建议使用[业务空间](../concepts/workspace.md)专属域名以获得更好的性能和稳定性。 +2. **配置会话**： + - 发送 `session.update` 事件（或调用 SDK `update_session()`），传入所需参数。成功后服务端返回 `session.updated`。 -## 客户端事件 +3. **输入数据**： + - **音频**：持续发送 `input_audio_buffer.append`（Base64 PCM），VAD 模式下由服务端自动提交；Manual 模式下需显式发送 `input_audio_buffer.commit`。 + - **图像**：发送 `input_image_buffer.append`（Base64 JPG/JPEG），**必须在首次 `input_audio_buffer.append` 之后**，且与音频缓冲区一同提交。 -详细参数说明参见[客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 +4. **触发响应**： + - **VAD 模式**：语音结束自动触发，无需客户端操作。 + - **Manual 模式**：发送 `input_audio_buffer.commit` 后，再发送 `response.create`。 + - **工具调用后**：客户端执行工具并发送 `conversation.item.create`，再发 `response.create`（Manual 模式）或等待服务端自动触发（VAD 模式）。 -| 事件 | 用途 | -| --- | --- | -| `session.update` | 更新会话配置（模态、音色、VAD、工具等） | -| `input_audio_buffer.append` | 追加音频数据（Base64 编码） | -| `input_audio_buffer.commit` | 提交音频缓冲区（Manual 模式必需） | -| `input_audio_buffer.clear` | 清空音频缓冲区 | -| `input_image_buffer.append` | 追加图像数据（JPG/JPEG，Base64 编码） | -| `response.create` | 触发模型生成响应 | -| `response.cancel` | 取消正在进行的响应 | -| `conversation.item.create` | 回传工具调用结果 | +5. **处理输出**： + - 文本流：`response.text.delta` → `response.text.done` + - 音频流：`response.audio.delta` → `response.audio.done` + - 转录流：`conversation.item.input_audio_transcription.delta`（拼接 `text` + `stash` 获取实时预览）→ `conversation.item.input_audio_transcription.completed`（最终结果） -## 服务端事件 +## 限制和注意事项 -详细参数说明参见[服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md)。 +- **协议与兼容性**： + - 必须使用 WebSocket，HTTP REST 不支持。 + - `qwen-omni-turbo-realtime` 系列**禁止修改** `temperature`、`top_p`、`top_k`、`max_tokens`、`repetition_penalty`、`presence_penalty`、`seed`（文档 1、3、4 均明确标注）。 -| 事件 | 含义 | -| --- | --- | -| `session.created` | 连接建立，返回默认配置 | -| `session.updated` | 会话配置更新成功 | -| `error` | 错误信息 | -| `input_audio_buffer.speech_started` | VAD 检测到语音开始 | -| `input_audio_buffer.speech_stopped` | VAD 检测到语音结束 | -| `input_audio_buffer.committed` | 音频缓冲区已提交 | -| `response.audio.delta` | 增量音频输出 | -| `response.audio_transcript.delta` | 增量文本转录 | -| `response.done` | 响应完成 | -| `response.function_call_arguments.done` | 工具调用参数完成 | -| `conversation.item.input_audio_transcription.delta` | 实时语音识别中间结果 | +- **资源约束**： + - 单张图片 Base64 编码后 ≤ 256 KB；音频缓冲区最大 15 MiB（Manual 模式）；声音复刻音频 ≤ 60 秒且 < 10 MB。 + - `idle_timeout_ms` 仅对特定模型+VAD 组合生效，其他组合设置将被忽略。 -## 关键会话参数 +- **关键互斥规则**： + - `tools` 与 `enable_search` **不可同时启用**，否则 `session.update` 将返回 `invalid_request_error`。 + - `smooth_output` **仅对 `qwen3-omni-flash-realtime` 有效**，其他模型设置将被忽略（文档 1、3、4 一致）。 -通过 `session.update` 事件配置： +- **错误处理**： + - 所有错误均以 `error` 事件返回，含 `type`、`code`、`message`、`param` 字段，例如 `invalid_value` 错误会明确指出违规参数（如 `session.modalities`） [原文标题](../../raw/model-api-reference/omni-realtime-api/server-events.md)。 + - `input_audio_buffer.commit` 在缓冲区为空时返回错误；`response.cancel` 在无进行中响应时返回错误。 -| 参数 | 说明 | 默认值 | -| --- | --- | --- | -| `modalities` | 输出模态：`["text"]` 或 `["text","audio"]` | `["text","audio"]` | -| `voice` | 音色名称 | 因模型而异 | -| `input_audio_format` | 输入音频格式，仅支持 `pcm`（16kHz） | `pcm` | -| `output_audio_format` | 输出音频格式，仅支持 `pcm`（24kHz） | `pcm` | -| `instructions` | 系统消息 | - | -| `turn_detection.type` | VAD 类型：`server_vad` / `semantic_vad` | `server_vad` | -| `turn_detection.threshold` | VAD 灵敏度，范围 [-1.0, 1.0] | 0.5 | -| `turn_detection.silence_duration_ms` | 静音触发时间（ms），范围 [200, 6000] | 800 | -| `turn_detection.idle_timeout_ms` | 静默超时（ms），范围 [5000, 30000]，仅 qwen3.5 系列 | - | -| `enable_search` | 联网搜索，仅 Qwen3.5-Omni-Realtime | `false` | -| `tools` | 工具定义列表，仅 Qwen3.5-Omni-Realtime | `[]` | -| `smooth_output` | 口语化风格，仅 Qwen3-Omni-Flash-Realtime | `true` | - -> **注意**：`tools` 和 `enable_search` 不兼容，不可同时开启。 - -### 生成参数 - -| 参数 | Qwen3.5-Omni-Realtime | Qwen3-Omni-Flash-Realtime | Qwen-Omni-Turbo-Realtime | -| --- | --- | --- | --- | -| `temperature` | 0.7 | 0.9 | 1.0（不可改） | -| `top_p` | 0.8 | 1.0 | 0.01（不可改） | -| `top_k` | 20 | 50 | 20（不可改） | -| `repetition_penalty` | 1.0 | 1.05 | 1.05（不可改） | -| `presence_penalty` | 1.5 | 0.0 | 0.0（不可改） | - -> **注意**：`qwen-omni-turbo` 系列模型的生成参数不支持修改。 - -## SDK 使用 - -### Python SDK - -需要 [DashScope SDK](../concepts/dashscope-sdk.md) >= 1.25.17。核心类为 `OmniRealtimeConversation`，通过 `from dashscope.audio.qwen_omni import OmniRealtimeConversation` 引入。详见 [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md)。 - -```python -from dashscope.audio.qwen_omni import MultiModality, OmniRealtimeCallback, OmniRealtimeConversation - -conv = OmniRealtimeConversation(model="qwen3.5-omni-plus-realtime", callback=callback, url=url) -conv.connect() -conv.update_session( - output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], - voice="Tina", - enable_turn_detection=True -) -conv.append_audio(audio_base64) -conv.close() -``` - -### Java SDK - -需要 DashScope Java SDK >= 2.22.15。核心类为 `OmniRealtimeConversation`，通过 `OmniRealtimeParam` 和 `OmniRealtimeConfig` 配置参数。详见 [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md)。 - -```java -OmniRealtimeParam param = OmniRealtimeParam.builder() - .model("qwen3.5-omni-plus-realtime") - .url(url) - .build(); -OmniRealtimeConversation conversation = new OmniRealtimeConversation(param, callback); -conversation.connect(); -conversation.updateSession(OmniRealtimeConfig.builder() - .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT)) - .voice("Tina") - .enableTurnDetection(true) - .build()); -``` - -> **注意**：Java SDK 中 `instructions`、`smooth_output`、`enable_search`、`search_options`、`tools` 及生成参数（temperature/top_p/top_k 等）需通过 `OmniRealtimeConfig` 的 `parameters` 方法设置。 - -## 工具调用（Function Calling） - -仅 Qwen3.5-Omni-Realtime 模型支持。流程如下： - -1. 通过 `session.update` 配置 `tools` 列表 -2. 服务端识别到需要调用工具时，通过 `response.function_call_arguments.done` 返回函数名和参数 -3. 客户端执行工具函数，通过 `conversation.item.create` 回传结果（`type: "function_call_output"`） -4. VAD 模式下服务端自动生成响应；Manual 模式下需额外发送 `response.create` - -## 声音复刻 - -通过 `qwen-voice-enrollment` 模型创建自定义音色，然后在实时对话中使用。音频要求：WAV/MP3/M4A 格式，10-20 秒，采样率 >= 24kHz，单声道，文件 < 10MB。创建音色时指定的 `target_model` 必须与后续对话使用的模型一致。 - -支持的驱动模型：qwen3.5-omni-plus-realtime、qwen3.5-omni-flash-realtime。 - -## 输入限制 - -- 音频输入：16kHz 采样率 PCM，音频缓冲区最大 15MiB -- 图像输入：JPG/JPEG 格式，建议 480p-720p（最高 1080p），Base64 编码后不超过 256KB，建议 1 帧/秒 -- 图像需在至少一次 `input_audio_buffer.append` 之后发送，通过 `input_audio_buffer.commit` 与音频一起提交 +- **最佳实践**： + - VAD 模式推荐使用耳机避免回声打断；Manual 模式适用于“按住说话”类 UI。 + - 声音复刻音色必须与 Omni 模型版本严格匹配（如 `qwen3.5-omni-plus-realtime` 复刻的音色只能用于同系列模型）。 ## 来源文档 - [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) - [服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md) - [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) -- [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) - [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) +- [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) - [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md index b4840377..2c6e6a12 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md @@ -1,80 +1,68 @@ # preparations -本页汇总在阿里云百炼平台调用模型 API 前的准备工作，涵盖获取鉴权凭证（API Key）、安装官方或兼容 SDK、使用百炼 CLI 快速集成，以及常见错误码的排查思路。面向开发者，帮助你从零完成环境搭建并稳定发起第一次调用。 +在调用阿里云百炼平台的模型与服务前，开发者需完成 SDK 安装、API Key 获取与配置、CLI 工具部署等基础准备。这些步骤共同构成安全、稳定、可维护的调用链路，直接影响后续模型调用的可用性与合规性。本文档整合关键操作路径与约束条件，面向工程实践提供结构化指引。 -## 获取并配置 API Key +## 支持的模型/功能 -调用模型或应用前，需先获取 API Key 作为鉴权凭证。需使用主账号，或具备 `管理员` / `API-Key` 页面权限的子账号，在[阿里云百炼控制台](https://bailian.console.aliyun.com/)对应地域的 **API Key** 页面创建。详见 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md)。 +百炼平台支持全模态模型调用，包括文本生成（如 `qwen3.7-max`）、图像生成（`qwen-image-2.0`）、视频生成（`happyhorse-1.0-t2v`）、语音合成（`cosyvoice-v3-flash`）、语音识别（Paraformer）、向量嵌入（`text-embedding-v3`）、排序（`text-rerank-v3`）及多模态理解（`qwen3-vl-plus`、`qwen3.5-omni-plus`）。所有模型均通过统一 API 接口暴露，但协议兼容性存在差异：[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)适用于多数第三方 SDK，而 Anthropic 兼容接口仅限 Messages 协议（如 `anthropic.messages`），具体请以各模型文档为准。[安装SDK](../../raw/model-api-reference/preparations/install-sdk.md) 文档列出了各语言 SDK 对应的模型支持范围与调用示例。 -创建时的关键选项： +## 关键参数 -- **归属业务空间**：决定该 Key 的调用权限。同一空间内的 Key 权限相同，无需为不同模态（文生文、文生图、语音等）分别创建。默认业务空间的 Key 可调用所有标准模型及默认空间内的应用；子业务空间的 Key 只能调用已授权的模型及本空间应用。 -- **权限**：可选 **全部**（调用所有模型与应用），或 **自定义**（配置 IP 白名单最多 20 个 IPv4/IPv6 地址或网段，以及可访问的模型/应用范围）。 +调用时需关注以下核心参数及其取值约束： -> **注意**：百炼已对按量付费 API Key 做安全升级（美国（弗吉尼亚）地域除外）。升级后新建的 Key 以 `sk-ws` 开头，且**仅在创建时展示一次明文**，关闭弹窗后无法再次查看，务必立即复制保存；升级前 `sk-` 开头的旧 Key 仍可正常使用。此外，Token Plan / Coding Plan 使用以 `sk-sp-` 开头的专属 Key，不同于本文的按量付费 Key。 +- **`model`**：必须为百炼控制台模型市场中已开通的**标准模型 ID**（如 `qwen3.7-max`），不可混用 Hugging Face 格式（如 `Qwen/Qwen3-7B-Instruct`）；未开通模型将返回 `Model not exist` 或 `The product is not activated` 错误。 +- **`temperature`**：取值范围 `[0.0, 2.0)`，超出将触发 `400-InvalidParameter`。 +- **`top_p`**：取值范围 `(0.0, 1.0]`。 +- **`max_tokens`**：必须为 `[1, 模型最大输出 Token 数]` 内的整数，上限见各模型文档。 +- **`n`**（生成数量）：图像/视频类接口默认为 `1`，上限为 `6`；文本类接口上限为 `4`。 +- **`seed`**：DashScope 协议下有效范围为 `[0, 9223372036854775807]`。 +- **`enable_thinking`**：思考模式仅支持[流式输出](../concepts/streaming-output.md)（`stream=true`），且与 `response_format="json_object"` 互斥；部分模型（如 `qwen3-235b-a22b-thinking-2507`）强制要求设为 `true`。 +- **`messages` / `prompt`**：二者必须且仅存在其一；纯文本模型不接受 `content` 为数组或含 `image_url` 的多模态消息，否则报错 `Unexpected item type in content`。 -推荐将 API Key 配置到环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄漏。各系统配置方式（`~/.bashrc`、`~/.zshrc`、`~/.bash_profile`、Windows 系统属性 / `setx` / PowerShell）参见原文。调用时除 API Key 外，还需指定**服务端点** `base_url`（即创建弹窗中的 API Host），且 OpenAI 兼容协议与 Anthropic 兼容协议的 `base_url` 不同、随地域变化，请以对应接口文档为准。 +> **注意**：文档 3 中 CLI 的 `bl text chat` 默认模型为 `qwen3.7-max`，而文档 4 的错误码示例中多次出现 `qwen3-235b-a22b-thinking-2507` 等长 ID 模型。实际使用时须以 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) 后在控制台模型市场确认的**已开通模型列表**为准，避免因模型名过时或未开通导致 `Model not exist`。 -除控制台外，百炼还提供 OpenAPI（`CreateApiKey` / `GetApiKey` / `ListApiKeys` / `UpdateApiKey` / `DeleteApiKey` / `EnableApiKey` / `DisableApiKey` / `ResetApiKey`）以编程方式管理 Key，调用需使用阿里云账号 AccessKey 签名认证并具备相应 RAM 权限。 +## 使用方式 -## 安装 SDK - -百炼同时支持官方 **DashScope SDK**（Python、Java）与通过 **[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**调用的多语言 SDK。详见 [安装SDK](../../raw/model-api-reference/preparations/install-sdk.md)。 - -- **Python**（需 `python >= 3.8`）：`pip install -U openai` 或 `pip install -U dashscope` -- **Java**：DashScope 用 `com.alibaba:dashscope-sdk-java`；OpenAI 用 `com.openai:openai-java`（需 Java 8+，推荐 `3.5.0`），均通过 Maven / Gradle 引入。 -- **Node.js**：`npm install --save openai`（或 `yarn add openai`）；安装失败可配置镜像源 `npm config set registry https://registry.npmmirror.com/`。 -- **Go**（需 `Go 1.22+`）：`go get 'github.com/openai/openai-go/v3'`；超时可设 `go env -w GOPROXY=https://mirrors.aliyun.com/goproxy/,direct`。 - -安装后即可调用文本生成、图像生成、视频生成、语音合成/识别、向量、排序等模型。 - -## 使用百炼 CLI - -百炼 CLI（npm 包 `bailian-cli`，命令 `bl` / `bailian`）是面向 AI Agent 的命令行工具，可将平台能力集成到各类 AI 工具中。安装前置要求 **Node.js ≥ 22.12.0**，且**仅支持 npm 安装**（勿用 pnpm / yarn 安装该包）。详见 [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md)。 +### 1. SDK 集成 +推荐使用 DashScope SDK（官方维护）或 OpenAI 兼容 SDK（跨平台适配）。Python 开发者可任选： +```bash +pip install -U dashscope # DashScope 原生 SDK +pip install -U openai # OpenAI 兼容 SDK（需配置 base_url） +``` +Java、Node.js、Go 等语言同理，详见 [安装SDK](../../raw/model-api-reference/preparations/install-sdk.md)。调用时需显式指定 `base_url`（即 API Host），不同地域与协议的端点不同，不可复用。 +### 2. CLI 快速验证 +百炼 CLI（`bailian-cli`）适用于本地调试与自动化脚本： ```bash -# 1. 安装 CLI npm install -g bailian-cli -# 2. 安装 Skills（注册能力描述文件到各 Agent） -npx skills add modelstudioai/cli --all -g -# 3. 验证 -bl --version +bl auth login --console # 浏览器 OAuth 登录（推荐） +# 或 +bl auth login --api-key sk-xxx # 手动输入 API Key +bl text chat --message "ping" --non-interactive ``` +CLI 要求 Node.js ≥ 22.12.0，且**仅支持 npm 全局安装**（禁用 pnpm/yarn）。认证后可通过 `bl config set` 持久化模型、输出目录等参数。 -**认证方式**（可组合使用，互不覆盖）： - -| 方式 | 命令 | 适用场景 | -| --- | --- | --- | -| 控制台登录（推荐） | `bl auth login --console` | 模型调用 + 应用管理（浏览器 OAuth） | -| API Key | `bl auth login --api-key sk-xxx` | 模型调用；会先校验 Key 有效性 | -| 环境变量 | 配置 API Key 环境变量 | CI/CD、无界面环境 | -| 配置文件 | `bl config set --key api_key --value sk-xxx` | 持久化，**不校验** Key 有效性 | -| 临时传入 | `bl text chat --api-key sk-xxx ...` | 单次调用，不落盘 | +### 3. 环境变量安全配置 +**强烈建议**将 API Key 存入环境变量而非硬编码： +- Linux/macOS：写入 `~/.bashrc` 或 `~/.zshrc`，执行 `source` 生效 +- Windows：通过系统属性或 PowerShell 设置用户级变量 `DASHSCOPE_API_KEY` +详情见 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) 中的环境变量配置章节。 -常用全局参数：`--region `（默认 cn）、`--base-url`、`--output `、`--non-interactive`（Agent/CI）、`--dry-run`、`--concurrent ` 等。子命令覆盖文本对话（`bl text chat`）、全模态（`bl omni`）、图像（`bl image generate/edit`）、视频（`bl video generate/edit/ref`）、视觉理解（`bl vision describe`）、语音合成（`bl speech synthesize`）等。 +## 限制和注意事项 -> **注意**：CLI 文档中示例默认模型（如 `qwen3.7-max`、`qwen3.5-omni-plus`、`qwen-image-2.0`、`happyhorse-1.0-t2v` 等）为工具内置默认值，可能随版本变化；实际可用模型请以模型列表 / 控制台为准。安全约束上，禁止将真实 API Key 写入仓库、日志、Skill 或聊天记录的可公开部分。 +- **API Key 安全**：新创建的按量付费 API Key 以 `sk-ws` 开头，创建后**仅展示一次明文**，关闭弹窗即不可恢复；旧 `sk-` 密钥仍可用，但建议升级。切勿在代码、日志、Git 仓库中明文存储。 +- **地域隔离**：API Key 与服务端点（`base_url`）严格绑定地域（如华北2、新加坡、美国弗吉尼亚）。跨地域调用需单独创建对应地域的 Key 并配置 `--region` 参数。 +- **权限模型**：API Key 权限由其**归属业务空间**决定，同一空间内所有 Key 权限一致。子业务空间下的 Key 仅能调用该空间已授权的模型，需提前在控制台完成模型授权。 +- **文件限制**：Qwen-Long 等长文本模型仅支持 TXT/DOCX/PDF/EPUB/MOBI/MD 格式，单文件 ≤ 150 MB、≤ 1500 页；图片类文件需先用 Qwen-VL 提取文本。 +- **错误处理**：常见错误如 `Arrearage`（欠费）、`InvalidParameter`（参数越界）、`Model not exist`（未开通）均有明确修复路径，建议集成 [阿里云 AI 助理](https://www.aliyun.com/ai-assistant/) 实时解析错误响应。[错误码](../../raw/model-api-reference/preparations/error-code.md) 文档覆盖全部 HTTP 4xx/5xx 场景及解决方案。 -## 常见错误码与排查 - -调用过程中的报错多为 **400-InvalidParameter** 类的参数问题，可对照错误信息定位。完整清单见 [错误码](../../raw/model-api-reference/preparations/error-code.md)，以下为高频场景： - -- **思考模式相关**：思考模式模型需 `enable_thinking=true` 时配合[流式输出](../concepts/streaming.md)，并设 `incremental_output=true`、`result_format="message"`；部分模型（如 `qwen3-235b-a22b-thinking-2507`）不允许将 `enable_thinking` 设为 `false`。 -- **参数取值范围**：`temperature` ∈ [0.0, 2.0)、`top_p` ∈ (0.0, 1.0]、`top_k` ≥ 0、`presence_penalty` ∈ [-2.0, 2.0]、`n` ∈ [1, 4]；`max_tokens` 与输入长度上限以模型列表为准。 -- **模型不存在（Model not exist）**：核对 `model` 名称大小写与空格，勿混用开源社区名与百炼模型 ID（用 `qwen3-235b-a22b-instruct-2507` 而非 `Qwen/Qwen3-235B-A22B-Instruct-2507`）。 -- **content 类型错误**：纯文本模型的 `content` 必须为字符串，不能传数组或图片等多模态元素；需要图片输入请改用 Qwen-VL / Qwen3-VL 等多模态模型。 -- **结构化输出**：使用 `response_format` 的 `json_object` 时，提示词须包含 `json` 关键词，且不能同时开启思考模式。 -- **文件类（Qwen-Long）**：仅支持纯文本格式（TXT/DOCX/PDF/EPUB/MOBI/MD），单文件 < 150 MB、< 15000 页，file-id 数量 < 100。 -- **账号状态（Arrearage）**：账号欠费会导致访问被拒绝，需在费用与成本页面充值后等待系统更新。 - -排障时可借助[阿里云 AI 助理](https://www.aliyun.com/ai-assistant/)，直接粘贴报错信息即可获得原因与解决方案。 +> **注意**：文档 3 中 CLI 的 `bl image generate --n 6` 允许单次生成 6 张图，但文档 4 的错误码说明中 `n` 参数上限为 `4` —— 此矛盾源于**接口协议差异**：CLI 封装层对图像类接口做了特殊处理（非标准 OpenAI 协议），而错误码文档描述的是通用文本生成接口约束。开发者应以具体接口文档（如 [图像生成 API](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference)）为准。 ## 来源文档 -- [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) - [安装SDK](../../raw/model-api-reference/preparations/install-sdk.md) -- [错误码](../../raw/model-api-reference/preparations/error-code.md) +- [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) - [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md) - +- [错误码](../../raw/model-api-reference/preparations/error-code.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md index f4dc1072..be27d4a6 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md @@ -1,37 +1,57 @@ # qwen api reference -百炼平台为文本生成模型提供了多种调用接口，开发者可根据迁移成本、功能完整度和生态兼容性选择合适的入口。当前共有四类接口：OpenAI 兼容 Chat Completions、OpenAI 兼容 Responses、Anthropic 兼容 Messages 以及百炼原生的 DashScope 接口。详见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 - -## 支持的接口 - -百炼针对不同的接入场景提供了以下四种接口，功能定位各有侧重： - -- **OpenAI 兼容 Chat Completions**：与 OpenAI 客户端库直接兼容，迁移现有应用或接入第三方工具的成本最低。适合已经基于 OpenAI SDK 构建的应用平滑迁移。 -- **OpenAI 兼容 Responses**：内置联网搜索、代码解释器和网页内容提取工具，并自动管理对话历史，无需手动维护上下文。 -- **Anthropic 兼容 Messages**：兼容 Anthropic Messages API，支持思考（thinking）和工具调用（tool use）。适合基于 Anthropic 生态构建的应用接入。 -- **DashScope**：百炼原生接口，提供最完整的功能集和参数支持，是需要使用平台全部能力时的首选。 - -以上接口的完整清单与说明参见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 - -## 如何选择 - -- 追求**最低迁移成本**、已有 OpenAI 应用：选择 OpenAI 兼容 Chat Completions。 -- 需要**内置工具（联网搜索/代码解释器/网页提取）与自动对话管理**：选择 OpenAI 兼容 Responses。 -- 处于 **Anthropic 生态**、需要思考与工具调用：选择 Anthropic 兼容 Messages。 -- 需要**最完整的功能与参数**、使用平台全部能力：选择 DashScope 原生接口。 - -## 使用方式与注意事项 - -- [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)可直接复用官方 OpenAI 客户端库，仅需替换 base URL 和 API Key，改动量小。 -- 若依赖联网搜索、代码解释器等内置工具，需使用 Responses 接口，而非普通的 Chat Completions。 -- 不同接口在参数集合和功能覆盖上存在差异：DashScope 参数最全，OpenAI/Anthropic 兼容接口以对应生态的字段约定为准，跨接口迁移时需核对参数映射。 - -> **注意**：本页仅为文本生成模型各接口的入口索引，具体的请求参数、字段格式与调用示例请查阅对应接口的专属文档；随着平台迭代，接口能力可能变化，请以 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) 为准。 +Qwen 系列大模型通过百炼平台提供多种 API 接入方式，支持文本生成、工具调用、多轮对话等核心能力。开发者可根据技术栈兼容性、功能需求和运维复杂度选择合适接口。所有接口均需通过阿里云认证（AccessKey 或 STS [Token](../concepts/token.md)）调用。 + +## 支持的模型与功能 + +当前 Qwen 系列支持以下主流接入协议： + +- **OpenAI 兼容 Chat Completions**：适用于已有 OpenAI 客户端的应用迁移，支持 `qwen-max`、`qwen-plus`、`qwen-turbo` 等模型，但不支持原生工具调用（需自行封装）。详见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 +- **OpenAI 兼容-Responses**：内置联网搜索、代码解释器、网页内容提取三类工具，自动维护对话历史，适合快速构建智能助手。该接口对 `messages` 格式有特定约束，详见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 +- **Anthropic 兼容 Messages**：支持 `tool_use`、`thinking` 等结构化输出，适用于需要可控推理链路的场景；注意其 `max_tokens` 语义与 OpenAI 不同（指输出 token 上限，不含输入）。详见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 +- **DashScope 原生接口**：提供最全参数控制（如 `incremental_output`、`enable_search`、`tools` schema 注册），支持流式响应、[函数调用](../concepts/function-calling.md)、长上下文（最高 128K tokens）及私有模型部署。推荐新项目优先选用。 + +> **注意**：`qwen-vl`（多模态）和 `qwen-audio` 模型**不支持** [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，仅可通过 DashScope 原生接口调用，相关限制请参阅最新 [DashScope 文档](https://help.aliyun.com/zh/dashscope/developer-reference/quick-start)。 + +## 关键参数 + +| 参数名 | 类型 | 必填 | 说明 | +|--------|------|------|------| +| `model` | string | 是 | 模型标识符，如 `qwen-max`、`qwen-plus`、`qwen-turbo`；不同接口对取值范围要求不同（例如 Anthropic 接口仅支持 `qwen-max`） | +| `messages` | array | 是 | 对话消息列表，格式为 `[{"role": "user", "content": "..."}, ...]`；[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)中 `content` 可为字符串或数组（含 text/image_url），DashScope 接口支持更丰富的 content 结构 | +| `temperature` | number | 否 | 采样温度，默认 `0.8`；取值范围 `0.0–1.0`，`0` 表示确定性输出 | +| `top_p` | number | 否 | 核采样阈值，默认 `0.95`；与 `temperature` 互斥使用效果更佳 | +| `stream` | boolean | 否 | 是否启用流式响应；[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)默认 `false`，DashScope 默认 `true`（若未显式设置） | + +## 使用方式 + +1. **认证**：使用阿里云 AccessKey ID/Secret 或 STS [Token](../concepts/token.md)，通过 `Authorization: Bearer ` 或 `X-DashScope-Signature` 头传递； +2. **Endpoint**： + - OpenAI 兼容：`https://dashscope.aliyuncs.com/v1/chat/completions` + - Anthropic 兼容：`https://dashscope.aliyuncs.com/v1/messages` + - DashScope 原生：`https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation`； +3. **示例请求（DashScope）**： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "model": "qwen-max", + "input": {"messages": [{"role":"user","content":"你好"}]}, + "parameters": {"temperature": 0.5} + }' + ``` + +## 限制和注意事项 + +- 单次请求最大 `input` 长度受模型 context window 限制（`qwen-turbo`: 8K, `qwen-plus`: 32K, `qwen-max`: 128K），超出将返回 `400 Bad Request`； +- OpenAI 兼容接口**不支持** `system` 角色消息（会被忽略），需改用 `user` + 提示词前置方式模拟； +- 所有接口均按 `input_tokens + output_tokens` 计费，`output_tokens` 包含工具调用返回内容； +- 流式响应中，OpenAI 兼容接口返回 `delta.content` 字段，DashScope 返回 `output.text` 字段，解析逻辑需区分； +- 调用失败时优先检查 `X-DashScope-Request-ID` 响应头，便于问题定位。 ## 来源文档 - [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md b/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md new file mode 100644 index 00000000..61eeba2d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md @@ -0,0 +1,87 @@ +# realtime api user guide + +Realtime API 是一套面向低延迟、弱网对抗和多模态实时交互场景的协议化接入方案，支持 WebSocket、WebRTC 和 AOQ 三种传输协议，开发者可根据终端类型、部署环境与业务需求选择最适配的接入方式。所有协议均基于统一的事件驱动模型，通过 `session.update` 配置会话、`input_audio_buffer.append` 等事件流式输入数据，并接收 `response.text.delta`、`response.audio.delta` 等增量响应。 + +## 支持的模型/功能 + +Realtime API 当前支持以下核心模型与应用类型，不同协议的支持能力存在差异： + +| 模型/应用类型 | AOQ | WebRTC | WebSocket | +|---------------|-----|--------|-----------| +| 实时全模态（qwen3.5-omni-plus-realtime、qwen3.5-omni-flash-realtime） | ✅ | ✅ | ✅ | +| 实时语音翻译（qwen3.5-livetranslate-flash-realtime） | ✅ | ✅ | ✅ | +| 多模态开发套件（multimodal-dialog） | ❌ | ✅ | ✅ | +| 实时语音识别（Fun-ASR系列） | ❌ | ❌ | ✅ | +| 实时语音合成（CosyVoice系列） | ❌ | ❌ | ✅ | +| 实时语音对话（qwen-audio-3.0-realtime-plus/flash） | ❌ | ❌ | ✅ | + +> **注意**：文档 [通过WebRTC使用多模态交互套件实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) 明确指出 `multimodal-dialog` 仅支持 WebRTC 和 WebSocket；而 [Realtime API简介](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md) 的表格中亦标注 AOQ 不支持该套件。二者一致，无矛盾。 + +AOQ 与 WebRTC 均内置回声消除（AEC）和降噪能力，WebSocket 方案需客户端自行处理；AOQ 和 WebRTC 支持音视频+文本混合传输，WebSocket 仅支持文本/音频/图像分通道传输，不支持原生多模态融合。 + +## 关键参数 + +### 协议级通用参数 +- `Authorization: Bearer `：建连阶段必需的鉴权头，**严禁硬编码于客户端**，生产环境应由业务 AppServer 代理请求或使用 [Token](../concepts/token.md) 机制（AOQ 必须）。 +- `model`：URL 查询参数，指定目标模型或应用，如 `qwen3.5-omni-plus-realtime` 或 `multimodal-dialog`。 +- `workspace_id`：百炼工作空间唯一标识，用于路由至对应服务实例，格式为 `llm-xxxxxxxxxx`。 + +### 会话配置参数（`session.update` 事件） +- `modalities`: `["text", "audio"]` 等数组，声明期望输出模态。 +- `voice`: 输出音频音色名（如 `"Ethan"`），仅当 `audio` 在 `modalities` 中时生效。 +- `input_audio_format` / `output_audio_format`: 当前仅支持 `"pcm"`。 +- `turn_detection`: VAD 配置对象，`type` 可选 `"server_vad"` 或 `"semantic_vad"`（推荐用于 qwen3.5-omni 系列）；`silence_duration_ms` 控制静音阈值，默认 800ms。 +- `instructions`: 系统角色提示词，影响模型行为。 + +### 协议特有参数 +- **WebRTC**: SDP 交换需 `Content-Type: application/sdp`；服务端 endpoint 格式为 `{workspace_id}.{region}.maas.aliyuncs.com`。 +- **AOQ**: 必须使用 `x-dashscope-rtc-transport: moq` 请求头；建连凭证含 `aoqTokenForClient`、`sid`、`clientRelayEndpoints` 等字段，详见 [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md)。 +- **WebSocket**: 连接 URL 为 `wss://dashscope.aliyuncs.com/...`，握手阶段携带 `Authorization` 头。 + +## 使用方式 + +### 协议选择指南 +- **WebSocket**: 适用于服务端集成、快速原型验证、浏览器外环境（如 Node.js CLI 工具）。接入成本最低，但弱网对抗与 AI 场景适配性较弱。 +- **WebRTC**: 适用于浏览器端实时互动（如网页版智能客服），依赖原生浏览器能力，需处理 CORS 限制（SDP 交换需后端代理）。 +- **AOQ**: 适用于 Android/iOS/HarmonyOS 原生 App，对延迟、弱网、多模态有极致要求，需集成 [AOQ SDK](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md) 并管理 [Token](../concepts/token.md)。 + +### 核心流程（以 AOQ 为例） +1. **初始化引擎**：调用 `createEngine`，设置 `AoqEngineDelegate` 回调。 +2. **启动采集**：`startAudioCapture()` + `startVideoCapture()`（可选），支持内部/外部采集模式。 +3. **获取凭证**：业务 AppServer 向百炼网关请求 [Token](../concepts/token.md)（见 [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md)）。 +4. **建连控制**：调用 `connect(config)` 前，务必 `enableSendMediaStream(.audio, false)` 暂停发送；收到 `session.updated` 事件后，再 `enableSendMediaStream(.audio, true)` 开启（见 [媒体流发送管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md)）。 +5. **事件交互**：通过 `send()` 发送 `session.update`、`input_audio_buffer.append` 等事件；通过 `onDataMsg` 处理 `response.text.delta`、`response.audio.delta` 等响应。 + +### WebRTC 注意事项 +- 浏览器端无法直连百炼网关（CORS 限制），SDP 交换必须由业务后端代理，**禁止在前端代码中暴露 API Key**。 +- `RTCPeerConnection` 应配置 `iceServers: []`（服务端 ICE-lite 模式），无需 STUN/TURN 服务器。 +- 视频发送需通过 Canvas 降帧（如 2fps）并 `replaceTrack(null)` 实现门控，确保 `session.created` 后才推送媒体流（见 [通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md)）。 + +## 限制和注意事项 + +- **并发与限流**：所有协议共享百炼平台的并发限流策略，具体配额请参考 [限流文档](https://help.aliyun.com/zh/model-studio/rate-limit)，未明确说明的默认值以控制台为准。 +- **浏览器兼容性**：WebRTC 仅支持现代浏览器（Chrome/Edge/Firefox/Safari），AOQ 不支持浏览器环境。 +- **媒体流同步**：AOQ 协议下，`enableSendMediaStream` 是强制控制点——模型未返回 `session.updated` 前发送媒体流将被丢弃或导致连接异常；WebSocket/WebRTC 无此严格约束，但强烈建议遵循相同模式以保证稳定性。 +- **SDK 版本**：AOQ Client SDK v1.0.1 起支持 Opus 编解码插件（需单独下载 [libPluginOpus](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md)），未加载插件时仅支持 PCM。 +- **音频设备管理**：AOQ SDK 的 `isVoipMode` 参数影响硬件 AEC 行为，且扬声器切换（`enableSpeakerphone`）仅在 VoIP 模式下有效；非 VoIP 模式下调用将触发 `AoqECAudioDeviceEarpieceRequiresVoipMode` 错误（见 [音频常用功能介绍](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md)）。 +- **自定义采集/播放**：若需外部音频/视频源（如 TTS 输出、屏幕录制），必须先调用 `startAudioCapture({isExternal:true})` 或 `startVideoCapture({isExternal:true})`，再通过 `pushAudioExternalStreamData` / `pushExternalVideoCapturedFrame` 推送数据，否则 SDK 不消费帧（见 [自定义音频采集](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md) 和 [自定义视频输入](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md)）。 + +## 来源文档 + +- [Realtime API简介](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md) +- [通过WebRTC使用多模态交互套件实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) +- [通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md) +- [通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md) +- [SDK下载](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md) +- [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md) +- [实现接通模型/应用](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md) +- [AOQ SDK简介](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md) +- [连接状态管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md) +- [音频常用功能介绍](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md) +- [媒体流发送管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) +- [自定义音频采集](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md) +- [自定义音频播放](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md) +- [自定义视频输入](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md) +- [视频常用功能介绍](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md index fad0f724..fb16f5d3 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md @@ -1,111 +1,94 @@ # toolkits and [frameworks](frameworks.md) -阿里云百炼的通义千问等模型提供了一套与 OpenAI 高度兼容的接口体系，覆盖 Chat Completions、Responses、Completions、Embedding、文件、Batch、Conversations 等能力，并可直接接入 LangChain/LangChain4j 等主流框架。对于已有 OpenAI 应用，通常只需替换 `api_key`、`base_url` 与 `model` 三项即可完成迁移，无需改动业务逻辑。 - -## 迁移三要素与服务地址 - -将 OpenAI 应用迁移到百炼的核心是配置以下三项（详见 [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)）： - -- **`api_key`**：替换为[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。**各地域的 API Key 不同**，切换地域时需同步更换。建议配置到环境变量 `DASHSCOPE_API_KEY` 以降低泄露风险。 -- **`base_url`**：OpenAI SDK 调用统一使用 `/compatible-mode/v1` 路径；HTTP 调用在其后追加具体资源路径（如 `/chat/completions`、`/responses`、`/embeddings`、`/files`）。 -- **`model`**：替换为百炼支持的模型名称。 - -各地域 SDK `base_url`： - -| 地域 | base_url | -| --- | --- | -| 华北2（北京） | `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` | -| 新加坡 | `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| 日本（东京） | `https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| 德国（法兰克福） | `https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1` | -| 美国（弗吉尼亚） | `https://dashscope-us.aliyuncs.com/compatible-mode/v1` | - -其中 `{WorkspaceId}` 为业务空间 ID，可在百炼控制台**业务空间详情**页面查看。 - -> **注意**：百炼为北京、新加坡地域推出了业务空间专属域名，性能与稳定性更佳，建议从旧域名迁移：北京 `https://dashscope.aliyuncs.com` → `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`；新加坡 `https://dashscope-intl.aliyuncs.com` → `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`。现有域名仍可正常使用。 - -> **注意**：Responses 与 Conversations 接口的旧版路径 `/api/v2/apps/protocols/compatible-mode/v1/...` 即将停止维护，请尽快迁移至新版 `/compatible-mode/v1/...` 路径。 - -## 各兼容接口一览 - -### Chat Completions（对话补全） - -最常用的兼容接口，支持非流式、流式（`stream=True`，配合 `stream_options={"include_usage": True}` 返回 Token 统计）与 function call（工具调用）。支持模型广泛：Qwen 大语言模型（商业版/开源版）、Qwen-VL、Qwen-Coder、Qwen-Omni、Qwen-Math，以及 DeepSeek、Kimi、GLM、MiniMax 等三方模型。 - -> **注意**：三方直供模型仅在中国站的中国内地地域可用，调用前需先在百炼控制台开通对应服务。Qwen-Audio 不支持 OpenAI 兼容协议，仅支持 DashScope 协议。 - -### Responses（智能体原生接口） - -作为 Chat Completions 的演进版本，Responses API 内置联网搜索、网页抓取、代码解释器、文搜图/图搜图等工具，输入更灵活（可直接传字符串），并通过 `previous_response_id` 自动管理多轮上下文，无需手动拼接消息历史。详见 [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 - -- 支持模型示例：`qwen3-max`、`qwen3.7-plus`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus` 等。 -- `previous_response_id` 需传入上一轮响应的顶层 `id`（`resp_xxx`），而非 `output` 数组内消息的 `id`；当前响应 `id` 有效期为 **7 天**。 - -### Conversations（会话管理） - -提供会话的创建、查询、更新、删除及消息项管理。配合 Responses API 可自动注入历史上下文，实现跨设备、跨会话的对话延续。初始消息项 `items` 最多 20 条，`metadata` 最多 16 对键值对（key ≤ 64 字符、value ≤ 512 字符）。删除会话时其消息项不会被删除。 - -### Completions（文本补全） - -专为代码补全、内容续写设计，当前仅支持 `qwen-coder-turbo`，且**仅适用于中国内地（北京地域）**。通过 `<|fim_prefix|>...<|fim_suffix|>...<|fim_middle|>` 模板可实现「前缀生成后续」或「前缀+后缀生成中间」两种补全（暂不支持仅凭后缀生成前缀）。关键参数包括 `max_tokens`、`temperature`、`top_p`、`stop`、`seed`、`presence_penalty` 等，详见 [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 - -### Embedding（文本向量） - -兼容 OpenAI Embedding 规范，支持 `text-embedding-v1/v2/v3/v4`。其中 v3、v4 支持通过 `dimensions` 参数指定向量维度（v4 可选 64~2048 多档，默认 1024）。 - -> **注意**：多模态 Embedding 模型（如 qwen3-vl-embedding、tongyi-embedding-vision 系列）不支持 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，需改用[多模态向量接口](https://help.aliyun.com/zh/model-studio/multimodal-embedding-api-reference)。 - -### Vision（视觉理解） - -Qwen-VL、QVQ、Qwen-OCR 兼容 OpenAI Chat 接口，通过 `content` 数组中的 `image_url` 传入图片。各地域支持的模型有差异。QVQ 模型仅支持[流式输出](../concepts/streaming.md)。 - -### 文件接口与 Batch - -文件上传接口（`client.files.create`）通过 `purpose` 区分用途，详见 [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)： - -| purpose | 用途 | 单文件大小上限 | -| --- | --- | --- | -| `file-extract` | Qwen-Long / Qwen-Doc-Turbo 文档问答与数据提取 | 150 MB | -| `batch` | 批量推理输入（jsonl） | 500 MB | -| `fine-tune` | 模型调优数据集（jsonl） | 300 MB | - -百炼存储空间上限为 10000 个文件、总计 100 GB，达到任一上限后新上传会失败，需删除文件释放配额。上传返回的文件 ID（如 `file-batch-xxx`）可重复使用。 - -百炼提供两种批量推理方式，费用均约为实时调用的 **50%**： - -- **Batch（文件输入）**：上传 jsonl 文件异步批处理，适合大批量、时效性要求不高的场景（数据分析、模型评测）。可先用测试模型 `batch-test-model` 做全链路验证（文件 ≤ 1 MB、≤ 100 行、最大并行 2 个任务，不产生推理费用）。 -- **Batch Chat**：保持与实时 API 一致的同步调用方式，仅需将 `base_url` 改为 `https://batch.dashscope.aliyuncs.com/compatible-mode/v1`，单次仅支持一个请求；默认等待超时 3600 秒（可设 60~3600 秒）。 - -> **注意**：Batch 场景下 `enable_thinking` 须作为请求 body 的顶层参数（与 `model` 同级）传入，不能放在 `extra_body` 中；`qwen3.7`/`qwen3.6`/`qwen3.5` 系列默认开启思考模式，会产生额外思考 Token 成本，建议显式设置。 - -## 框架集成（LangChain） - -百炼可通过两条路径接入 LangChain（Python / JavaScript / Java），详见 [在LangChain中使用阿里云百炼](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)： - -- **OpenAI 兼容路径**：使用 `langchain_openai.ChatOpenAI`（JS 为 `@langchain/openai`，Java 为 `langchain4j-open-ai`），配置 `base_url` 指向 `compatible-mode/v1`。**仅支持 OpenAI 兼容模式覆盖的部分模型**。 -- **DashScope 原生路径**：使用 `ChatTongyi`（`langchain-community` + `dashscope`）或 JS 的 `ChatAlibabaTongyi`，**支持百炼所有文本生成模型（含部署后的模型）**。 - -> **注意**：LangChain4j 1.0.0-beta3 需要 Java 17 及以上版本，使用 Java 11 编译会报 `Unsupported class file major version 61` 错误。 - -## 限制与注意事项 - -- **地域隔离**：API Key 与 `base_url` 均按地域区分，跨地域调用必须成对更换；不同接口/模型在各地域的可用性存在差异，以[百炼控制台](https://bailian.console.aliyun.com/)为准。 -- **协议差异**：并非所有模型都支持 OpenAI 兼容协议（如 Qwen-Audio、多模态 Embedding），此类模型需使用 DashScope 原生协议。 -- **端点区别**：普通请求走各地域 `compatible-mode/v1`，而 Batch Chat 使用独立的 `batch.dashscope.aliyuncs.com` 域名。 -- 调用失败时请参考[错误码](https://help.aliyun.com/zh/model-studio/error-code)排查。 +阿里云百炼平台提供多种 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)及专用工具链，支持开发者快速迁移现有应用或构建新场景。核心能力覆盖文本生成（Chat/Completions/Responses）、多模态理解（Vision）、向量嵌入（Embedding）、文件处理（Files）、批量推理（Batch）、会话管理（Conversations）以及主流框架集成（如 LangChain）。所有接口均通过统一的 `compatible-mode/v1` 路径暴露，但模型支持范围、参数行为和地域端点存在差异，需按场景谨慎选型。 + +## 支持的模型/功能 + +百炼支持的 OpenAI 兼容能力按功能维度划分如下： + +- **标准 Chat 接口**：兼容 `chat/completions`，支持 Qwen 系列（`qwen-plus`, `qwen-flash`, `qwen3-*`）、Qwen-VL、Qwen-Coder、Qwen-Omni、Qwen-Math 及第三方直供模型（DeepSeek、Kimi、GLM、MiniMax）等 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 +- **Responses API（智能体原生）**：专为复杂任务设计，内置联网搜索、网页抓取、代码解释器等工具，支持 `qwen3.7-plus`、`qwen3.5-flash`、`qwen3-coder-next` 等 20+ 个 Qwen3 系列模型及 `qwen-plus` [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 +- **Vision 接口**：支持[多模态输入](../concepts/multi-modal-input.md)（文本+图像 URL/Base64），适配 `qwen3-vl-plus`、`QVQ`、`Qwen-OCR` 模型 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md)。 +- **Embedding 接口**：提供 `text-embedding-v1` 至 `v4` 四代文本向量模型，支持可调维度（如 `v4` 支持 64–2048）及多语种，但**多模态 Embedding（如 `qwen3-vl-embedding`）不支持 OpenAI 兼容协议** [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md)。 +- **Files 接口**：用于上传文档供 `Qwen-Long`（长文档问答）、`Qwen-Doc-Turbo`（数据提取）或 Batch/Fine-tune 任务使用，支持 TXT/DOCX/PDF/图片等格式 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)。 +- **Batch 接口**：分两种模式： + - **文件输入（JSONL）**：异步批量处理，支持 `qwen3.7-max`（256K 上下文）、`qwen-vl-plus` 等 30+ 模型； + - **同步 Batch Chat**：单请求阻塞式调用，仅需切换 `base_url` 即可复用现有 Chat 代码 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md)。 +- **Conversations API**：管理跨设备会话状态，支持创建、查询、更新、删除会话及追加消息项，与 Responses API 配合实现上下文自动注入 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md)。 +- **Completions 接口**：专用于代码补全，当前**仅支持 `qwen-coder-turbo` 模型**，且仅限华北2（北京）地域 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 + +> **注意**：文档 1 和文档 2 均强调业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）的性能与稳定性优势，但文档 3 的 Completions 接口示例仍使用旧域名 `https://dashscope.aliyuncs.com`，存在迁移指引不一致问题，建议以文档 1 和 2 的推荐为准。 + +## 关键参数 + +各接口共性参数与关键差异如下： + +| 参数 | 说明 | 注意事项 | +|------|------|----------| +| `base_url` | 必填，服务端点。地域不同路径不同，北京/新加坡需替换 `{WorkspaceId}` | 北京/新加坡必须使用业务空间专属域名；弗吉尼亚/东京/法兰克福无需 WorkspaceId；Batch Chat 使用独立域名 `https://batch.dashscope.aliyuncs.com` [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) | +| `model` | 模型名称，严格区分大小写与版本后缀（如 `qwen3.7-plus` vs `qwen3.7-plus-2026-01-23`） | Vision 接口不支持 `qwen-audio`；Completions 接口仅支持 `qwen-coder-turbo`；Embedding 不支持多模态模型 | +| `stream` / `stream_options` | 控制[流式输出](../concepts/streaming-output.md)。`stream_options={"include_usage": true}` 可在流末尾返回 token 统计 | QVQ 模型**强制[流式输出](../concepts/streaming-output.md)**，非流式调用将失败 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) | +| `enable_thinking` | Batch 场景下控制思考模式（影响 token 成本）。`qwen3.5/3.6/3.7` 系列默认开启，**必须作为 `body` 顶层参数传入，不可置于 `extra_body`** | 此参数在 Chat/Responses 接口中无效，仅 Batch JSONL 文件中生效 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) | +| `previous_response_id` | Responses API 多轮对话的核心参数，传入上一轮响应的顶层 `id`（UUID 格式） | **不可传入 `output` 数组内消息的 `id`**（如 `msg_xxx`），否则上下文关联失败 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) | +| `dimensions` | Embedding 接口特有，指定向量维度（仅 `v3`/`v4` 支持） | `v1`/`v2` 不支持该参数，设置将导致错误 | + +## 使用方式 + +### 通用步骤 +1. **获取并配置 API Key**：通过百炼控制台获取，**强烈建议配置至环境变量 `DASHSCOPE_API_KEY`**，避免硬编码泄露风险； +2. **选择 `base_url`**：根据地域与接口类型确定（见上表），北京/新加坡务必替换 `{WorkspaceId}`； +3. **安装 SDK**：Python 推荐 `pip install -U openai langchain_openai`；Java/Node.js/Go 等参照对应文档； +4. **构造请求**：按接口规范传入 `model`、`input`/`messages`/`prompt` 等核心字段。 + +### 典型调用示例 +- **Chat（非流式）**： + ```python + from openai import OpenAI + client = OpenAI( + api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" + ) + resp = client.chat.completions.create(model="qwen-plus", messages=[{"role":"user","content":"你好"}]) + ``` +- **Responses（多轮）**： + ```python + # 第一轮 + resp1 = client.responses.create(model="qwen3.7-plus", input="我的名字是张三") + # 第二轮（自动继承上下文） + resp2 = client.responses.create(model="qwen3.7-plus", input="你还记得我的名字吗？", previous_response_id=resp1.id) + ``` +- **Vision（图文理解）**： + ```python + completion = client.chat.completions.create( + model="qwen3-vl-plus", + messages=[{"role":"user","content":[{"type":"text","text":"这是什么"},{"type":"image_url","image_url":{"url":"https://..."}}]}], + stream=True + ) + ``` +- **LangChain 集成**： + - `langchain_openai.ChatOpenAI`：仅支持 OpenAI 兼容模型（如 `qwen-plus`）； + - `langchain_community.chat_models.tongyi.ChatTongyi`：支持全部百炼文本模型（含部署模型）[原文标题](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)。 + +## 限制和注意事项 + +- **地域与模型绑定**：部分模型仅在特定地域可用（如三方直供模型仅限中国内地），需在控制台开通后方可调用； +- **文件限制**：Files 接口上传文件总大小 ≤100 GB，总数 ≤10,000 个；`file-extract` 单文件 ≤150 MB，`batch`/`fine-tune` 单文件 ≤500 MB/300 MB； +- **Batch 超时**：Batch Chat 同步调用默认超时 3600 秒（1 小时），不可超过此值；Batch 文件处理最长等待时间由 `completion_window` 参数控制（如 `"24h"`）； +- **Qwen-Audio 不兼容**：明确不支持 OpenAI 兼容协议，仅能通过 DashScope 原生协议调用； +- **旧路径弃用**：`/api/v2/apps/protocols/compatible-mode/v1/responses` 和 `/api/v2/apps/protocols/compatible-mode/v1/conversations` 已标记为“即将停止维护”，必须迁移到 `/compatible-mode/v1/{responses|conversations}`； +- **参数作用域**：`enable_thinking` 仅在 Batch JSONL 请求体中有效，且必须与 `model` 同级；`previous_response_id` 仅适用于 Responses API，Chat 接口需自行维护消息历史。 ## 来源文档 - [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) - [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md) -- [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) - [OpenAI Vision接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) +- [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) - [OpenAI兼容-Batch（文件输入）](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) -- [OpenAI Embedding接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - [OpenAI兼容-Batch Chat](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) +- [OpenAI Embedding接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - [OpenAI Conversations接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md) - [在LangChain中使用阿里云百炼](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md index 2a8da565..29712c03 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md @@ -1,96 +1,110 @@ # vector and sort -百炼平台围绕"向量化"与"排序"提供了一整套模型 API，覆盖通用文本向量、多模态向量与文本/多模态重排序三大能力。它们共同服务于语义搜索、推荐、聚类、分类与 RAG 检索：向量模型负责把文本、图片、视频编码为同一语义空间中的数值向量，排序（rerank）模型则在召回阶段之后对候选结果做二次精排，提升最终相关性。 - -## 能力与模型总览 - -按用途可分为三类接口，分别对应不同的 endpoint 与调用方式： - -- **通用文本向量（同步）**：将字符串 / 字符串列表 / 文件转为向量，实时返回。支持 `qwen3.7-text-embedding`、`text-embedding-v4/v3/v2/v1`。详见 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 -- **通用文本向量（批处理）**：面向大规模离线向量化，仅支持异步模式，通过文件 URL 输入。支持 `text-embedding-async-v2/v1`。详见 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **多模态向量**：将文本、图片、视频编码到同一语义空间，支持跨模态检索与融合表征。支持 `qwen3-vl-embedding`、`qwen2.5-vl-embedding`、`tongyi-embedding-vision-plus/flash`（含 `2026-03-06` 快照版）、`multimodal-embedding-v1`。详见 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 -- **文本 / 多模态排序**：对召回文档做精排，支持 `qwen3-rerank`、`qwen3-vl-rerank`（多模态）、`gte-rerank-v2`。详见 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 - -> **注意**：`gte-rerank` 模型将于 2026-05-30 下线，官方推荐迁移到 `qwen3-rerank`。新项目请直接选用 `qwen3-rerank` / `qwen3-vl-rerank`。 - -## 通用文本向量 - -### 同步接口 - -- **兼容方式**：提供 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，可用 OpenAI SDK 直连。 - - base_url：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - - endpoint：`POST .../compatible-mode/v1/embeddings` - - 调用前需将 `{WorkspaceId}` 替换为真实业务空间 ID。 -- **关键参数**： - - `model`（必选）：模型名称。 - - `input`（必选）：`string` / `array` / `file` 三种形态。 - - `dimensions`（可选）：仅 `text-embedding-v3/v4`（及 `qwen3.7-text-embedding` 的 2560 维）支持自定义维度，取值 2560/2048/1536/1024/768/512/256/128/64，默认 1024。 - - `encoding_format`（可选）：当前仅支持 `float`。 -- **输入上限（按模型区分）**： - - `qwen3.7-text-embedding`：单条字符串最长 128,000 Token；列表/文件最多 20 条。 - - `text-embedding-v3/v4`：单条 8,192 Token；列表/文件最多 10 条。 - - `text-embedding-v1/v2`：单条 2,048 Token；列表/文件最多 25 条。 - -> **注意**：`dimensions` 只对部分模型生效——`text-embedding-v1/v2` 为固定维度（分别 1536 / 1536），传入该参数无意义；`v4` 才支持 2048/1536 等高维度。选维度前请对照模型概览表。 - -### 批处理接口 - -批处理专用于大批量离线场景，特点是**仅支持异步**： - -- endpoint：`POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding`。 -- HTTP 请求**必须**带请求头 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。 -- 输入通过 `input.url` 传入文件 HTTP URL（一行一条），单行最长 2,048 Token、最多 100,000 行、文件不超过 200MB。 -- `parameters.text_type` 可选 `document`（默认）或 `query`；检索类非对称任务建议区分 query / document。 -- 调用两步走：创建任务拿到 `task_id` → `GET .../api/v1/tasks/{task_id}` 轮询结果。任务状态含 PENDING / RUNNING / SUCCEEDED / FAILED / CANCELED / UNKNOWN。 -- **数据时效**：任务结果 URL 仅保留 24 小时，务必及时下载，详见 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **限流**：`text-embedding-async-v2` 任务下发 RPS 为 1，排队+运行作业不超过 50 个，同时并发运行不超过 3 个。 - -## 多模态向量 - -多模态向量把 text / image / video 编码进**同一语义空间**，可直接用余弦相似度做跨模态匹配（以文搜图、以图搜视频等）。 - -- endpoint：`POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding`。 -- 输入通过 `input.contents` 数组传入，每个元素为 `{"模态类型": "值"}`，支持 `text` / `image` / `video` / `multi_images` 四种类型。图片可用 URL 或 Base64 Data URI，视频仅支持公开 URL。 -- **独立向量 vs 融合向量**： - - 独立向量：为每个输入分别生成一个向量，适合逐项对比（以图搜图）。 - - 融合向量：将所有输入融合为 1 个向量，适合整体语义表征（如商品图+描述文本）。 - - `qwen3-vl-embedding` 通过 `enable_fusion=true` 开启融合；`tongyi-embedding-vision-*-2026-03-06` 则通过把 text/image/video 放进同一个 content 对象来生成融合向量（不使用 `enable_fusion`）。 -- **关键参数（在 `parameters` 内）**：`dimension`（不同模型取值不同）、`output_type`（仅 `dense`）、`fps`（视频帧率比例 [0,1]）、`instruct`（任务说明，建议英文）、`res_level`（分辨率档位 0/1/2/3，仅 2026-03-06 版）、`max_video_frames`（最大采样帧，≤64，仅 2026-03-06 版）。 - -> **注意**：各模型的向量类型能力差异明显——`qwen2.5-vl-embedding` **仅**支持融合向量、不支持独立向量与多图；`tongyi-embedding-vision-plus/flash`（非快照版）**仅**支持独立向量；`multimodal-embedding-v1` 与 `tongyi-embedding-vision-plus/flash` 不支持 `dimension` 参数（维度固定）。选型前务必核对 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) 中的"模型能力对照"表。 - -## 文本 / 多模态排序（Rerank） - -排序模型对召回文档二次精排，返回相关性分数。不同模型使用不同接口： - -- `qwen3-rerank`：`POST .../compatible-api/v1/reranks`，且 `query` / `documents` / `top_n` / `instruct` 与 `model` **同层级**（不使用 `input` / `parameters` 包装）。 -- `qwen3-vl-rerank`（多模态）/ `gte-rerank-v2`：`POST .../api/v1/services/rerank/text-rerank/text-rerank`，参数需包装进 `input` 与 `parameters` 对象。 - -关键参数与返回： - -- `query`（必选）：最大 4,000 Token；`qwen3-vl-rerank` 支持 `{"text": ...}` 或 `{"image": ...}` 对象形式。 -- `documents`（必选）：候选文档数组；`qwen3-vl-rerank` 每项可为 `text` / `image` / `video`。 -- `top_n`（可选）：返回前 N 条，默认全部。 -- `return_documents`（可选，默认 `false`）：是否回带原文，仅 `gte-rerank-v2` / `qwen3-vl-rerank` 支持。 -- `instruct`（可选）：仅 `qwen3-rerank` / `qwen3-vl-rerank` 生效，用于切换问答检索 / 语义相似度等排序策略，建议英文。 -- `fps`（可选）：仅 `qwen3-vl-rerank` 支持，控制视频抽帧比例。 -- 返回 `results` 按 `relevance_score`（0.0–1.0）降序排列，`index` 对应输入原始位置。 - -> **注意**：`relevance_score` 是**单次请求内的相对分数**，仅用于本次请求内排序，不可作为跨请求比较的绝对阈值。此外两类接口响应结构不同——`qwen3-rerank` 的 `results` 位于响应顶层且无 `output` 对象，其余模型结果在 `output.results` 内，详见 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 - -## 通用限制与注意事项 - -- **前提条件**：所有接口都需先[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并配置到环境变量 `DASHSCOPE_API_KEY`；SDK 调用还需安装 DashScope SDK。 -- **地域**：同步向量与 rerank 走 `maas.aliyuncs.com`（需替换 `{WorkspaceId}`），多模态与批处理走 `dashscope.aliyuncs.com`；新加坡地域需将 base_url 换为 `dashscope-intl.aliyuncs.com`。 -- **SDK 与 HTTP 差异**：HTTP 使用嵌套的 `input` / `parameters` 结构，DashScope SDK 多为扁平参数，开发时注意区分。 -- **超长截断**：rerank 中单条超过"单条最大输入 Token"会被截断，API 仅基于截断后内容计算，可能影响排序准确性。 -- **限流与错误码**：触发条件参见平台[限流](https://help.aliyun.com/zh/model-studio/rate-limit)文档，失败响应通过 `code` / `message` 指明原因，对照[错误码](https://help.aliyun.com/zh/model-studio/error-code)排查。 +百炼平台提供文本向量化（vector）、多模态向量化（multimodal vector）和文本排序（rerank）三类核心能力，覆盖语义搜索、RAG、跨模态检索等典型场景。所有能力均通过标准化 API 提供，支持同步/异步调用、OpenAI 兼容模式及 DashScope SDK 封装，适用于从单条文本处理到百万级批量任务的全量需求。详细模型能力与参数约束请参考各接口文档。 + +## 支持的模型/功能 + +### 文本向量化 +- **同步模型**：`qwen3.7-text-embedding`（最高 128K token 输入）、`text-embedding-v4`（默认 1024 维，支持 `dimensions` 参数）、`text-embedding-v3`、`text-embedding-v2`、`text-embedding-v1` +- **异步批处理模型**：`text-embedding-async-v2`（单次最多 100,000 行）、`text-embedding-async-v1` +- **[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**：支持 `embeddings.create` 调用，需配置 `base_url` 为兼容模式地址（详见 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)） + +### 多模态向量化 +- **支持模态**：文本、图像（JPEG/PNG/WEBP 等）、视频（MP4/AVI/MOV 等 URL）及多图序列（`multi_images`） +- **关键模型**： + - `qwen3-vl-embedding`：支持独立向量与融合向量（通过 `enable_fusion=true`），默认 2560 维 + - `tongyi-embedding-vision-plus-2026-03-06`：支持融合向量（同 content 对象内[多模态输入](../concepts/multi-modal-input.md)）、`res_level` 和 `max_video_frames` 参数 + - `qwen2.5-vl-embedding`：仅支持融合向量，不支持 `multi_images` +- 详见 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) + +### 文本排序（Rerank） +- **主流模型**： + - `qwen3-rerank`：纯文本排序，[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)（`/compatible-api/v1/reranks`），支持 `instruct` 任务指令 + - `qwen3-vl-rerank`：跨模态排序（文本/图片/视频混合），支持图片/视频作为 query + - `gte-rerank-v2`（即将下线）：仅文本，最大支持 30,000 文档，[官方公告](https://www.aliyun.com/notice/118217) 明确推荐迁移至 `qwen3-rerank` +- 接口路径与请求结构因模型而异，详见 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) + +> **注意**：`gte-rerank` 系列模型（含 `gte-rerank-v2`）将于 2026 年 05 月 30 日下线，新项目应直接使用 `qwen3-rerank` 或 `qwen3-vl-rerank`。 + +## 关键参数 + +| 参数 | 适用模型 | 说明 | 示例值 | +|------|----------|------|--------| +| `model` | 全部 | 必选，指定模型名称 | `"text-embedding-v4"`, `"qwen3-vl-rerank"` | +| `input` / `query` / `documents` | 按模型区分 | 向量化：`input`（string/array/file）；排序：`query` + `documents`（array） | `"衣服的质量杠杠的..."`, `[{"text":"doc1"},{"image":"url"}]` | +| `dimensions` | `text-embedding-v3/v4`, `qwen3-vl-embedding`, `tongyi-embedding-vision-plus-2026-03-06` 等 | 可选，指定输出向量维度；部分模型（如 `tongyi-embedding-vision-plus`）不支持 | `1024`, `2560` | +| `encoding_format` | 同步文本向量 | 仅支持 `"float"` | `"float"` | +| `enable_fusion` | 仅 `qwen3-vl-embedding` | bool，启用融合向量生成 | `true` | +| `top_n` | `qwen3-rerank`, `qwen3-vl-rerank`, `gte-rerank-v2` | 返回前 N 个结果 | `5` | +| `instruct` | `qwen3-rerank`, `qwen3-vl-rerank` | 任务指令，影响排序策略（如问答检索 vs 语义相似度） | `"Given a web search query, retrieve relevant passages..."` | +| `text_type` | 异步批处理（`text-embedding-async-*`） | 区分 `query`（检索查询）或 `document`（底库文本），影响向量表征 | `"query"` | + +## 使用方式 + +### 同步调用（文本向量） +- **HTTP**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/embeddings` +- **SDK（OpenAI 兼容）**：使用 `openai` 客户端，设置 `base_url` 为兼容模式地址 +- **示例**（Python）： + ```python + from openai import OpenAI + client = OpenAI( + api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" + ) + resp = client.embeddings.create( + model="text-embedding-v4", + input=["hello", "world"], + dimensions=1024 + ) + ``` + +### 异步批处理（大文件） +- **HTTP**：两步调用 —— 先 `POST /api/v1/services/embeddings/text-embedding/text-embedding` 创建任务，再 `GET /api/v1/tasks/{task_id}` 查询结果 +- **SDK**：`BatchTextEmbedding.call()`（同步封装）或 `BatchTextEmbedding.async_call()`（异步） +- **注意**：输入必须为公开可访问的 URL 文件（如 OSS），单文件 ≤ 200MB，单行 ≤ 2048 token + +### 多模态向量 +- **HTTP**：`POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding` +- **输入格式**：`input.contents` 为数组，每个元素为 `{"text": "..."}`, `{"image": "url"}`, `{"video": "url"}` 或 `{"multi_images": [...]}` +- **融合向量**：`qwen3-vl-embedding` 需 `parameters.enable_fusion=true`；`tongyi-embedding-vision-plus-2026-03-06` 需将多模态字段置于同一对象内 + +### 文本排序 +- **纯文本**：`qwen3-rerank` 使用 `/compatible-api/v1/reranks`，参数扁平化（`query`, `documents`, `top_n` 同级） +- **跨模态**：`qwen3-vl-rerank` 使用 `/api/v1/services/rerank/text-rerank/text-rerank`，`query` 和 `documents` 均支持模态对象 +- **SDK**：统一使用 `dashscope.TextReRank.call()`，参数自动适配对应模型 + +## 限制和注意事项 + +- **[Token](../concepts/token.md) 限制**： + - `qwen3.7-text-embedding` 单文本最长 128,000 token；`text-embedding-v4` 为 8,192；`qwen3-vl-embedding` 文本为 32,000；`qwen3-rerank` 单条文档为 4,000 + - 总 [Token](../concepts/token.md) 计算规则：`qwen3-rerank` 为 `query_tokens + sum(doc_tokens)`；`qwen3-vl-rerank` 为 `query_tokens × doc_count + sum(doc_tokens)`，上限 120,000 + +- **批量规模**： + - 同步向量：`qwen3.7-text-embedding` 最多 20 条；`text-embedding-v4` 最多 10 条 + - 异步批处理：`text-embedding-async-v2` 单次最多 100,000 行 + - 排序：`qwen3-rerank` 最多 500 文档；`qwen3-vl-rerank` 文本文档最多 100，图片最多 40，视频最多 4 + +- **地域与 endpoint**： + - 同步/兼容接口：北京地域为 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/...` + - 多模态/排序 HTTP 接口：公共 endpoint 为 `https://dashscope.aliyuncs.com/...`，新加坡地域需替换 host + - 异步批处理 SDK：需显式设置 `dashscope.base_http_api_url` + +- **免费额度与计费**： + - 所有模型均提供开通后 90 天内的免费额度（如 `text-embedding-v4` 为 100 万 token），具体额度见各模型概览表 + - 多模态模型按模态分别计费（文本/图片/视频单价不同），详见 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) + +- **限流**： + - 同步接口受通用 [限流](https://help.aliyun.com/zh/model-studio/rate-limit) 约束 + - 异步批处理：单用户并发运行中任务 ≤ 3 个，排队中 + 运行中任务 ≤ 50 个 + - `qwen3-vl-rerank` 视频处理依赖 `fps` 参数控制帧数，避免超时 + +> **注意**：`text-embedding-v1` 与 `text-embedding-v2` 的语种支持范围（50+/100+）在文档中存在表述差异，以 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) 中表格为准；`multimodal-embedding-v1` 不支持 `dimension` 参数，固定 1024 维，与文档 3 中其他模型形成明确区分。 ## 来源文档 - [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) -- [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) -- [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) - [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) +- [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) +- [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md index f4a55920..db9a6d1b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md @@ -1,102 +1,119 @@ # video generation api -阿里云百炼平台提供覆盖多家厂商（万相 Wan、爱诗 PixVerse、Vidu、可灵 Kling、HappyHorse 等）的视频生成 API，支持文生视频、图生视频（首帧/首尾帧）、参考生视频、视频编辑、数字人、人像驱动、视频超清与对口型等能力。所有视频生成任务均通过统一的异步调用模式完成，开发者先提交任务拿到 `task_id`，再轮询查询结果。 +百炼平台的 Video Generation API 提供多种视频生成与编辑能力，覆盖文生视频、图生视频、参考生视频、视频编辑、数字人播报、口型替换、风格重绘等核心场景。所有接口均采用异步调用模式（`X-DashScope-Async: enable`），任务创建后返回 `task_id`，需轮询获取结果，`task_id` 有效期为 24 小时。开发者必须确保模型、Endpoint URL 与 API Key 属于同一地域，跨地域调用将失败。 -## 统一调用模式：异步「创建任务 → 轮询获取」 +## 支持的模型/功能 -由于视频生成耗时较长（通常 1-5 分钟，个别统一编辑模型约 5-10 分钟），API 全部采用异步方式，流程分两步： +API 支持多系列模型，按能力可分为以下几类： -1. **创建任务**：向 `video-synthesis` 端点发起 `POST` 请求，请求头必须带 `X-DashScope-Async: enable`（缺少会报错 `current user api does not support synchronous calls`），返回一个 `task_id`。 -2. **轮询获取**：用 `task_id` 发起 `GET https:///api/v1/tasks/{task_id}` 查询任务状态，直到完成并拿到视频 URL。 +- **通用生成类**： + - `wan2.7` 系列（推荐）：支持首帧/首尾帧/视频续写（[万相2.7-图生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md)）、文生视频（[万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md)）、参考生视频（[万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md)）、视频编辑（[万相2.7-视频编辑API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md)）。 + - `happyhorse` 系列：支持图生视频、文生视频、参考生视频、视频编辑（[HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) 等）。 + - `pixverse`（爱诗）系列：支持文生视频、图生视频（首帧/首尾帧）、参考生视频、对口型、动作模仿、超清等（[爱诗-文生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) 等）。 + - `vidu`、`kling`：提供高性能文生视频与图生视频能力（[Vidu-文生视频API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md)、[可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md)）。 -其余通用约定： +- **人物驱动类（数字人/肖像动画）**： + - `liveportrait`、`emo`、`videoretalk`、`animateanyone`、`emoji`：均需先调用检测模型（如 `liveportrait-detect`）验证输入合规性，再生成视频（[图生播报视频-灵动人像LivePortrait](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md)、[图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) 等）。 + - `wan2.2-s2v`（数字人）：基于单图+音频生成说话/唱歌视频，流程为 `wan2.2-s2v-detect` → `wan2.2-s2v`（[万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md)）。 -- `task_id` 有效期为 **24 小时**，过期无法查询（返回状态 `UNKNOWN`）；请勿重复创建任务，轮询即可。 -- 请求头 `Content-Type: application/json`、`Authorization: Bearer $DASHSCOPE_API_KEY` 为必填。 -- 新手可参考 [Postman 首次调用指引](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md)。 +- **专用增强类**： + - `video-style-transform`：8种预设艺术风格重绘（[视频风格重绘API参考](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md)）。 + - `pixverse/pixverse-upscale`：4K超分（[爱诗-视频超清API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md)）。 + - `wan2.2-animate-move` / `wan2.2-animate-mix`：图生动作、视频换人（[万相-图生动作API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md)、[万相-视频换人API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md)）。 -> **注意**：绝大多数视频生成模型使用端点路径 `/api/v1/services/aigc/video-generation/video-synthesis`，但部分数字人/换人/图生动作类模型（[万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md)、[万相-图生动作](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md)、[万相-视频换人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md)、[万相2.2-首尾帧](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md)）使用的是 `/api/v1/services/aigc/image2video/video-synthesis`。接入时请以对应文档端点为准。 - -## 支持的模型与功能 - -按厂商与任务类型划分，主要能力如下： - -- **万相 Wan（2.7 新版协议）**： - - 文生视频（`wan2.7-t2v-*`），支持通过 `prompt` 自然语言控制单/多镜头。 - - 图生视频（`wan2.7-i2v-*`），支持多模态输入（文本/图像/音频/视频），可完成首帧生视频、首尾帧生视频、视频续写三大任务。 - - 参考生视频（`wan2.7-r2v-*`），多主体参考（图像+视频+音色）。 - - 视频编辑（`wan2.7-videoedit`），指令编辑与视频迁移。 -- **万相 Wan（旧版协议，2.1-2.6）**：文生视频、图生视频-基于首帧、参考生视频（`wan2.6-r2v-flash`）、首尾帧生视频（`wan2.2-kf2v-flash`）、视频编辑统一模型（`wanx2.1-vace-plus`，支持多图参考、视频重绘等 `function`）。 -- **万相人物/数字人系列**：数字人 `wan2.2-s2v`（图片+音频，需先用 `wan2.2-s2v-detect` 检测图片）、图生动作 `wan2.2-animate-move`、视频换人 `wan2.2-animate-mix`（均含 `wan-std`/`wan-pro` 两种模式）。 -- **爱诗 PixVerse**：文生视频、图生视频、首尾帧生视频（`pixverse/pixverse-c1-*`、`-v6-*`、`-v5.6-*`）、参考生视频（`-r2v`）、视频超清（`pixverse/pixverse-upscale`，固定输出 4K）、视频对口型（`pixverse/pixverse-lipsync`，支持音频驱动或 TTS 文本）、视频动作模仿（`pixverse/pixverse-motioncontrol`）。 -- **Vidu**：文生视频、图生视频、首尾帧生视频、参考生视频（`vidu/viduq3-*`、`viduq2_*`）。 -- **可灵 Kling**：一个模型（`kling/kling-v3-video-generation`、`kling/kling-v3-omni-video-generation`）统一支持文生视频、图生视频（首帧/首尾帧）、参考生视频、视频编辑，并支持智能分镜/多镜头（`multi_shot`、`shot_type`、`multi_prompt`）。 -- **HappyHorse**：文生视频、图生视频-基于首帧、参考生视频（多图像）、视频编辑。 -- **人像驱动系列（两步调用：先检测后生成）**：舞动人像 AnimateAnyone（图生舞蹈）、悦动人像 EMO（图生唱演，`style_level` 控制风格强度）、灵动人像 LivePortrait（图生播报）、表情包 Emoji（预设模板 `driven_id`）、声动人像 VideoRetalk（口型替换）、视频风格重绘 `video-style-transform`（8 种预设风格）。 +> **注意**：`wan2.1`–`wan2.6` 等旧版模型（如 [万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md)）已逐步被 `wan2.7` 新版协议替代；新版统一使用 `/api/v1/services/aigc/video-generation/video-synthesis` 路径，而旧版 `wan2.2-kf2v` 等部分模型仍使用 `/api/v1/services/aigc/image2video/video-synthesis`（见文档33），二者 endpoint 不兼容。 ## 关键参数 -请求体主要由 `model`、`input`、`parameters` 三部分构成： - -- `model`：模型名称，决定能力与协议版本。 -- `input`：任务输入。文生类用 `prompt`；图生/参考/编辑类多用 `media` 数组（`type` 可为 `image_url`/`first_frame`/`last_frame`/`video_url`/`audio_url`/`reference_image` 等），部分旧版模型用 `image_url`/`video_url`/`audio_url`/`first_frame_url`/`last_frame_url`/`ref_images_url` 等独立字段。 -- `parameters`：常见有 `resolution`（如 `480P`/`540P`/`720P`/`1080P`）、`size`（如 `1280*720`）、`duration`（秒）、`watermark`、`prompt_extend`（智能改写）、`audio`、`shot_type`/`multi_shot`（分镜）、`seed`、`style`/`style_level` 等，具体取值随模型不同。 - -> **注意**：多镜头控制方式在不同模型间不一致。万相 2.7 与 PixVerse-c1 通过 `prompt` 自然语言描述控制，设置 `shot_type` 不生效；而旧版万相 2.6（见 [万相-文生视频（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md)）需显式设置 `shot_type: "multi"` 且 `prompt_extend: true` 才能启用多镜头。接入前务必确认所用模型的具体协议。 - -## 地域与域名 - -- **必须保证模型、Endpoint URL 与 API Key 属于同一地域**，跨地域调用会失败（鉴权失败或服务报错）。 -- 多数第三方模型（PixVerse、Vidu、Kling、数字人等）**仅支持华北2（北京）地域**；万相与 HappyHorse 部分能力还支持新加坡、美国（弗吉尼亚）、德国（法兰克福）等地域。 -- 百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，提供更高性能与稳定性，建议迁移： - - 华北2（北京）：`https://dashscope.aliyuncs.com` → `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - - 新加坡：`https://dashscope-intl.aliyuncs.com` → `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` - - `{WorkspaceId}` 为业务空间 ID，可在控制台「业务空间详情」查看；现有域名仍可正常使用。 - -## 限制与注意事项 - -- **版本选型**：万相已推出 2.7 新版协议，[万相2.7-图生视频](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) 支持首帧/首尾帧/续写三大任务，官方推荐优先选用；旧版 wan2.6 及更早模型仅支持首帧生视频。新旧协议接口不通用，`wan2.7-*` 只走新版协议。 -- **服务开通**：PixVerse、Vidu、Kling 等第三方模型需先在百炼控制台模型广场搜索并「立即开通」授权后方可调用。 -- **两步式模型**：数字人、AnimateAnyone、EMO、LivePortrait、Emoji 等需先调用对应的 `-detect` 检测模型确认图片合规（如清晰度、单人、正面），再调用生成模型。检测模型多为同步调用（如 `wan2.2-s2v-detect` 0.004 元/张）。 -- **限流**：视频生成模型通常「同时处理中任务数量」限制较低（多为 1，即同一时刻仅 1 个作业运行，其余排队），任务下发接口 RPS/QPS 约为 5，接入时需做好排队与重试。 -- **计费**：多按生成视频时长计费（如 LivePortrait 0.02 元/秒、EMO/VideoRetalk/AnimateAnyone 0.08 元/秒、数字人 720P 0.9 元/秒），`wan-pro` 等专业模式价格高于标准模式。 -- VideoRetalk 目前仅支持 API 调用，不支持控制台在线体验。 +所有请求需包含以下必选 Header： +- `Content-Type: application/json` +- `Authorization: Bearer $DASHSCOPE_API_KEY` +- `X-DashScope-Async: enable` + +请求体（JSON）核心字段： +- `model`: 模型标识符（如 `"wan2.7-t2v-2026-06-12"`、`"pixverse/pixverse-c1-t2v"`），必须与所选模型精确匹配。 +- `input`: 输入数据容器，结构因模型而异： + - 文生视频：`{"prompt": "..."}` + - 图生视频：`{"media": [{"type": "image_url", "url": "..."}], "prompt": "..."}` + - 首尾帧：`{"media": [{"type": "first_frame", "url": "..."}, {"type": "last_frame", "url": "..."}], "prompt": "..."}` + - 参考生视频：`{"media": [{"type": "reference_image", "url": "..."}, ...], "prompt": "..."}` + - 数字人：`{"image_url": "...", "audio_url": "..."}`（需先通过 detect 模型） +- `parameters`: 可选配置项，常见参数包括： + - `duration`: 视频时长（秒），通常为 3–5 秒（部分模型支持最长 10 秒） + - `resolution` / `size`: 分辨率（如 `"720P"`、`"1280*720"`、`"540P"`） + - `watermark`: 布尔值，控制是否添加水印（默认 `true`） + - `aspect_ratio`: 宽高比（如 `"16:9"`） + - `style_level`: 动作风格强度（`emo` 模型特有，如 `"active"`） + - `mode`: 生成模式（`kling` 模型特有，如 `"std"`） + +## 使用方式 + +1. **环境准备**： + - 在百炼控制台开通对应模型服务（如 PixVerse、Vidu、Wan2.7）。 + - 获取目标地域的 API Key，并配置为环境变量 `DASHSCOPE_API_KEY`。 + - 获取业务空间 ID（WorkspaceId），用于构造专属域名（推荐）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`。 + +2. **发起[异步任务](../concepts/asynchronous-task.md)**： + ```bash + curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ + -H 'X-DashScope-Async: enable' \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H 'Content-Type: application/json' \ + -d '{ + "model": "wan2.7-t2v-2026-06-12", + "input": {"prompt": "一只猫在花园里追逐蝴蝶"}, + "parameters": {"duration": 5, "resolution": "720P"} + }' + ``` + 成功响应含 `task_id`（如 `"task-abc123"`）。 + +3. **轮询获取结果**： + 使用 `task_id` 查询任务状态（示例 URL：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`），直至 `status` 为 `"SUCCESS"`，响应中 `output.video_url` 即为生成视频地址。 + +## 限制和注意事项 + +- **地域强绑定**：模型、Endpoint、API Key 必须同属一个地域（北京/新加坡/美国等），混用将导致鉴权失败或 `401 Unauthorized`（所有文档均强调此点，如 [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md)）。 +- **任务生命周期**：`task_id` 仅在创建后 24 小时内有效，超时需重新提交任务。 +- **并发与限流**：各模型有独立 QPS/RPS 与同时处理任务数限制（如 `liveportrait` 同时仅支持 1 个任务运行），详见各模型资费文档。 +- **输入合规性**：数字人/肖像类模型（`liveportrait`, `emo`, `animateanyone`）**必须前置调用 detect 模型**验证图片质量，否则生成失败（[图生播报视频-灵动人像LivePortrait](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) 明确要求此流程）。 +- **路径差异**：`wan2.7` 及主流新模型统一使用 `/video-generation/video-synthesis`，但 `wan2.2-animate-move`、`wan2.2-animate-mix`、`wan2.2-s2v` 等旧模型仍使用 `/image2video/video-synthesis`（见文档9、10、11），不可混用。 +- **弃用提示**：`wan2.6` 及更早版本（文档30–34）已被明确标注为“旧版协议”，官方推荐迁移至 `wan2.7` 新版（[万相2.7-图生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md)）。 ## 来源文档 - [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-文生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) - [HappyHorse-参考生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) -- [HappyHorse-视频编辑API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) - [万相2.7-图生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) -- [万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) +- [HappyHorse-视频编辑API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) - [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-视频编辑API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) +- [万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) - [万相-图生动作API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) -- [万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - [万相-视频换人API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) +- [万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - [爱诗-文生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) - [爱诗-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) -- [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - [爱诗-参考生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) - [爱诗-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) -- [图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) +- [爱诗-视频对口型API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) +- [爱诗-视频动作模仿API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) +- [爱诗-视频超清API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) +- [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - [图生播报视频-灵动人像LivePortrait](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) +- [图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) - [视频口型替换-声动人像VideoRetalk](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) -- [图生表情包视频-表情包Emoji](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) -- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [视频风格重绘API参考](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) +- [图生表情包视频-表情包Emoji](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) +- [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - [Vidu-文生视频API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) - [Vidu-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) +- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-参考生视频 API 参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) -- [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - [万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) -- [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) +- [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-视频编辑API参考（2.1）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) -- [爱诗-视频超清API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) -- [爱诗-视频对口型API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) -- [爱诗-视频动作模仿API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md deleted file mode 100644 index c5ca2d6f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md +++ /dev/null @@ -1,75 +0,0 @@ -# Qwen API vs 应用调用 vs 托管智能体API对比 - -百炼平台为开发者提供了三种主要的 API 调用方式：直接调用 Qwen 系列大模型（Qwen API）、调用已在控制台编排好的应用（应用调用 API）、以及通过托管智能体运行时管理完整的智能体生命周期（Managed Agents API）。三者在抽象层级、使用复杂度和适用场景上差异显著，本文帮助开发者根据业务需求做出技术选型。 - -## 核心定位 - -- **Qwen API**：直接访问基础大模型能力，开发者完全掌控对话编排与工具集成。 -- **应用调用 API**：调用控制台已配置好的应用（智能体/工作流），平台负责模型选择、提示词和工具编排。 -- **Managed Agents API**：平台托管智能体全生命周期（会话、沙箱、工具执行、事件流），开发者通过 REST 管理资源。 - -## 关键维度对比 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -| --- | --- | --- | --- | -| **抽象层级** | 模型层（底层） | 应用层（中层） | 运行时层（高层） | -| **调用对象** | Qwen 系列模型 | 控制台编排的应用（APP ID） | 平台托管的 Agent 实例 | -| **兼容协议** | OpenAI / Anthropic / DashScope 原生 | OpenAI Responses / DashScope 原生 | 百炼专有 REST API | -| **Endpoint 示例** | `POST /chat/completions` | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST /api/v1/agentstudio/sessions/{id}/events` | -| **认证方式** | [API Key](../concepts/api-key.md) (DASHSCOPE_API_KEY) | [API Key](../concepts/api-key.md) (DASHSCOPE_API_KEY) | [API Key](../concepts/api-key.md)（Bearer [Token](../concepts/token.md)） | -| **必需标识** | model 名称 | APP ID（+ 可选 Workspace ID） | workspace_id + agent_id | -| **对话历史管理** | 调用方自行维护（Responses 接口除外） | OpenAI Responses 模式自动管理；DashScope 模式需自行维护 | 平台托管，通过 Session/Event 机制自动管理 | -| **工具/插件** | Responses 接口内置联网搜索、代码解释器、网页提取；其余需自行定义 | 由控制台应用配置决定，调用时无需关心 | Agent 配置挂载 Skill（zip 包）、Environment（沙箱） | -| **流式输出** | 支持 | 支持（stream=True） | SSE 事件流订阅 | -| **[异步调用](../concepts/async-invocation.md)** | 不支持 | 支持（background=True） | 原生异步：Session 状态机驱动 | -| **[多模态](../concepts/multimodal.md)** | 取决于具体模型能力 | 支持（OpenAI Responses 模式） | 支持（通过 File 资源挂载） | -| **沙箱/执行环境** | 无 | 无（平台内部处理） | 开发者可创建和管理 Environment | -| **版本控制** | 无（指定模型版本即可） | 无 | Agent 自动版本递增，Session 锁定创建时版本 | -| **SDK 兼容** | OpenAI SDK / Anthropic SDK / [DashScope SDK](../concepts/dashscope-sdk.md) | OpenAI SDK / [DashScope SDK](../concepts/dashscope-sdk.md) | 需直接 HTTP 调用或自封装 | -| **迁移成本** | 低（直接复用 OpenAI/Anthropic 代码） | 中（需先在控制台配置应用） | 高（专有 API，需学习资源模型） | - -## [计费](../concepts/billing.md)与配额 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -| --- | --- | --- | --- | -| **[计费](../concepts/billing.md)粒度** | [Token](../concepts/token.md) 用量（按模型计价） | [Token](../concepts/token.md) 用量（应用内模型调用） | Token 用量 + 可能的沙箱资源费用 | -| **文件配额** | 无 | 无 | 单文件 20MB，空间总量 100GB，保留 30 天 | - -## 适用场景建议 - -### 选择 Qwen API - -- 需要直接控制模型参数（temperature、top_p 等）进行精细调优 -- 已有 OpenAI/Anthropic 代码希望低成本迁移到百炼 -- 构建自定义 RAG、Agent 框架，需要底层模型能力 -- 对工具调用逻辑有完全自主的编排需求 - -### 选择应用调用 API - -- 已在百炼控制台完成应用编排（提示词、知识库、插件），希望快速集成到业务系统 -- 团队中有非开发角色负责应用配置，开发者只需调用 -- 需要工作流（多步骤串联）能力但不想自行编排 -- 希望通过 OpenAI SDK 兼容方式接入已编排好的应用 - -### 选择 Managed Agents API - -- 需要平台托管智能体完整生命周期（创建、会话、工具执行、文件管理） -- 有复杂的工具执行需求，需要安全沙箱环境 -- 需要细粒度的会话状态管理和事件流订阅 -- 构建多智能体协作系统，需要独立管理每个 Agent 的版本和配置 -- 希望将工具包（Skill）作为可复用资产跨智能体共享 - -## 选型决策路径 - -1. **是否已在控制台配置好应用？** 是 -> 应用调用 API（最快集成） -2. **是否需要平台托管工具执行沙箱和会话状态机？** 是 -> Managed Agents API -3. **是否需要直接访问模型底层能力并自行编排？** 是 -> Qwen API -4. **从 OpenAI/Anthropic 迁移？** 优先 Qwen API 的兼容接口，迁移成本最低 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md deleted file mode 100644 index b3a96d1b..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md +++ /dev/null @@ -1,70 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供应用评测和模型评测两套独立的评测体系，分别面向不同的评测对象和使用场景。应用评测聚焦于智能体应用和工作流应用的端到端输出质量，覆盖 RAG 链路的各个环节；模型评测则聚焦于大语言模型本身的推理能力，用于模型选型和基础能力基准测试。理解两者的定位差异，有助于开发者在不同阶段选择合适的评测工具。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -|---------|---------|---------| -| **评测对象** | 智能体应用、工作流应用（已发布） | 文本生成类大模型 | -| **核心目标** | 评估应用端到端输出质量，定位 RAG 链路问题 | 评估模型推理能力，辅助模型选型与调优验证 | -| **评测方式** | 自动评测（单应用/多应用横向）、手动评测 | 自定义评测（AI/规则/人工）、基线评测 | -| **评测集来源** | 基于知识库自动生成，或手动上传（xls/xlsx/jsonl） | 手动上传评测数据集（Prompt + Completion），或使用公开基线数据集 | -| **评估机制** | 新版：评估器（LLM/Code）+ 标签；旧版：内置评分模型 | 评测维度（大模型评估/规则评估/人工评估） | -| **自动评分方式** | LLM 评估器（语义理解）、Code 评估器（规则判断） | 裁判模型打分（数值型/分类型）、规则评估（ROUGE/BLEU/Cosine 等） | -| **人工标注** | 旧版手动评测；新版通过标签体系支持 | 人工评估维度（Pass/Fail 标注） | -| **横向对比能力** | 多应用横向评测（最多 8 个应用） | 排行榜（相同维度下对比多个模型） | -| **归因分析** | 支持（模型理解有误/重排不佳/检索无效/切片不完整/未获取知识） | 不支持链路归因，仅提供维度评分明细 | -| **前提条件** | 应用已发布、已配置知识库、已开通应用观测 | 无特殊前提，上传数据集即可 | -| **支持的评分模型** | 评测集生成和评估仅支持 qwen-max、qwen-plus | 裁判模型推荐千问-Max，可选其他模型 | -| **基线能力评测** | 不支持 | 支持（C-Eval、MMLU、ARC、GSM8K、BBH、HellaSwag） | -| **版本管理** | 新版评测集支持版本管理 | 不支持评测集版本管理 | -| **操作入口** | 控制台（应用管理模块） | 控制台（模型管理模块） | -| **API/SDK 支持** | 通过应用调用间接支持 | 不提供公开 API/SDK，仅控制台操作 | -| **地域限制** | 无特殊地域限制 | 基线评测仅北京地域可用 | - -## 评测数据与评分体系对比 - -| 对比项 | 应用评测 | 模型评测 | -|-------|---------|---------| -| **数据集类型** | 旧版：对话分析（xls）、知识问答（jsonl）；新版：智能体/工作流/自定义 | 评测数据集（Prompt + Completion）、推理结果集 | -| **评分输出** | 1-5 分制（正确率 = 得分 >= 4 的占比） | 自定义评分范围（如 0-5）+ 通过阈值 | -| **结果分析** | 总正确率、BadCase 分析、调优建议、RAG 评价 | 综合得分、通过率、分数分布、逐样本明细 | -| **评估器/维度上限** | 每个任务最多 10 个评估器 | 无明确数量限制 | - -## 计费对比 - -| 对比项 | 应用评测 | 模型评测 | -|-------|---------|---------| -| **费用构成** | 评测集生成 [Token](../concepts/token.md) + 评估模型 [Token](../concepts/token.md) | 被评测模型推理 [Token](../concepts/token.md) + 裁判模型评分 Token | -| **免费方式** | 评估器模型当前限时免费 | 规则评估无裁判模型费用；已部署调优模型不额外计费 | -| **成本优化** | 合理控制评测集规模 | 先小规模验证（50-100 条）→ 保存推理结果集复用 → 优先规则评估 | - -## 适用场景建议 - -**选择应用评测的场景：** - -- 已构建完整的 RAG 智能体应用，需要评估端到端回答质量 -- 需要定位问题环节（模型理解、检索、重排、切片、知识库内容） -- 对多个应用版本进行 A/B 对比，验证迭代效果 -- 知识库更新、Prompt 调整、检索策略变更后的回归测试 -- 需要将人工标注经验固化为自动化评估规则（评估器） - -**选择模型评测的场景：** - -- 项目初期进行模型选型，对比多个候选模型的基础能力 -- 使用公开基准数据集（C-Eval、MMLU 等）快速了解模型水平 -- 模型微调后验证调优效果 -- 翻译、摘要、NL2SQL 等有确定性评判标准的任务，优先用规则评估降低成本 -- 需要在排行榜上持续跟踪模型表现 - -**组合使用建议：** - -在实际项目中，两种评测往往互补使用。典型流程是先通过模型评测筛选出基础能力最优的候选模型，再将其集成到应用中，通过应用评测验证端到端效果并持续优化 RAG 链路。模型评测解决"哪个模型更好"的问题，应用评测解决"应用整体表现如何优化"的问题。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-evaluation-vs-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-evaluation-vs-monitoring.md deleted file mode 100644 index 6adae0c7..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-evaluation-vs-monitoring.md +++ /dev/null @@ -1,65 +0,0 @@ -# 应用评估与应用监控对比 - -阿里云百炼平台在应用上线运营的不同阶段提供了两类互补能力：**应用评估**负责在质量维度上系统化地衡量并调优应用输出，**应用监控（应用观测）**负责端到端追踪线上应用的调用链路与运行指标。两者常被同时使用，但目标、数据来源和使用方式差异明显。本文从技术选型角度对两者做对比，帮助开发者理清各自定位与协同关系。 - -## 核心定位 - -- **应用评估**：面向**质量**。通过评测集、评估器、标签构建多维度评测闭环，回答「应用回答得好不好、哪里出了问题、如何优化」。 -- **应用监控（应用观测）**：面向**运行时可观测性**。追踪应用内部调用链路、延时与 Token 消耗，回答「应用跑起来发生了什么、慢在哪里、花了多少 Token」。 - -## 关键维度对比 - -| 维度 | 应用评估 | 应用监控（应用观测） | -|------|----------|----------------------| -| 主要目标 | 系统化衡量并优化应用输出质量 | 端到端追踪线上运行状态与性能指标 | -| 支持应用类型 | 智能体应用、工作流应用（自动评测仅面向已发布且已配置知识库的智能体应用） | 智能体应用、工作流应用、高代码应用（暂不支持 Assistant API 创建的智能体；高代码仅观测入口 CHAIN 节点） | -| 数据来源 | 评测集（自动生成 / 手动上传 / 从应用观测导入） | 线上真实调用（自动追踪，分钟级同步） | -| 输入格式 | 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）；新版：智能体/工作流/自定义评测集 | 无需上传，应用发布并添加到观测列表后自动采集 | -| 输出格式 | 评测报告（总正确率、BadCase 分析、归因分析、调优建议） | Trace 列表、监控统计图表；可导出 JSONL / EXCEL | -| 评估/统计方式 | 自动评测（LLM 评估器）+ 手动评测（人工标注）+ Code 评估器 | 调用次数/失败率、Token 量、平均首 Token 耗时、平均调用时长等性能指标 | -| 评测/观测粒度 | 应用级别的回答质量、单应用或最多 8 个应用横向对比 | 调用链路节点级别（CHAIN/AGENT/RETRIEVER/LLM/TOOL 等），支持 Root/All/Model Span | -| 是否有 API | 有评测任务相关能力（评测过程调用大模型） | **无 API**，仅可通过控制台操作 | -| 数据留存 | 评测集与报告持久化，评测集支持版本管理 | 调用记录最长保留 30 天 | -| 计费方式 | 评测调用大模型产生 Token 费用；Code 评估器无额外费用 | 功能本身不收费，观测数据存储由 OpenTelemetry 服务收费 | -| 前提条件 | 应用已发布 + 已配置知识库 + 开通应用观测 + 相应权限（管理员/应用评测-操作） | 完成应用观测配置（授权链路角色、开通 OpenTelemetry、初始化 LogStore）+ 相应权限 | -| 典型场景 | 版本迭代验证、模型/Prompt 调优、多应用选型、回归评测 | 线上问题排查、性能优化、成本分析、调用链路诊断 | - -## 组件与能力差异 - -### 应用评估的核心组件 - -- **评测模式**：自动评测（大模型基于知识库自动生成评测集并评分，支持单应用与最多 8 个应用横向评测）与手动评测（人工构建评测集并打分，属旧版能力）。 -- **评测集**：旧版分对话分析/知识问答两类；新版分智能体/工作流/自定义三类，支持版本管理与从应用观测导入。 -- **评估器**：预置模板、LLM 评估器、Code 评估器，或基于历史评测任务抽象生成；每个评测任务最多 10 个评估器，建议组合 3-5 个。 -- **归因分析**：将 BadCase 定位到 RAG 具体环节（模型理解有误、重排不佳、检索无效、切片不完整、未获取知识）并给出优化方向。 - -### 应用监控的核心能力 - -- **调用链路追踪**：按节点类型（CHAIN、AGENT、RETRIEVER、REWRITER、EMBEDDING、RERANKER、LLM、TOOL、GUARDRAIL 及工作流专属节点）展示嵌套调用关系。 -- **监控统计**：调用次数与失败率、Token 总量、平均单次请求 Token、平均首 Token 耗时、平均调用时长，支持分钟/小时/天聚合，最长 30 天。 -- **数据筛选与标注**：Span 筛选模式与多条件过滤器；支持对 Span 打标签（布尔/分类/数字/文本）。 -- **数据导出与回流**：导出 JSONL/EXCEL，并可将线上 Span 数据直接加入评测集（最多 50 个字段映射）。 - -## 两者的协同关系 - -应用评估与应用监控并非二选一，而是形成闭环： - -1. **观测提供样本**：应用监控采集的真实线上调用可直接导入评测集，让评测更贴近生产分布。 -2. **共享标签体系**：两者的标签管理统一，观测阶段的标注可复用到评测任务，反之亦然。 -3. **评测的前提依赖观测**：自动评测要求应用已开通应用观测并加入观测列表。 -4. **优化闭环**：监控发现异常（如延时高、失败率上升）→ 评估定位质量问题与归因 → 实施优化 → 再次观测与评测确认改进。 - -## 技术选型建议 - -- **要衡量「答得好不好」并做调优**：选择**应用评估**。尤其在知识库更新、Prompt 调整、模型升级、检索/重排策略变更后，或需要多应用/多版本选型时。 -- **要看「跑得怎么样、慢在哪、花多少 Token」**：选择**应用监控**。用于线上性能诊断、成本分析和调用链路排查。 -- **高代码应用**：只能用应用监控（且仅观测入口 CHAIN 节点），当前不在自动评测的支持范围内。 -- **需要程序化集成**：应用监控无 API，只能控制台操作；若强依赖自动化上报，需结合 AgentScope-AI 的 Tracing 模块与部署时的 `--telemetry enable` 参数。 -- **最佳实践**：先用应用监控开启持续观测积累真实调用数据，再用应用评估以这些数据为样本建立定期回归评测机制，形成「观测 → 评测 → 优化 → 再观测」的持续改进闭环。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [application monitoring](../guides/application-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md deleted file mode 100644 index c8136a11..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md +++ /dev/null @@ -1,70 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两种互补的可观测性能力：**应用观测**和**模型监控**。应用观测聚焦于应用内部的端到端调用链路追踪，帮助开发者理解智能体应用、工作流应用的执行过程；模型监控则聚焦于模型调用层面的运行状态与成本管理，提供用量统计、性能指标、日志审计和主动告警。两者分别从"应用维度"和"模型维度"保障系统的可观测性，开发者通常需要同时使用。 - -## 关键维度对比 - -| 对比维度 | 应用观测 | 模型监控 | -| --- | --- | --- | -| **监控对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（大语言模型、视觉模型、语音模型、向量模型等） | -| **核心目标** | 追踪应用内部调用链路，优化运营效果与成本 | 监控模型运行状态与用量，保障稳定性与成本可控 | -| **数据维度** | 按应用 + [业务空间](../concepts/workspace.md) | 按模型 + [业务空间](../concepts/workspace.md) | -| **数据刷新频率** | 分钟级 | 普通监控：小时级；高级监控：分钟级 | -| **数据保留期** | 最长 30 天 | 普通用量最长 30 天，更早需查账单 | -| **支持的应用/模型范围** | 智能体应用、工作流应用、高代码应用（不支持 Assistant API 创建的智能体应用） | 所有模型均支持用量统计；普通监控支持全地域全模型；高级监控限北京、新加坡、弗吉尼亚 | -| **链路追踪** | 支持（CHAIN、AGENT、LLM、RETRIEVER、TOOL 等多节点类型，可展开查看嵌套调用） | 不支持应用级链路追踪，仅记录单次模型调用 | -| **关键指标** | 调用次数/失败率、[Token](../concepts/token.md) 总量（输入/输出）、平均单次请求 [Token](../concepts/token.md) 量、平均首 [Token](../concepts/token.md) 耗时、平均调用时长 | 调用总量/失败量/失败率、平均调用时长、平均首包时长、RPM、TPM、限流错误次数(429)、内容安全错误次数 | -| **监控分类** | 无分类，统一展示性能与调用指标 | 四类：安全、成本、性能、错误 | -| **告警能力** | 不支持 | 支持主动告警（仅北京、新加坡地域），可设置超时、Token 消耗突增等阈值 | -| **日志/历史对话** | 支持查看 Prompt 内容、输出、延时等完整调用记录 | 支持推理日志和历史对话记录（仅北京地域部分模型，需手动开通） | -| **Token 消耗追踪** | 按 Span 节点记录 Token 消耗 | 三层管理：汇总统计、单次调用追踪、阈值告警 | -| **数据筛选** | 按状态、Span Name、输入/输出内容、延时、Token 量、标签等多维筛选；支持 Request ID / Trace ID / Span ID 检索 | 按 API-KEY、推理类型、时间范围、时间精度筛选 | -| **数据标注** | 支持（布尔值/分类/数字/文本四种标签类型，与评测共享） | 不支持 | -| **导入评测集** | 支持将 Span 数据直接加入评测集作为评测样本 | 不支持 | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 不支持直接导出，需通过费用与成本页面查询 | -| **费用管理** | 不提供费用管理功能 | 提供费用概览、账单趋势、免费额度管理及用完即停开关 | -| **计费** | 功能本身免费，OpenTelemetry 存储费用另计 | 功能本身免费，高级监控可能涉及额外费用 | -| **开通方式** | 控制台手动开通（授权 OpenTelemetry 服务角色 + 开通服务 + 初始化 LogStore） | 系统自动采集（普通监控）；高级监控和日志需手动开通 | -| **API 支持** | 无 API，仅控制台操作 | 无专用 API，仅控制台操作 | -| **地域限制** | 无明确地域限制 | 普通监控无限制；高级监控限北京/新加坡/弗吉尼亚；告警限北京/新加坡；日志限北京 | - -## 适用场景建议 - -### 应用观测适用于 - -- **调试应用内部逻辑**：需要查看智能体应用或工作流应用的完整调用链路，定位某个节点（如检索、模型推理、插件调用）的异常或性能瓶颈。 -- **优化 RAG 效果**：通过 RETRIEVER、REWRITER、RERANKER 等节点的详细数据，分析检索召回质量和排序效果。 -- **构建评测数据集**：将线上真实调用数据标注后导入评测集，用于持续优化应用效果。 -- **应用级性能分析**：关注单个应用的整体调用时长、Token 消耗趋势，评估应用的运营效率。 - -### 模型监控适用于 - -- **模型稳定性保障**：监控模型调用的失败率、限流错误(429)、内容安全错误等，及时发现异常。 -- **成本核算与控制**：按模型、按[业务空间](../concepts/workspace.md)统计 Token/图片/视频的用量与费用，管理免费额度。 -- **主动告警**：对关键模型设置超时、Token 消耗突增等告警规则，防止静默失败。 -- **合规审计**：通过推理日志记录每次模型调用的输入与输出，满足内容审计需求。 -- **多模型对比**：在监控列表中横向对比不同模型的性能与错误率，辅助模型选型。 - -### 建议同时使用的场景 - -当应用上线后需要全面保障服务质量时，建议同时开启两项能力：用应用观测定位"哪个环节出了问题"，用模型监控回答"模型本身是否正常、成本是否可控"。例如，当应用观测发现某次调用的 LLM 节点延时异常时，可切换到模型监控确认该模型是否存在全局性的性能劣化或限流。 - -## 技术选型参考 - -| 选型考量 | 推荐方案 | -| --- | --- | -| 需要追踪应用内部多节点调用链路 | 应用观测 | -| 需要监控模型全局运行状态和失败率 | 模型监控 | -| 需要对异常指标设置主动告警 | 模型监控 | -| 需要将线上数据导入评测集 | 应用观测 | -| 需要按模型维度统计费用和用量 | 模型监控 | -| 需要对调用数据打标签做质量分析 | 应用观测 | -| 需要审计模型调用的输入输出内容 | 模型监控（推理日志） | -| 需要端到端的可观测性保障 | 两者配合使用 | - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md deleted file mode 100644 index 02011a75..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md +++ /dev/null @@ -1,64 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供两套独立的评测体系：**应用评测**面向智能体应用和工作流应用的端到端输出质量评估，**模型评测**面向文本生成类模型的基础能力评估。两者在评测对象、数据集格式、评分机制和使用场景上存在本质差异，开发者需根据自身需求选择合适的评测路径。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -|---------|---------|---------| -| 评测对象 | 智能体应用、工作流应用 | 文本生成类模型（含调优模型） | -| 核心目标 | 评估应用端到端输出质量 | 评估模型推理能力，辅助选型或验证调优效果 | -| 评测方式 | 自动评测（大模型生成评测集并评分）、手动评测（人工逐条标注） | 自定义评测（自有数据集 + 自定义维度）、基线评测（公开标准数据集） | -| 评测集格式 | 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）；新版：按应用出入参自动生成模板 | 统一格式：Prompt（用户问题）+ Completion（参考答案）两列 | -| 评测集创建 | 支持自动生成（知识问答类型，限 qwen-max/qwen-plus）和手动上传 | 在数据管理模块上传评测集（EvaluationSet）类型数据 | -| 评分机制 | 评估器体系：LLM 评估器（语义理解）、Code 评估器（规则判断）、预置模板 | 评测维度体系：大模型评估（数值型/分类型）、规则评估（文本相似度/字符串匹配）、人工评估 | -| 评分配置单元 | 评估器（最多 10 个/任务，建议 3-5 个组合） | 评测维度（按需配置，类型创建后不可修改） | -| 裁判模型 | 由评估器内部选择模型 | 推荐千问-Max | -| 规则评分算法 | Code 评估器自定义 Python 函数 | 内置 7 种算法（ROUGE-1/2/L、BLEU、Cosine、Fuzzy Match、Accuracy）+ 字符串匹配 | -| 人工标注 | 标签系统（分类/布尔值/数字/文本四种类型），支持快速标注模式 | 人工评估维度（Pass/Fail 分类型） | -| 排行榜 | 不支持 | 支持（相同维度下横向对比多模型，得分 0-100） | -| 横向对比 | 自动评测最多同时评测 8 个应用 | 通过排行榜对比多个模型 | -| 结果下载 | 支持 | 支持（基线评测除外） | -| 基线评测 | 不支持 | 支持（C-Eval、MMLU、ARC、GSM8K、BBH、HellaSwag），仅北京地域 | -| 版本管理 | 新版/旧版两套界面并存 | 统一界面 | -| 操作方式 | 控制台 | 仅控制台（无公开 API/SDK） | -| 前置要求 | 自动评测需已发布应用并配置知识库，需开通应用观测 | 无特殊前置，上传数据集即可 | - -## 评分体系差异 - -应用评测和模型评测虽然都支持"大模型评分"和"规则评分"，但实现方式不同： - -- **应用评测**的评估器是独立可复用的组件，同一评估器可在不同评测任务间共享。LLM 评估器和 Code 评估器各有优势，建议组合使用以覆盖语义理解和精确规则两个层面。 -- **模型评测**的评测维度在创建时即绑定评分方式，类型不可修改。规则评估提供开箱即用的标准算法（ROUGE、BLEU 等），无需编写代码，适合翻译、摘要等有确定性标准的场景。 - -## 计费差异 - -两套评测体系的费用结构相似，均包含推理费用和评分费用，但细节有所不同： - -- **应用评测**：调用大模型的 Token 费用正常计费，Code 评估器无额外费用。 -- **模型评测**：被评测模型推理费用按 Token 计费（使用推理结果集时不产生），大模型评估维度额外产生裁判模型费用，规则评估和人工评估无裁判模型费用。已部署的调优模型评测不额外计费。 - -两者共同的成本优化策略：先小规模验证（50-100 条），确认配置无误后再扩大规模。模型评测还支持保存推理结果集复用，避免重复推理。 - -## 适用场景建议 - -**选择应用评测**： -- 已构建完整的智能体或工作流应用，需要评估端到端输出质量 -- 需要领域专家介入进行人工标注和多维度质量评估 -- 需要结合应用观测进行线上真实数据的持续质量监控 -- 评测关注点是应用整体表现而非底层模型能力 - -**选择模型评测**： -- 处于技术选型阶段，需要在多个候选模型间做横向对比 -- 对模型进行了微调/调优，需要量化对比调优前后的能力变化 -- 需要使用公开基准数据集（C-Eval、MMLU 等）快速了解模型基础能力 -- 评测关注点是模型本身的推理和生成能力 - -**组合使用**：在实际项目中，建议先通过模型评测选定基础模型，再通过应用评测验证集成到应用后的端到端效果，形成"模型选型 → 应用构建 → 应用评测 → 持续监控"的完整闭环。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md deleted file mode 100644 index 291634a5..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md +++ /dev/null @@ -1,66 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的可观测能力：**应用观测**聚焦应用内部调用链路的端到端追踪，帮助开发者理解智能体、工作流等应用的执行过程；**模型监控**则聚焦模型维度的运行指标与成本核算，保障模型调用的稳定性与经济性。两者观测粒度、数据来源和使用场景各有侧重，开发者需要根据排查目标选择合适的工具，也可配合使用以获得从应用到模型的全链路可观测性。 - -## 关键维度对比 - -| 维度 | 应用观测 | 模型监控 | -| --- | --- | --- | -| **观测对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（所有模型，含调优后的自定义模型） | -| **核心目标** | 追踪应用内部调用链路，定位延时瓶颈与逻辑问题 | 监控模型运行状态与用量，保障稳定性与控制成本 | -| **数据粒度** | 单次请求的完整 Trace / Span 链路 | 按"模型 + [业务空间](../concepts/workspace.md)"维度的聚合指标 | -| **数据更新频率** | 分钟级 | 普通监控小时级，高级监控分钟级 | -| **数据保留时长** | 最长 30 天调用记录 | 用量统计保留 30 天，更早数据需到费用与成本页面查询 | -| **关键指标** | 延时、Token 量（输入/输出）、首 Token 耗时、调用次数、失败率 | 调用总量、失败率、平均调用时长、首包时长、RPM、TPM、Token 消耗 | -| **指标分类** | 按节点类型（CHAIN、LLM、RETRIEVER 等）查看 | 按安全、成本、性能、错误四类分类查看 | -| **链路追踪** | 支持，可展开查看完整 Span 树及每个节点的输入输出 | 不支持链路追踪，仅提供模型级聚合数据 | -| **日志能力** | Trace 详情中可查看 Prompt 内容与模型输出 | 需额外开通推理日志，仅北京地域部分模型支持 | -| **告警能力** | 不支持 | 支持主动告警（仅北京、新加坡地域） | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 不支持直接导出（需通过费用与成本页面获取账单数据） | -| **数据标注** | 支持对 Span 数据添加标签（布尔值/分类/数字/文本） | 不支持 | -| **与评测集联动** | 支持将 Span 数据直接加入评测集 | 不支持 | -| **地域限制** | 无特殊地域限制 | 高级监控限北京/新加坡/弗吉尼亚，告警限北京/新加坡，日志限北京 | -| **费用** | 功能免费，OpenTelemetry 存储另计 | 功能免费，高级监控的底层存储可能产生费用 | -| **开通方式** | 需手动配置：授权服务角色 → 开通 OpenTelemetry → 初始化 LogStore | 普通监控自动采集；高级监控和日志需手动开通 | -| **操作方式** | 仅控制台，无 API | 仅控制台，无 API | -| **用量/计费统计** | 不涉及计费统计 | 提供按模型类型的用量统计（Token/张/秒），可用于成本核算 | - -## 适用场景建议 - -### 应用观测适用于 - -- **应用调试与优化**：需要查看智能体或工作流应用内部各环节（检索、重写、模型推理、插件调用等）的执行顺序与耗时，定位性能瓶颈。 -- **Prompt 审查**：需要查看每次调用的具体 Prompt 内容与模型输出，排查回答质量问题。 -- **评测数据积累**：将线上真实调用数据标注后加入评测集，用于持续改进应用效果。 -- **单次请求排障**：通过 Request ID / Trace ID / Span ID 精确定位某一次调用的异常环节。 - -### 模型监控适用于 - -- **模型稳定性保障**：监控模型调用失败率、限流错误（429）、内容安全错误等指标，及时发现服务异常。 -- **成本管理**：按[业务空间](../concepts/workspace.md)和模型维度统计 Token 消耗和调用量，核算各模型的使用成本。 -- **容量规划**：通过 RPM（每分钟请求数）、TPM（每分钟 Token 数）等指标评估负载水平，预判扩容需求。 -- **主动告警**：设置 Token 消耗或错误率阈值，在异常发生时第一时间收到通知，避免静默故障。 -- **合规审计**：开通推理日志后可回溯历史对话的输入输出，满足内容审计需求。 - -### 建议组合使用 - -在生产环境中，建议同时开启应用观测和模型监控：应用观测负责应用层面的链路追踪与调试，模型监控负责基础设施层面的稳定性与成本管理。当模型监控发现某模型失败率上升时，可在应用观测中按时间范围筛选相关 Trace，进一步定位是应用逻辑问题还是模型服务问题。 - -## 技术选型参考 - -| 排查目标 | 推荐工具 | -| --- | --- | -| 某次调用为什么返回了错误答案 | 应用观测（查看 Trace 中各节点的输入输出） | -| 某个模型最近的失败率是否正常 | 模型监控（查看错误类指标） | -| 应用整体响应变慢，瓶颈在哪个环节 | 应用观测（对比各 Span 延时） | -| 本月 Token 消耗是否超出预算 | 模型监控（查看用量统计与费用概览） | -| 需要对线上 bad case 做标注并加入评测集 | 应用观测（数据标注 + 添加到评测集） | -| 模型调用量突增需要告警 | 模型监控（配置告警规则） | -| 回溯某次模型调用的完整输入输出 | 模型监控（推理日志，仅北京地域部分模型） | - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-apis-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-apis-comparison.md new file mode 100644 index 00000000..9bbe4b9a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-apis-comparison.md @@ -0,0 +1,83 @@ +# 应用层核心 API 对比：Application Calling、Managed Agents 与 Application Component + +为帮助开发者在百炼平台构建智能应用时做出清晰、可靠的技术选型，本文系统对比三大应用层核心 API 能力：**Application Calling**（应用调用）、**Managed Agents**（托管智能体）与 **Application Component**（应用组件）。三者定位不同——Application Calling 面向“已发布应用的端到端执行”，Managed Agents 面向“可编程、可编排的智能体运行时”，Application Component 则聚焦“数据与知识基础设施的原子化管理”。理解其差异是避免架构错配、提升开发效率与生产稳定性的关键前提。 + +--- + +## 关键维度对比 + +| 维度 | Application Calling | Managed Agents | Application Component | +|------|---------------------|----------------|------------------------| +| **核心定位** | 调用已发布、已配置完成的智能体或工作流应用（黑盒执行） | 构建、部署、运行可版本化、带沙箱与工具能力的智能体实例（灰盒运行时） | 管理应用底层数据资产与知识基础设施（白盒数据/知识/提示工程能力） | +| **输入格式** | • 字符串（单轮文本）
• 消息数组（`role`/`content`/`type`），支持 `text`/`imageList`/`input_file`（仅智能体）
• `biz_params` 透传业务参数 | • 严格遵循事件驱动消息结构：
 `{"role": "user", "type": "user_message", "content": [...]}`
• `content` 支持文本、文件引用（`file_id`）、结构化工具调用结果
• 所有输入需经 Session 绑定的 Agent + Environment 处理 | • 按模块分离：
 – 数据连接：`AddFile`（二进制+元数据）、`AddTable`（结构化 schema）
 – 知识库：`SubmitIndexJob`（文档路径列表）、`Retrieve`（query + filter）
 – Prompt：`CreatePromptTemplate`（含 `${variable}` 占位符的字符串） | +| **输出格式** | • 同步：JSON 响应体含 `output.text` / `output.images` / `output.files`；流式响应为 SSE 或 chunked JSON
• 异步：返回 `task_id`，后续 `GET /tasks/{id}` 获取最终结果（非流式） | • 全事件流（SSE）：
 `session_status`（`idle`/`running`/`terminated`）
 `message`（含 `role`/`content`/`tool_calls`）
 `tool_result`（工具执行返回）
• 无传统“终态响应体”，需客户端聚合事件流 | +| **支持模型** | • 智能体/工作流所绑定的任意百炼模型（Qwen-VL、Qwen-Max、Qwen-Plus、Qwen2 等）
• 模型选择在应用创建/编辑阶段完成，API 调用时不指定 | • 仅支持百炼托管模型（当前明确支持 `qwen-plus` 等有限 ID）
• 模型通过 `Agent.model.id` 显式声明，不可动态切换或自定义接入 | • **不直接调用大模型**
• 为上层应用（如 Application Calling 中的智能体）提供数据源与知识支撑 | +| **API 端点（典型）** | • Responses API（OpenAI 兼容）：
 `POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`
• DashScope API（原生）：
 `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` | • 统一基础路径：
 `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`
• 资源路由示例：
 `POST /agents` / `POST /environments` / `POST /sessions/{id}/events` | • ROA 风格，按能力域划分：
 数据连接：`POST /bailian/2023-12-29/data-connection/files`
 知识库：`POST /bailian/2023-12-29/knowledge-base/indexes`
 Prompt：`POST /bailian/2023-12-29/prompt-engineering/templates` | +| **认证方式** | • APP ID + Workspace ID（部分地域必需） + DashScope API Key | • Workspace ID + Region + DashScope API Key（统一鉴权） | • Workspace ID + RAM AccessKey（AK/SK） + ROA 签名（需 SDK 或手动实现） | +| **计费方式** | • 按调用次数 + 模型 token 消耗计费（同底层模型计费规则）
• [异步任务](../concepts/asynchronous-task.md)按实际执行时长与资源占用折算为等效 token 计费 | • 按 Session 运行时长（秒级） + 工具调用次数 + 文件处理量计费
• Agent/Environment 创建、File/Skill 上传等管理操作免费 | • 按资源使用量计费：
 – 文件存储（GB/月）
 – 知识库索引构建与查询（QPS + 文档页数）
 – Prompt 模板调用量（次） | +| **会话状态管理** | • DashScope API：通过 `session_id`（有效期 1 小时）维护上下文
• Responses API：**不支持自动会话管理**，需显式传递完整历史消息数组 | • 内置全生命周期会话状态机（`idle` → `running` → `idle`/`terminated`）
• 状态变更通过 SSE 实时推送，客户端无需维护状态快照 | • **无会话概念**
• 所有接口均为无状态 RESTful 调用，状态由业务侧自行管理（如缓存 `IndexId`） | +| **多模态支持** | • ✅ 图像输入（需 VL 模型 + 应用配置）
• ✅ 文件输入（仅智能体应用，支持全文引用/切片检索） | • ✅ 图像/音频作为 `content` 项上传（需先 `POST /files` 审核）
• ✅ 文件可挂载至沙箱供工具读写 | • ✅ 文件上传（`AddFile` 支持 PDF/DOCX/PNG/JPG 等）
• ❌ 不直接处理图像语义，仅作存储与索引源 | +| **典型场景** | • 客服对话机器人（Web/App 接入）
• 工作流自动化（审批流、报告生成）
• 第三方系统集成（ERP/CRM 触发智能分析） | • 需深度定制执行逻辑的智能体（如多步骤工具协同、复杂错误恢复）
• 高隔离性需求场景（金融合规沙箱、客户专属环境）
• 需细粒度观测与调试的 AI 工程化项目 | • 构建企业级知识库（产品文档、客服知识）
• 管理多源异构数据（数据库连接、Excel 表格、非结构化文件）
• 标准化 Prompt 模板库（营销文案生成、代码解释） | + +--- + +## 适用场景建议 + +### ✅ 优先选用 **Application Calling** +- 你的应用已在百炼控制台完成开发、测试与发布，只需“调用”而非“重构”； +- 场景对实时性要求高（如在线客服），且输入以文本/图像为主，无需复杂工具链； +- 团队熟悉 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，希望最小成本迁移现有 SDK 代码； +- 业务逻辑相对稳定，无需频繁变更执行流程或沙箱环境。 + +### ✅ 优先选用 **Managed Agents** +- 你需要完全掌控智能体的执行过程：例如插入自定义日志、拦截工具调用、实现重试/回滚策略； +- 应用涉及敏感操作（如调用内部 API、读写客户数据库），必须运行在隔离沙箱中； +- 智能体需组合多个 Skill（如“查天气”+“订机票”+“发邮件”），且各 Skill 版本需独立演进； +- 项目处于 AI 工程化探索期，需要可观测事件流（`tool_result`、`session_status`）辅助调试与监控。 + +### ✅ 优先选用 **Application Component** +- 你正在搭建应用的数据底座：统一纳管客户资料、产品手册、销售合同等非结构化/结构化数据； +- 需要构建可复用、可审计的知识服务（如 `Retrieve` 接口供多个前端调用）； +- Prompt 设计已成为团队标准实践，需集中管理、AB 测试与灰度发布模板； +- 当前使用 Application Calling 或 Managed Agents 时，频繁遇到“知识更新滞后”“数据源分散”“提示词散落各处”等问题。 + +> ⚠️ **重要提醒**：三者并非互斥，而是典型的**分层协作关系**。 +> **最佳实践架构**： +> `Application Component`（提供知识库 Index + 数据连接 File） +> ↓（作为数据源注入） +> `Managed Agents`（构建具备工具能力的智能体，从知识库检索并调用业务 API） +> ↓（封装为标准化应用） +> `Application Calling`（供前端/第三方系统一键调用） + +--- + +## 技术选型决策树（面向开发者） + +```mermaid +graph TD + A[你的核心目标是什么？] --> B{是否在调用一个
已发布、功能完备的应用？} + B -->|是| C[✅ Application Calling
→ 快速集成，开箱即用] + B -->|否| D{是否需要构建一个
可编程、可沙箱、可事件追踪的智能体？} + D -->|是| E[✅ Managed Agents
→ 精细控制执行，强工程化] + D -->|否| F{是否在建设
数据/知识/提示等基础设施？} + F -->|是| G[✅ Application Component
→ 原子化管理，支撑上层] + F -->|否| H[请重新审视需求边界：
• 是否混淆了“能力提供”与“能力调用”？
• 是否遗漏了数据准备环节？] +``` + +**补充判断要点**： +- 若需 **跨地域部署**：Application Calling 和 Managed Agents 当前均**仅支持华北2（北京）**；Application Component 支持多地域接入点（需显式配置 endpoint）。 +- 若需 **最小权限管控**：Application Component 依赖 RAM 精细策略（如 `sfm:Retrieve`）；Application Calling 与 Managed Agents 使用统一 DashScope API Key，权限粒度较粗。 +- 若需 **流式响应**：Application Calling（同步+stream=true）与 Managed Agents（SSE）均支持；Application Component 的 `Retrieve` 为同步 JSON 响应，不支持流式。 +- 若存在 **长期会话需求**（>1 小时）：Application Calling 的 `session_id` 会过期，需业务侧续期；Managed Agents 的 Session 可长期保持 `idle` 状态，支持更灵活的保活策略。 + +--- +*最后更新：2024年6月* +*适用百炼平台 v2.3.x 及以上版本* + +## 被对比主题页 + +- [application call](../api/application-call.md) +- [managed agents api](../api/managed-agents-api.md) +- [application component api reference](../api/application-component-api-reference.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md deleted file mode 100644 index b2325ef4..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md +++ /dev/null @@ -1,45 +0,0 @@ -# 应用调用方式对比 - -阿里云百炼的应用（智能体应用、[工作流](../concepts/workflow.md)应用、新版智能体 Agent 2.0）可通过 API 集成到业务系统中。官方文档中存在两篇高度相关的主题页：一篇偏 **API 参考**（`application call`），重点介绍两套调用 API（OpenAI 兼容 Responses API 与 DashScope 原生 `/completion`）的端点、参数与同步/异步模式；另一篇偏 **使用指南**（`bailian application calling`），重点介绍通过 DashScope SDK / HTTP 调用智能体与[工作流](../concepts/workflow.md)应用的实操步骤、多轮对话与自定义参数透传。本文对两者做维度对比，帮助开发者在技术选型时快速定位所需信息。 - -## 关键维度对比 - -| 维度 | [application call](../api/application-call.md)（API 参考） | bailian [application call](../api/application-call.md)ing（使用指南） | -| --- | --- | --- | -| 文档定位 | API 参考，强调端点、参数、调用模式 | 使用指南，强调 SDK 实操与场景示例 | -| 所属分类 | api | guides | -| 覆盖的 API 模式 | 两套：OpenAI 兼容 Responses API + DashScope 原生 `/completion` | 一套：DashScope 原生 `/completion`（`Application.call` / `POST /apps/{app_id}/completion`） | -| 主 Endpoint | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`（OpenAI 兼容）
`POST /api/v1/apps/{APP_ID}/completion`（DashScope 原生） | `POST /api/v1/apps/{APP_ID}/completion` | -| SDK 推荐 | OpenAI 兼容模式用 OpenAI SDK；DashScope 原生模式用 DashScope SDK | DashScope SDK（Python / Java），Node.js 用 `axios` | -| 支持应用类型 | 智能体、[工作流](../concepts/workflow.md)、新版智能体 Agent 2.0 | 智能体应用、工作流应用（智能体编排应用已被工作流应用替代） | -| 输入格式 | `input` 为 string 或 messages 数组；多模态支持 `input_text` / `input_image` / `input_file` | `input.prompt` 字符串 或 自行管理 `messages` 数组 | -| 多轮对话 | 通过 messages 数组传递完整对话历史；`pre_response_id` / `conversation_id` 上下文后续支持 | 两种方式：`session_id`（云端托管，1 小时 / 50 轮）或自行管理 `messages`（推荐） | -| 同步/异步 | 支持 `stream` 流式、`background` 异步；异步暂不支持流式 | 默认同步；多轮对话通过 `session_id` 或 `messages` 实现 | -| 多模态 | 显式支持图像、文件（`input_image` / `input_file`，文件仅智能体应用支持） | 未专门展开 | -| 自定义参数透传 | 未展开 | 支持 `biz_params.user_defined_params` 透传业务参数到自定义插件 / 工作流插件节点 | -| 业务空间 | 明确说明子业务空间需 Workspace ID，多地域 Base URL 含 Workspace ID | 提及业务空间对插件与应用关联的约束（同一业务空间内） | -| 地域说明 | 明确给出华北2（北京）默认 Endpoint，并列出德国、新加坡、日本等地域需带 Workspace ID | 未专门说明地域差异 | -| 典型代码示例 | Python（OpenAI SDK 同步多轮） | Python / Java / Node.js / curl（基础调用） | -| 凭证准备 | APP ID + Workspace ID + API Key + SDK | API Key + APP_ID + DashScope SDK | - -## 适用场景建议 - -- **选 `application call`（API 参考）**：当你需要复用现有 OpenAI 生态代码库与工具链；需要使用[流式输出](../concepts/streaming-output.md)或异步执行；需要调用新版智能体 Agent 2.0；需要多模态输入（图像、文件）；或部署在非北京地域需要明确 Workspace ID 与 Base URL 拼接规则时。 -- **选 `bailian application calling`（使用指南）**：当你首次接入百炼应用、需要 Python / Java / curl 的最小可运行示例；需要通过 `session_id` 实现云端托管多轮对话；需要向自定义插件或工作流插件节点透传业务参数（`biz_params.user_defined_params`）；或团队已习惯使用 DashScope SDK 时。 - -## 技术选型参考 - -两篇文档并非互斥，而是互补：`application call` 给出"调哪套 API、用什么端点、传哪些字段"的契约层信息，`bailian application calling` 给出"用哪个 SDK、怎么写代码、怎么传业务参数"的实操层信息。建议的选型路径： - -1. 先读 `application call` 确定调用模式（OpenAI 兼容 vs DashScope 原生），明确端点与参数契约； -2. 若选择 DashScope 原生 `/completion`，再读 `bailian application calling` 获取多语言 SDK 示例与多轮、自定义参数等进阶能力； -3. 若选择 OpenAI 兼容 Responses API，则以 `application call` 为主，参考 OpenAI SDK 既有用法，`bailian application calling` 中的 `session_id` / `biz_params` 等能力在该模式下暂不适用。 - -> 注：两篇文档对应用类型的命名略有差异（"新版智能体 Agent 2.0" vs "智能体编排应用已被工作流应用替代"），接入前请以控制台实际应用类型与最新 API 参考为准。 - -## 被对比主题页 - -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md deleted file mode 100644 index c3c897e5..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md +++ /dev/null @@ -1,57 +0,0 @@ -# 应用调用方式对比：API 直调与百炼应用调用 - -阿里云百炼平台为已编排好的应用（智能体、工作流、新版智能体 Agent 2.0）提供两套对外调用路径：一是面向 OpenAI 生态的 **Responses API（OpenAI 兼容模式）**，二是面向百炼原生的 **DashScope `Application.call` / `/completion` API**。两者底层均指向同一个 `APP_ID`，但在端点形态、SDK 选型、输入结构、多轮与多模态能力、扩展参数等方面存在差异。本文从技术选型视角对比两种方式，帮助开发者根据现有技术栈与功能需求做出取舍。 - -## 关键维度对比 - -| 维度 | OpenAI 兼容 Responses API（API 直调） | DashScope 原生 API（百炼应用调用） | -| --- | --- | --- | -| 调用端点 | `POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` | -| SDK 选型 | OpenAI SDK（多语言） | DashScope SDK（Python / Java），或直接 HTTP | -| base_url 配置 | `https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1` | 无需 base_url，SDK 内置或直接 POST | -| 输入格式 | `input` 为字符串或消息数组，`role` 取 `system`/`user`/`assistant`，多模态 `content` 为数组（`input_text`/`input_image`/`input_file`） | `input.prompt` 字符串，或 `messages` 数组（自行管理多轮历史） | -| 输出格式 | OpenAI Responses 结构，`response.output` 等 | `{"output": {"finish_reason","session_id","text"}, "usage":{...}, "request_id":"..."}`，业务侧消费 `output.text` | -| 支持模型 | 智能体、工作流、新版智能体 Agent 2.0 | 智能体应用、工作流应用（[智能体编排](../concepts/agent-orchestration.md)应用已被工作流应用替代） | -| 同步/异步 | 支持 `background` 异步执行，同步默认；流式 `stream=true` | 主要为同步调用，`session_id` 由云端管理历史 | -| [流式输出](../concepts/streaming-output.md) | 支持（`stream=true`），异步暂不支持流式 | 通过 SDK / HTTP 支持（详见调用文档） | -| 多轮对话 | 传递完整 `input` 消息数组；基于 `pre_response_id`/`conversation_id` 的上下文能力后续支持 | 两种方式：`session_id`（云端托管，1 小时有效、最多 50 轮）或自行维护 `messages`（推荐，更灵活） | -| 多模态 | 原生支持文本、图像、文件（`input_file` 仅智能体应用支持） | 通过 `messages` 与应用内编排支持 | -| 自定义参数透传 | 通过 `input`/应用编排间接实现 | `biz_params.user_defined_params` 透传至自定义插件与工作流插件节点 | -| 业务空间 | 默认空间仅需 APP ID；子空间或海外地域需在请求中包含 Workspace ID | 同样需要 APP_ID；子空间按地域 Base URL 处理 | -| 典型场景 | 复用现有 OpenAI 代码库与工具链、多模态交互、统一 OpenAI 协议接入 | 全面功能与更高性能、自定义插件参数透传、Java/Node.js 直接 HTTP 集成 | - -## 适用场景建议 - -**OpenAI 兼容 Responses API 适合：** - -- 已有 OpenAI SDK 代码资产、希望以最小改动接入百炼应用的团队。 -- 需要多模态输入（文本 + 图像 + 文件）的智能体交互场景。 -- 希望统一在 OpenAI 协议生态下做模型/应用切换、保持代码中立。 -- 需要异步执行（`background`）与[流式输出](../concepts/streaming-output.md)能力的实时或长任务交互。 - -**DashScope 原生 `/completion` API 适合：** - -- 追求更全面功能与更高性能，使用百炼原生能力（如自定义插件参数透传 `biz_params`）。 -- Java/Node.js 项目希望直接以 HTTP 方式集成，不引入 OpenAI SDK 依赖。 -- 工作流应用需要通过 `session_id` 让云端托管对话历史，简化多轮实现。 -- 需要在工作流大模型节点中配合 `historyList` 变量精细控制提示词与上下文。 - -## 技术选型建议 - -1. **优先看协议生态**：若团队代码栈已围绕 OpenAI SDK 构建（含观测、重试、流式解析），选 Responses API 可降低迁移与维护成本；若以阿里云/DashScope 体系为主，选原生 API 更顺。 -2. **看扩展能力**：自定义插件参数透传（`biz_params.user_defined_params`）目前是原生 API 的明确能力，需要此能力的场景应选原生 API。 -3. **看多轮管理偏好**：希望云端托管历史、降低客户端状态复杂度，用原生 API 的 `session_id`；希望完全自控历史与上下文，两套 API 都支持 `messages` 数组方式。 -4. **看多模态需求**：图像、文件等多模态输入在 Responses API 中有标准化的 `content` 数组结构，接入更直接；原生 API 需结合应用编排实现。 -5. **看地域与业务空间**：两套 API 均支持默认空间仅凭 APP ID 调用；子业务空间或海外地域需携带 Workspace ID，选型不影响该约束，但需在请求中正确拼装。 -6. **凭证一致**：两套方式都使用同一份 `DASHSCOPE_API_KEY`，无需为不同调用方式分别管理密钥，切换成本主要在 SDK 与请求结构层面。 - -综上，两种方式并非互斥：同一 `APP_ID` 可同时被两套 API 调用，团队可按业务模块分别选型——面向外部生态集成用 Responses API，面向内部能力扩展用原生 API。 - -## 被对比主题页 - -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md deleted file mode 100644 index 044ec3fc..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md +++ /dev/null @@ -1,45 +0,0 @@ -# 应用评测与应用监控对比 - -阿里云百炼平台同时提供"应用评测"与"应用监控（应用观测）"两套围绕应用质量的能力：前者面向**离线**场景，用评测集 + 评估器系统化衡量应用输出质量；后者面向**在线**场景，端到端追踪已发布应用的真实调用链路并采集延时、Token 量等运行时指标。两者通过共享的"标签"组件打通——线上 Span 可标注后直接沉淀为评测样本，实现"线上观测 → 离线评测 → [模型调优](../concepts/fine-tuning.md)"的闭环。本文从目的、数据来源、输入输出、计费、典型场景等维度对比两者，帮助开发者做技术选型。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 应用监控（应用观测） | -| --- | --- | --- | -| 核心目的 | 离线评估应用输出质量，产出评分、BadCase、归因与调优建议 | 在线追踪应用真实调用链路，采集延时、Token 量等运行时指标 | -| 数据来源 | 评测集（手动上传或大模型自动生成） | 已发布应用在线接收的真实 Prompt 与调用 | -| 触发方式 | 手动发起评测任务，运行评测集 | 应用添加到观测列表后自动追踪，分钟级同步 | -| 应用范围 | 智能体应用（自动评测需配置知识库）、[工作流](../concepts/workflow.md)应用；新版支持自定义类型 | 智能体应用、[工作流](../concepts/workflow.md)应用、高代码应用；暂不支持 Assistant API 创建的智能体应用 | -| 输入格式 | 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）；新版：`.xls`/`.xlsx`，按应用出入参生成模板，单文件 ≤ 20MB，单次最多 10 个文件 | 真实线上请求，无需上传文件；导出支持 JSONL 与 EXCEL | -| 输出形式 | 评分（1-5 / 0-100 / 0-1 等）、评估器结果、BadCase 分析、归因报告、调优建议 | 调用链路树（CHAIN/AGENT/RETRIEVER/LLM 等节点）、延时、Token 量、监控统计图表 | -| 评估器/评分机制 | 预置模板 + LLM 评估器 + Code 评估器；每任务最多 10 个评估器 | 不做评分，仅记录指标；可对 Span 人工打标签 | -| 标签能力 | 分类/布尔值/数字/文本四种类型，与监控共用 | 同一套标签体系，可对每个 Span 标注，自动保存 | -| API 端点 | 通过控制台操作（评测集、评测任务、评估器、标签管理） | 无 API，仅控制台操作；底层依赖 OpenTelemetry 服务 | -| 数据时效 | 任务执行后产出，按版本留存 | 指标分钟级更新，调用记录最长可查 30 天 | -| 计费方式 | 评测任务调用大模型产生 Token 费用，正常计费 | 应用观测功能本身不收费；观测数据存储费用由 OpenTelemetry 服务收取 | -| 关键限制 | 自动评测仅面向已发布且配置知识库的智能体应用，单次最多 8 个应用横向评测；评测任务发起后配置不可修改 | 高代码应用仅能观测到入口 CHAIN 节点，不支持内部链路追踪；暂不支持长期记忆检索过程观测 | -| 版本差异 | 区分"旧版"与"新版"两套界面，评测集类型、关联应用范围、评估器机制不同 | 无新旧版本区分 | - -## 两者协同关系 - -应用监控与评测并非孤立能力，平台在数据层做了打通： - -- **Span → 评测集**：应用观测支持将 Span 数据直接加入评测集（选择目标评测集、导入方式、字段映射，每评测集最多 50 个字段映射），把真实线上调用沉淀为评测样本。 -- **共享标签**：标签管理为评测与监控共用，线上 Span 标注的标签可在评测任务中复用，反之亦然，便于跨阶段质量分析。 - -## 适用场景建议 - -- **选应用评测** when：需要系统化、可重复地评估应用输出质量；要对响应做相关性、有害性、幻觉等语义判断；有领域专家介入做端到端标注；发布前做回归验证或横向对比多个应用（最多 8 个）。 -- **选应用监控** when：应用已上线，需要排查真实调用的延时、失败率、Token 成本；想查看智能体/[工作流](../concepts/workflow.md)内部检索、重写、重排、LLM 等子节点的执行链路；需要分钟级性能监控与最长 30 天的调用记录检索；想从线上真实数据中沉淀评测样本。 -- **组合使用** when：希望构建"线上观测 → 标注/BadCase 沉淀 → 离线评测 → 调优迭代"的闭环质量运营体系。此时建议先开通应用观测（自动评测的前置条件之一），再用观测到的 Span 数据补充评测集。 - -## 技术选型小结 - -若问题偏"质量好不好"（输出是否准确、完整、合规）→ 用应用评测；若问题偏"跑得稳不稳"（延时、失败、成本、链路追踪）→ 用应用监控；若希望用线上真实数据驱动持续调优，则两者配合使用，通过共享标签与 Span 导入评测集的能力打通闭环。开通自动评测前需先开通应用观测，这也是两者协同的一个隐含前置条件。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [application monitoring](../guides/application-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md deleted file mode 100644 index 2194f12d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md +++ /dev/null @@ -1,59 +0,0 @@ -# 应用监控与应用评测对比 - -阿里云百炼平台围绕[智能体应用](../concepts/agent-application.md)、工作流应用提供两类数据驱动的运营能力：**应用监控（应用观测）**与**应用评测**。两者都服务于"用数据评估应用质量、辅助调优"这一目标，但定位不同——应用监控聚焦**线上真实流量的运行时可观测性**（延时、[Token](../concepts/token.md)、调用链路），应用评测聚焦**离线/预上线数据集的质量评估**（评分、BadCase、归因）。本页通过关键维度对比帮助开发者在不同阶段做出技术选型。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 应用评测 | -| --- | --- | --- | -| 核心定位 | 线上运行时可观测，追踪调用链路与性能指标 | 离线/预上线数据集质量评估，量化输出好坏 | -| 数据来源 | 自动采集应用实际调用（Prompt、输出、延时、[Token](../concepts/token.md)） | 评测集（自动生成或手动上传）+ 被测应用输出 | -| 触发方式 | 开启观测后自动同步，分钟级更新 | 手动发起评测任务，任务完成后产出报告 | -| 支持应用类型 | [智能体应用](../concepts/agent-application.md)、工作流应用、高代码应用（仅入口 CHAIN 节点） | [智能体应用](../concepts/agent-application.md)、工作流应用（自动评测需配置[知识库](../concepts/knowledge-base.md)且已发布） | -| 不支持场景 | 暂不支持 Assistant API 创建的智能体；高代码内部链路不可追踪；无 API | 自动评测仅面向配置[知识库](../concepts/knowledge-base.md)的已发布智能体，单次最多 8 个应用 | -| 输入格式 | 实际线上 Prompt 及调用数据（无需用户准备数据集） | 评测集：旧版（`.xls`/`.xlsx` 对话分析、`.jsonl` 知识问答）；新版（`.xls`/`.xlsx`，单文件 ≤20MB，单次 ≤10 文件） | -| 输出格式 | 调用记录、Trace 列表、监控统计图表；可导出 **JSONL** / **EXCEL** | 评测报告：总正确率、BadCase、归因分析、调优建议；任务详情含数据明细与指标统计 | -| 评估方式 | 不评分，仅呈现延时/[Token](../concepts/token.md)/失败率等指标，支持人工标注标签 | 自动评测用大模型评 1-5 分；手动评测人工打标（较差/一般/较好或 1-5 分）；新版支持 LLM/Code 评估器 | -| 评估器/评分 | 无评分器，依赖指标与人工标注 | 预置模板（通用质量/智能体/文本匹配/相似度/格式校验）+ 自定义 LLM/Code 评估器，每任务最多 10 个 | -| 关键指标 | 调用次数（含失败率）、Token 总量（输入/输出）、平均首 Token 耗时、平均调用时长、节点级延时 | 总正确率（≥4 分占比）、BadCase Top-5、RAG 智能体按问题类型分档得分、各评估器通过率 | -| 调用链路追踪 | 支持，按节点类型展示（CHAIN/AGENT/RETRIEVER/LLM/TOOL 等，可嵌套） | 不追踪链路，仅按评测集条目组织输入输出 | -| 数据时效与保留 | 分钟级更新，调用记录最长查 30 天，支持按 Request ID/Trace ID/Span ID 检索 | 任务结果持久保存，评测集每次发布生成新版本，可选特定版本评测 | -| 标签与标注 | 支持对 Span 标注标签（布尔/分类/数字/文本），与应用评测共享统一管理 | 标签为核心组件，用于数据明细筛选；支持普通模式与快速标注 | -| 与对方联动 | 可将 Span 数据直接加入评测集（选目标集、导入方式、字段映射，最多 50 字段映射） | 评测集可承接来自应用观测的真实线上样本 | -| 前提条件 | 需开通 OpenTelemetry 服务并初始化 LogStore；子账号需 `AliyunBailianFullAccess` + 应用观测权限 + `ram:CreateServiceLinkedRole` | 自动评测需已开通应用观测；评测集需发布后可用；任务配置后不可修改 | -| 计费方式 | 功能本身不收费，数据存储费用由 OpenTelemetry 服务收取 | 评测任务调用大模型产生的 Token 费用正常计费；Code 评估器无额外费用 | -| API 支持 | 无 API，仅控制台操作 | 无 API，仅控制台操作（评测集/任务/评估器/标签均通过控制台管理） | -| 典型场景 | 线上性能监控、故障排查、Token 成本分析、瓶颈节点定位、把真实调用回流为评测样本 | 上线前质量验收、Prompt/检索/[知识库](../concepts/knowledge-base.md)调优验证、多应用横向对比、BadCase 归因与优化建议 | - -## 适用场景建议 - -**优先选择应用监控（应用观测）的场景**： - -- 应用已上线，需要持续观察**真实流量**下的延时、Token 消耗、失败率等运行时指标。 -- 需要**端到端调用链路追踪**，定位某个检索/重排/模型节点是性能瓶颈或错误来源。 -- 需要把线上真实调用**回流为评测样本**，让评测集贴近实际分布。 -- 关注**成本治理**，希望按应用/节点维度统计 Token 用量。 - -**优先选择应用评测的场景**： - -- 应用上线前或迭代中需要**系统化评估输出质量**，得到可量化的评分与归因。 -- 需要用**领域专家或自动评估器**对一批固定用例打分，产出 BadCase 与调优建议。 -- 需要对比多个应用/版本在同一评测集上的**横向表现**（自动评测单次最多 8 个应用）。 -- 需要**可复用的评分逻辑**（LLM 评估器做语义判断，Code 评估器做格式/数值精确校验）。 - -## 技术选型参考 - -两者并非二选一，而是**互补的闭环**：应用监控提供"线上发生了什么"的运行时事实，应用评测提供"这些输出到底好不好"的质量判定。推荐组合使用—— - -1. **开发/调优阶段**：用手动或自动评测在固定评测集上迭代 Prompt、检索配置、知识库切片，借助评估器与归因建议收敛质量。 -2. **上线/运营阶段**：开启应用观测监控真实流量，识别性能瓶颈与成本异常，并用 Span 标注或"加入评测集"把线上问题转化为下一轮评测输入。 -3. **数据回流**：将应用观测中标记为错误或低质的 Span 直接加入评测集，形成"线上 → 评测 → 调优 → 线上"的持续优化闭环。 - -选型时关键判据是**数据来源**：要用真实流量就看监控，要用受控数据集就看评测；要量化"好坏"就用评测，要看"运行状态"就用监控。两者共享标签体系，便于跨阶段贯通标注与筛选。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [application evaluation](../guides/application-evaluation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md deleted file mode 100644 index 9c8ed744..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md +++ /dev/null @@ -1,56 +0,0 @@ -# 百炼应用与托管 Agent 对比 - -阿里云百炼平台提供两条构建 AI 应用的路径：**[智能体应用](../concepts/agent-application.md)**（含新版 Agent 2.0、旧版 Agent 1.0、工作流、高代码应用）与 **Managed Agents（托管 Agent）**。二者都能组合模型、提示词、工具与 MCP 服务，但在运行模式、状态管理、执行环境和适用任务上有本质差异。本页帮助开发者理解两者定位，做出正确的技术选型。 - -核心区别在于：**[智能体应用](../concepts/agent-application.md)是无状态的应用构建与发布形态**，由应用侧维护上下文，面向问答、对话等交互式场景；而 **Managed Agents 是服务端托管的运行时**，在独立云端沙箱中持久化会话状态并支持中断续接，面向多步工具调用、代码执行、文件处理等长时运行任务。 - -## 关键维度对比 - -| 对比维度 | [智能体应用](../concepts/agent-application.md)（LLM Application） | Managed Agents（托管 Agent） | -|---------|------------------------------|------------------------------| -| 定位 | 应用构建与发布形态（Agent / 工作流 / 高代码） | 服务端托管的智能体运行时 | -| 运行模式 | 无状态调用，应用侧维护上下文 | 服务端维护会话状态，支持中断与续接 | -| 执行环境 | 共享运行时 | 独立沙箱，百炼托管的云端容器 | -| 状态持久化 | 不持久化，依赖调用方传递历史 | 事件历史在服务端持久化 | -| 事件模型 | 响应级[流式输出](../concepts/streaming.md) | 会话级 SSE 事件流（User/Agent/Tool/Tool_output/Error/Model/System） | -| 核心对象 | 应用（Agent/工作流/高代码应用） | 智能体 / 运行环境 / 会话 / 事件（四对象解耦复用） | -| 工具能力 | 内置沙箱工具（bash/read/write/edit/glob/grep/download_file）、知识库、MCP、插件 | 7 个内置工具（bash/read/write/edit/glob/grep/download_file）、MCP、Skill | -| 依赖安装 | 不支持自定义环境依赖 | 环境可通过 `config.packages` 预装 apt / pip 依赖并设网络策略 | -| 文件挂载 | 单会话最多 10 个文件、单文件 ≤10MB（文件问答） | 资源挂载到 `/mnt/session/uploads`，独立于会话、可多会话复用；单文件 ≤10MB | -| 记忆能力 | 新版 Agent 短期记忆 0-30 轮，长期记忆未支持 | 会话状态由状态机管理，天然承载长程上下文 | -| 主要 API 端点 | 各应用发布后经统一 API 集成调用 | `POST /agents`、`POST /environments`、`POST /sessions`、`POST /sessions/{id}/events`、`GET /sessions/{id}/events/stream` | -| 开发方式 | 零代码（Agent）/ 低代码（工作流）/ Python（高代码） | 控制台向导或 API 编排 | -| 计费方式 | 按模型 Token、知识库召回、MCP/插件、高代码资源等分项计费 | 按模型 Token 用量及托管沙箱资源计费 | -| 典型场景 | 问答、对话、RAG 检索、固定流程自动化、轻量任务 | 多步工具调用、代码执行、文件批处理等长时任务 | - -## 适用场景建议 - -### 选择智能体应用 - -- **快速交付交互式问答/对话**：新版智能体（Agent 2.0）以自然语言配置即可上线，适合业务人员和产品经理。 -- **知识库 / RAG 问答**：需要挂载企业私有知识库并做标签过滤检索时，智能体应用的知识库能力开箱即用。 -- **流程固定、精确可控**：意图分类、多步骤审批等确定性流程，用工作流做可视化节点编排更稳定。 -- **需要完整代码控制或企业级部署**：用高代码应用（Serverless Function / K8s），配套网关、可观测、日志等企业能力。 -- **多渠道发布**：需要发布到钉钉、微信公众号等第三方平台时，智能体应用原生支持。 - -### 选择 Managed Agents - -- **长时运行任务**：单次任务需要多轮工具调用、可能被中断后续接，服务端持久化事件历史更可靠。 -- **代码执行与文件处理**：需要在独立沙箱中执行 shell 命令、读写文件、安装依赖（apt/pip）、下载资源等。 -- **智能体自主规划**：让智能体在云端容器中自主决定命令与工具调用顺序，无需应用侧管理复杂上下文。 -- **资源复用与会话隔离**：同一环境 / 挂载资源被多个会话复用，且会话间修改互不影响时。 - -## 技术选型参考 - -- 若你的应用本质是**一问一答或短流程交互**，优先用智能体应用；无旧版依赖时选新版 Agent 2.0。 -- 若任务是**服务端长跑、含代码执行或多步工具编排**，选 Managed Agents，避免在应用侧自行拼接和维护会话状态。 -- 两者并非互斥：可以用智能体应用承载前端交互体验，用 Managed Agents 承接后端的长时执行任务。 -- 注意文档中 Managed Agents 的模型名存在不一致（向导示例 `qwen3.7-plus`、API 示例 `qwen3-max`），实际以控制台可选模型 ID 为准；智能体应用推荐使用工具调用能力强的千问-Max 系列。 -- 两类方案的文件上传上限均为单文件 10MB，超限需走文件上传 API。 - -## 被对比主题页 - -- [llm application](../guides/llm-application.md) -- [managed agents](../guides/managed-agents.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md deleted file mode 100644 index d23dd1fd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md +++ /dev/null @@ -1,61 +0,0 @@ -# 知识库与数据连接对比 - -在阿里云百炼平台中，**知识库**与**数据连接**是为大模型应用补充外部数据的两条主要路径，二者定位互补但常被混淆：知识库面向 RAG 检索增强生成，强调"先建索引、再语义召回"；数据连接面向外部数据源的统一接入与实时访问，强调"按需查询、按需引用"。本文从输入格式、输出形态、支持模型、调用方式、计费模型、典型场景等维度对两者做技术选型对比，帮助开发者根据数据形态与访问模式选择合适方案。 - -## 关键维度对比 - -| 维度 | 知识库（RAG） | 数据连接（Data Connection） | -| --- | --- | --- | -| 定位 | 基于向量化检索增强生成，为模型补充私有数据与最新信息 | 外部数据源统一入口，安全访问企业数据库、文档系统、对象存储 | -| 数据形态 | 非结构化文档、表格、图片、音视频（向量化后检索） | 非结构化文件、结构化表格、关系型数据库、对象存储、语雀文档 | -| 支持的数据源类型 | 本地上传、OSS、数据连接器导入 | 文件、表格、MySQL、PostgreSQL、PolarDB-X 2.0、语雀、OSS | -| 输入格式 | pdf/docx/pptx/xlsx/txt/markdown/html/png/jpg/mp4/mkv 等 | 文件类同知识库；表格 xlsx/xls；数据库走 SQL；不支持直接导入 JSON/CSV/YAML | -| 处理方式 | 切片 + 向量化 + 召回排序（建立索引） | 文件类解析入库；数据库走流式实时查询；OSS 走向量检索服务 | -| 访问模式 | 语义检索召回切片（TopK 1–20，相似度阈值过滤） | 文件/表格按类目导入与查询；数据库实时执行 SQL；OSS 工具调用 | -| 支持模型 | 千问 QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research、千问 VL 系列、开源版、第三方（DeepSeek-R1/V3.1、abab6.5s、Llama3.1、Yi-Large 等） | 由挂载的应用决定，本身不绑定模型；数据库/SQL 访问依赖应用编排 | -| 向量模型 | text-embedding-v4/v3（512 维）、multimodal-embedding-v1（1024 维） | OSS 连接器需开通向量检索服务；其他类型不强制向量化 | -| API 端点 | 知识库 API（上传租约→文件→类目→索引任务轮询） | 数据连接 API + 应用内调用（searchOSSFile、searchOSSFileByFileName 等工具） | -| 计费方式 | 标准版 0.03 元/知识库/小时；旗舰版 0.2 元/RCU/小时（1 RCU≈50 QPS） | 文件/表格平台存储限时免费、超额按量；数据库走源实例计费；OSS 走 OSS 计费 | -| 地域限制 | 仅中国站华北2（北京）可开通使用 | 数据库类需私网/公网可达；PolarDB-X 2.0 仅私网；语雀仅公网版本 | -| 数据时效 | 增量导入后需重新解析与建索引，存在更新延迟 | 数据库与语雀类为实时访问；文件/表格/OSS 导入后立即可用 | -| 配额要点 | 业务空间类目 500、文件 10 万、数据表 1 千；标准版 100 GB、旗舰版 9,999 GB | 平台存储 10 万文件、1 TB 免费额度；导入文件仅保留 90 天可查看 | -| 典型场景 | 文档问答、企业知识问答、图文并茂回复、音视频检索问答 | 实时查询业务库、引用语雀文档、检索 OSS 对象、结构化表格问答 | - -## 适用场景建议 - -### 优先选择知识库 - -- 数据以**非结构化文档**为主，需要**语义检索**（即使关键词不匹配也能召回）。 -- 需要图文并茂回复、视觉理解、音视频内容检索等 RAG 场景。 -- 应用形态为智能体应用、工作流应用，或通过 SDK 集成 RAG 检索能力。 -- 对**召回质量**有可调优诉求（相似度阈值、TopK、切片策略、元数据过滤）。 - -### 优先选择数据连接 - -- 数据源是**关系型数据库**（MySQL、PostgreSQL、PolarDB-X 2.0），需要在对话中**实时执行 SQL** 查询最新业务数据。 -- 数据存储在**语雀**或**OSS**，希望按需引用而非预先向量化。 -- 数据以**结构化表格**为主，且字段含义清晰，适合模型直接理解。 -- 团队希望以"连接器"方式统一管理企业数据资产，按权限安全访问。 - -### 二者结合使用 - -- 同一批数据既需语义召回又需结构化查询：可先将结构化部分接入表格连接器，再将非结构化部分导入知识库，在应用编排中按查询意图分流。 -- 文件已在 OSS：既可作为知识库数据来源导入（向量化召回），也可通过 OSS 连接器在对话中按需检索——前者适合"问答"，后者适合"取文件"。 -- 语雀文档：通过语雀连接器实时访问，避免频繁同步；若需历史版本语义检索，仍需导入知识库。 - -## 技术选型参考 - -1. **看数据形态**：非结构化文档/图片/音视频 → 知识库；结构化数据库/表格 → 数据连接；混合 → 两者并行。 -2. **看时效要求**：实时业务数据（订单、库存）→ 数据连接（流处理类）；可容忍延迟的知识沉淀 → 知识库。 -3. **看访问模式**：语义相似度召回 → 知识库；精确 SQL 查询或工具取文件 → 数据连接。 -4. **看成本结构**：知识库按规格/RCU 计费，检索量大时旗舰版更经济；数据连接平台存储限时免费，主要成本在源数据库与 OSS。 -5. **看地域与权限**：知识库仅北京地域可用；数据库类连接器需保证网络可达并完成 DMS/EventBridge/DTS 等角色授权。 - -> 选型结论：**知识库解决"模型不知道"的问题，数据连接解决"模型拿不到最新数据"的问题**。在百炼平台中两者并不互斥，建议依据数据源类型与访问模式组合使用，由应用编排层按查询意图分发。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md deleted file mode 100644 index fd5d160e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md +++ /dev/null @@ -1,52 +0,0 @@ -# 应用[评测](../concepts/evaluation.md)与模型[评测](../concepts/evaluation.md)对比 - -百炼平台提供两套互相独立的[评测](../concepts/evaluation.md)能力：**应用评测**评估的是已编排好的智能体/工作流应用的端到端输出质量，**模型评测**评估的是大语言模型本身在标准或自定义数据集上的基础能力。两者面向的评测对象、数据组织方式、评估器机制、计费路径均不同，开发者需根据选型阶段（选模型 vs 验应用）选择合适工具。本文从评测对象、数据格式、评估机制、结果产出、计费、典型场景等维度进行对比，供技术选型参考。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -| --- | --- | --- | -| 评测对象 | 已发布的智能体应用、工作流应用（端到端整体输出） | 大语言模型本身（千问系列、开源版、第三方文本模型、调优模型） | -| 评测范式 | 自动评测（大模型生成评测集并评分 1–5）、手动评测（人工打标） | 自动评测（系统跑模型推理后评分）、人工评测（人工打标） | -| 评测集类型 | 旧版：对话分析（.xls/.xlsx）、知识问答（.jsonl）；新版：智能体/工作流/自定义三类（.xls/.xlsx，≤20MB，单次≤10 文件） | 自定义评测（评测数据集 / 推理结果集，Excel）、基线评测（系统预置榜单数据集，无需自备） | -| 评测集字段 | Prompt/Completion/SessionId（对话分析）、query/referenceAnswer/fineKeywords/coarseKeywords（知识问答），新版按应用出入参自动生成模板 | `${prompt}`（输入）、`${output}`（模型输出）、`${completion}`（参考答案）；评测数据集含 Prompt+Completion，推理结果集额外含 Output | -| 评估器/评测维度 | 评估器可复用：预置模板（通用质量/智能体/文本匹配/文本相似度/格式校验）+ 自定义（LLM 评估器、Code 评估器、基于评测任务创建）；单任务最多 10 个 | 评测维度 5 种：大模型评估-数值型、规则评估-文本相似度、大模型评估-分类型、规则评估-字符串匹配、人工评估-分类型；可组合多维度 | -| 评分尺度 | LLM 评估器自定义范围（0-100 / 1-5 / 0-1），Code 评估器返回数值；通过阈值决定 Pass/Fail | 数值型 1–5 分，分类型 Pass/Fail，规则型相似度/匹配度；综合得分=维度平均分 | -| 评分器 Prompt 变量 | 映射到评测集字段或应用输出（query、reference_response、context 等） | 固定三变量 `${prompt}` / `${output}` / `${completion}` | -| 预置模板 | 通用质量、智能体、文本匹配、文本相似度、格式校验五类 | 综合评测（5 维度）、语义相似度、自定义评测三种 | -| 关联范围 | 单次自动评测最多 8 个智能体应用横向对比 | 支持多模型同维度排行榜对比 | -| 基线/榜单能力 | 无 | 提供基线评测（C-Eval、MMLU、ARC、GSM8K、HellaSwag、BBH），仅支持调优模型，不支持预置模型 | -| API/触发方式 | 控制台编排评测任务，调用应用产生 Token 计费 | 控制台创建任务，系统自动推理或读取已有 Output | -| 计费方式 | 评测任务调用大模型产生的 Token 费用正常计费；LLM 评估器裁判评分产生 Token 费用；Code 评估器无额外费用 | System Prompt 产生被评测模型推理费用，评分器 Prompt 产生裁判模型评分费用；推理结果集直接读 Output 不推理，可降本 | -| 标签与人工标注 | 分类/布尔值/数字/文本四类标签，支持快速标注模式，单条标注页三栏布局 | 人工评估-分类型由人工打标签 | -| 结果产出 | 数据明细 + 指标统计；自动评测含总正确率、BadCase Top-5、归因分析、调优建议 | 数据明细（Prompt/Completion/Output/评分）+ 指标统计（综合得分、通过率、分数分布）；基线评测含分学科明细、能力雷达图、行业对比 | -| 版本差异 | 区分旧版/新版应用评测（评测集类型、关联范围、评估器机制显著不同） | 无新旧版之分 | - -## 适用场景建议 - -### 选应用评测 - -- 已完成智能体/工作流编排，需验证端到端输出质量（含知识库检索、Prompt 编排、工具调用整体效果）。 -- 需要 BadCase 归因与调优建议，指导应用迭代。 -- 需对线上真实 Span 做标签标注，与应用观测联动做质量监控。 -- 多应用横向对比（最多 8 个），选优上线。 -- 评测数据为多轮对话或带 fineKeywords 的知识问答，需要按 SessionId 或信息点粒度评估。 - -### 选模型评测 - -- 处于模型选型阶段，需对比千问系列、开源版或第三方模型的基础能力。 -- 需用标准榜单（C-Eval、MMLU、GSM8K 等）做基线验证，且被测模型为调优后模型。 -- 业务场景需自定义评分标准（问答质量、内容安全、Function Calling、NL2SQL、翻译摘要等），希望以数据驱动选型。 -- 已有模型推理结果，希望直接评分以降低推理成本（用推理结果集）。 -- 需要参与排行榜对比多模型同维度表现。 - -### 组合使用 - -复杂项目可先做模型评测锁定底座模型，再做应用评测验证编排后的端到端质量；两者评分器/评估器都依赖大模型裁判，需同时关注裁判模型 Token 成本。应用评测的 LLM 评估器与模型评测的大模型评估-数值型在机制上同源（都靠裁判模型按 Prompt 评分），但前者参数映射到应用输出，后者固定三变量，迁移评分逻辑时需调整变量绑定方式。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md deleted file mode 100644 index 673a548f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md +++ /dev/null @@ -1,50 +0,0 @@ -# 应用评测与模型评测对比 - -阿里云百炼平台提供两套面向不同对象的评测能力：**应用评测**关注智能体/工作流应用的端到端输出质量（含 RAG 归因、调优建议），**模型评测**关注单个文本生成模型在给定数据集上的能力表现（含基线评测、排行榜）。二者虽然都围绕"评测集 + 评估规则 + 报告"展开，但评测对象、可用能力、操作方式和计费构成差异明显。本页面为开发者做技术选型时提供横向参考。 - -## 关键维度对比 - -| 维度 | 应用评测 | 模型评测 | -|------|----------|----------| -| 评测对象 | 已发布的智能体应用 / 工作流应用（可含 RAG 知识库） | 单个文本生成类模型（含调优模型） | -| 评测目标 | 评估应用回答质量、定位 RAG BadCase、给出调优建议 | 评估模型基础能力、对比选型、验证调优效果 | -| 评测模式/方式 | 自动评测、手动评测；单应用 / 多应用横向（最多 8 个） | 自定义评测、基线评测（公开标准数据集） | -| 评测集/数据来源 | 旧版（对话分析 `.xls`/`.xlsx`、知识问答 `.jsonl`）、新版（智能体 / 工作流 / 自定义，含版本管理）；可从应用观测导入 | 评测数据集（Prompt + Completion，产生推理费用）或推理结果集（已含 Output，免推理费用） | -| 评分/评估器类型 | 评估器：LLM 评估器、Code 评估器、预置模板；每任务最多 10 个 | 评测维度：大模型评估（数值/分类）、规则评估（文本相似度/字符串匹配）、人工评估（分类） | -| 支持模型 | 评测集生成与评分当前仅 `qwen-max` / `qwen-plus` | 被评测：文本生成类模型；裁判模型推荐千问-Max | -| 报告能力 | 总正确率、BadCase 分析、RAG 归因分析、调优建议 | 综合得分、通过率、逐条评分、可参与排行榜 | -| 操作入口/API | 控制台操作（新旧两套系统） | 仅控制台，无公开 API/SDK（编程化可参考 PAI Judge Model API） | -| 地域限制 | 无特殊地域限制（需开通应用观测） | 基线评测仅北京地域可用 | -| 计费构成 | 调用大模型（评测集生成 + 评分）产生的 Token 费用 | 被评测模型推理费用 + 裁判模型评分费用（规则/人工评估无裁判费用） | -| 典型场景 | 智能体上线前质量把关、版本迭代对比、RAG 优化闭环 | 模型选型、调优前后能力对比、基础能力基准测试 | - -## 评分/评估规则对比 - -两者的评估规则思路相通，但组织方式不同：应用评测以"评估器"为组件挂载到评测任务，模型评测以"评测维度"作为可复用模板。 - -| 规则类别 | 应用评测（评估器） | 模型评测（评测维度） | -|----------|--------------------|----------------------| -| 语义理解打分 | LLM 评估器（相关性、幻觉、有害性等，产生 Token 费用） | 大模型评估-数值型 / 分类型（裁判模型打分或 Pass/Fail，有费用） | -| 确定性/规则判断 | Code 评估器（Python 规则，无额外费用） | 规则评估-文本相似度（ROUGE/BLEU/Cosine）、字符串匹配（无费用） | -| 人工标注 | 手动评测 + 标签管理（分类/布尔/数字/文本标签） | 人工评估-分类型（Pass/Fail 标注） | - -## 适用场景建议 - -- **优化一个已上线的智能体/RAG 应用** → 选应用评测。其归因分析能把 BadCase 精确定位到"模型理解有误 / 重排不佳 / 检索无效 / 切片不完整 / 未获取知识"等环节，并直接给出 Prompt、检索配置或知识库切片的优化建议，形成"识别—归因—优化—回归"闭环。 -- **在多个候选模型间选型，或验证微调效果** → 选模型评测。基线评测可用 C-Eval、MMLU、GSM8K、BBH 等公开数据集快速摸底；自定义评测配合排行榜可做定量对比。 -- **需要跨应用/跨版本横向对比** → 应用评测的多应用横向评测（同一评测基准下最多 8 个应用/版本，须关联相同知识库）。 -- **需要编程化、可自动化的评测流水线** → 两者当前都以控制台为主，模型评测无公开 API/SDK；如需 CI 集成，模型评测侧可参考 PAI Judge Model API。 - -## 技术选型参考 - -1. **先看评测对象**：评"应用整体表现（尤其 RAG）"用应用评测；评"模型本身能力"用模型评测。二者不可互相替代。 -2. **控制成本**：有确定性标准的场景优先使用规则评估 / Code 评估器（无裁判/LLM 费用）；模型评测可先用 50-100 条小规模验证，再保存推理结果集复用以免除重复推理费用。 -3. **注意能力边界**：应用评测的评测集生成与评分当前仅支持 `qwen-max` / `qwen-plus`；模型评测当前仅支持文本生成类模型，基线评测仅北京地域可用，任务提交后目标模型、评测维度类型均不可修改，选错需删除重建。 -4. **对待评测噪声**：模型评测中 1-3% 的分差通常为噪声，LLM 评分器存在位置偏差与自我偏好偏差，建议定期人工抽查校准；应用评测同样建议在知识库更新、Prompt 调整、模型升级后触发回归评测。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md deleted file mode 100644 index deadf313..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md +++ /dev/null @@ -1,71 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供两套独立的评测体系：**应用评测**面向已构建的智能体应用和工作流应用，评估端到端的输出质量与 RAG 链路效果；**模型评测**面向底层大模型本身，评估模型的推理能力和指令遵循表现。两者的评测对象、数据流、评分机制和适用场景均有显著差异，开发者需要根据当前所处的开发阶段选择合适的评测方式。 - -## 关键维度对比 - -| 维度 | 应用评测 | 模型评测 | -|------|----------|----------| -| **评测对象** | 智能体应用、工作流应用（已发布的完整应用） | 文本生成类大模型（基础模型或调优后模型） | -| **核心目标** | 验证应用端到端输出质量，定位 RAG 链路问题 | 评估模型推理能力，辅助模型选型或调优验证 | -| **评测方式** | 自动评测（单应用 / 多应用横向）、手动评测 | 自定义评测（AI / 规则 / 人工）、基线评测（公开数据集） | -| **评测集来源** | 基于应用关联知识库自动生成，或手动上传 | 手动上传评测数据集，或使用公开标准数据集（C-Eval、MMLU 等） | -| **评估器 / 评分机制** | 新版评估器（LLM 评估器 + Code 评估器 + 预置模板）；旧版由平台内置评分 | 评测维度模板（大模型评估数值型/分类型、规则评估相似度/匹配、人工评估） | -| **归因分析** | 支持 RAG 链路归因（模型理解有误、重排不佳、检索无效、切片不完整、未获取知识） | 不提供链路归因，仅输出维度得分和通过率 | -| **横向对比能力** | 最多 8 个应用同基准横向对比 | 支持多模型评测结果排行榜对比 | -| **人工标注** | 新版通过标签体系支持四种类型标注（分类 / 布尔值 / 数字 / 文本） | 人工评估维度（Pass/Fail 标注） | -| **前提条件** | 应用已发布、已配置知识库、已开通应用观测 | 无特殊前提，上传数据集即可评测 | -| **地域限制** | 无特殊地域限制 | 基线评测仅北京地域可用 | -| **API 支持** | 通过控制台操作 | 仅控制台操作，不提供公开 API/SDK（可参考 PAI Judge Model API） | -| **计费构成** | 评测集生成 + 应用调用 + 评估器模型的 Token 费用 | 被评测模型推理费用 + 裁判模型评分费用 | - -## 评分体系差异 - -| 对比项 | 应用评测 | 模型评测 | -|--------|----------|----------| -| **评分范围** | 1-5 分（正确率 = 得分 >= 4 的占比） | 可自定义整数区间（默认 0-5，建议不超过 10） | -| **自动评分方式** | LLM 评估器（语义）+ Code 评估器（规则） | 大模型评估（裁判模型）+ 规则评估（ROUGE/BLEU/Cosine/字符串匹配） | -| **评估器数量** | 每任务最多 10 个，建议组合 3-5 个 | 按评测维度配置，无上限说明 | -| **评分模型** | 评测集生成和评估仅支持 qwen-max 和 qwen-plus | 裁判模型推荐千问-Max，被评测模型不限 | - -## 适用场景建议 - -### 优先选择应用评测的场景 - -- 智能体应用已发布上线，需要持续监控输出质量 -- 需要定位 RAG 链路中的具体瓶颈（检索、重排、切片、模型理解） -- 知识库更新或 Prompt 调整后需要回归验证 -- 多个应用版本之间需要横向对比，选出最优配置 -- 需要将人工标注经验固化为自动评估规则（通过评估器模板化） - -### 优先选择模型评测的场景 - -- 项目初期的模型选型，需要在多个候选模型间对比基础能力 -- 模型微调（SFT）后需要验证调优效果是否达标 -- 使用公开基准（C-Eval、MMLU、GSM8K、BBH）快速了解模型通用能力 -- 需要用规则评估（ROUGE/BLEU）做确定性指标验证（如翻译、摘要场景） -- 关注模型推理能力本身，而非上层应用的端到端效果 - -### 组合使用建议 - -典型的开发流程中，两种评测可以分阶段配合使用：先通过**模型评测**完成基础模型选型和调优验证，确定最优模型后构建应用，再通过**应用评测**验证端到端效果并持续迭代优化。 - -## 成本优化对比 - -| 策略 | 应用评测 | 模型评测 | -|------|----------|----------| -| **减少推理费用** | 缩小评测集规模 | 使用推理结果集（复用已有推理输出） | -| **减少评分费用** | 使用 Code 评估器替代 LLM 评估器 | 使用规则评估或人工评估替代大模型评估 | -| **渐进式评测** | 先小规模自动评测，再针对 BadCase 人工复核 | 先 50-100 条验证，再扩大到 200-500 条正式评测 | - -## 来源文档 - -- [application evaluation](../guides/application-evaluation.md) (guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) (guides/model-evaluation-introduction.md) - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md deleted file mode 100644 index a6f32795..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md +++ /dev/null @@ -1,38 +0,0 @@ -# 框架、工具包与 MCP 对比 - -阿里云百炼为开发者提供了多种接入与扩展大模型能力的方式，常见的选择包括三类：**开源框架集成**（LlamaIndex、Spring AI Alibaba）、**[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与官方 SDK/工具包**（compatible-mode/v1、DashScope SDK、LangChain 适配）、以及**模型上下文协议（MCP）服务**。三者面向的诉求不同——框架侧重在既有编程语言生态中拼装 RAG/[智能体应用](../concepts/agent-application.md)；兼容接口族侧重用最小改动复用 OpenAI 代码与生态；MCP 则侧重让智能体/工作流动态调用外部工具与云资源。本文从接入方式、语言/运行时、能力范围、适用场景、计费与限制等维度做横向对比，供技术选型参考。 - -## 关键维度对比 - -| 维度 | 开源框架（LlamaIndex / Spring AI Alibaba） | [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与工具包 | MCP 服务（官方 + 自定义） | -| --- | --- | --- | --- | -| 接入方式 | 框架 SDK + 百炼云端[知识库](../concepts/knowledge-base.md) / 应用 ID | 替换 `api_key`、`base_url`、`model` 三参数，复用 OpenAI 路径 | 在智能体/工作流中挂载 MCP 服务，或外部通过 Streamable HTTP 调用 | -| 语言/运行时 | Python 3.9+（LlamaIndex）、Java JDK 17+ / Spring Boot 3.x（Spring AI Alibaba） | 任意支持 OpenAI SDK 的语言；官方同时适配 LangChain、LangChain4j | 语言无关（协议层）；自定义服务可由 npx/uvx/http 部署 | -| 鉴权 | [API Key](../concepts/api-key.md)（`DASHSCOPE_API_KEY` 等环境变量）；子[业务空间](../concepts/workspace.md)需[业务空间](../concepts/workspace.md) ID | [API Key](../concepts/api-key.md)（推荐 `DASHSCOPE_API_KEY`）；新加坡与北京地域 Key 不同 | [API Key](../concepts/api-key.md) + MCP 服务自身鉴权（如 `Authorization` 头）；仅主账号及授权 RAM 用户可访问自定义服务 | -| 主要能力 | 云端[知识库](../concepts/knowledge-base.md)构建、RAG 应用、调用百炼智能体/工作流应用、[知识库](../concepts/knowledge-base.md)检索 | Chat、Responses、Completions、Embedding、Vision、File、Batch、Conversations | 官方工具（地图、联网搜索等）、自定义脚本工具、封装 RESTful API、操作阿里云 OpenAPI（OSS、ECS 等） | -| 知识库/RAG | LlamaIndex 用云端智能切分与官方向量模型，不支持自定义切分/嵌入；Spring AI Alibaba 通过 `DashScopeDocumentRetriever` 检索百炼知识库 | Embeddings 接口做向量化；RAG 需自行在应用层编排 | 不直接提供 RAG；可作为工具被智能体调用，间接参与检索/查询 | -| 模型范围 | LlamaIndex 传 `qwen-max` 等；Spring AI Alibaba 调用智能体/工作流应用（应用背后绑定模型） | Qwen 全系（商业/开源/VL/Coder/Omni/Math）、DeepSeek、Kimi、GLM、MiniMax 等；Responses 支持 qwen3-max/plus/flash、qwen3-coder-plus 等 | 由承载 MCP 的智能体/工作流模型决定；调用准确性依赖提示词，必要时换用千问 3 系列等更强推理模型 | -| 调用形态 | 非流式与流式（Spring AI Alibaba `agent.call` / `agent.stream`） | 流式与非流式；Responses 支持 `previous_response_id` 多轮接续；Batch 异步批量（费用 50%） | 智能体自动判断是否调用；工作流中每节点单工具、手动串联；外部调用走 Streamable HTTP | -| 计费方式 | 按所调用模型/知识库的百炼标准计费 | 按模型 token 计费；Batch/Batch Chat 半价；Responses 上下文关联 7 天 | 云部署 MCP 限时免部署费；联网搜索 2000 次免费后 29 元/千次；自定义基础模式 0.000156 元/秒，极速模式另加 0.000036 元/秒部署时长 | -| 典型场景 | 已使用 Python/Java 生态、希望以框架方式构建 RAG 或集成百炼智能体 | 已有 OpenAI/LangChain 代码、希望低成本迁移或复用生态工具 | 让智能体/工作流动态调用第三方工具或阿里云资源，避免逐个写接口 | - -## 适用场景建议 - -- **选开源框架（LlamaIndex / Spring AI Alibaba）**：团队以 Python 或 Java/Spring 为主技术栈，希望以框架抽象快速搭建 RAG 应用或集成百炼智能体/工作流应用，且可接受云端智能切分与官方向量模型（LlamaIndex）或预先在控制台创建应用/知识库（Spring AI Alibaba）。若需要完全自定义文档切分与嵌入模型，LlamaIndex 路线并不适合，应改用本地知识库方案。 -- **选 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与工具包**：已有基于 OpenAI SDK 或 LangChain/LangChain4j 的存量代码，希望以最小改动（`api_key`/`base_url`/`model`）迁移到百炼，或需要使用 Completions（FIM 代码补全）、Batch（半价批量推理）、Responses（智能体原生能力、内置工具、多轮接续）等专项接口。适合追求协议兼容、跨语言复用与生态工具接入的团队。 -- **选 MCP 服务**：核心诉求是让百炼智能体或工作流在运行时动态调用外部工具（地图、联网搜索、自建脚本、RESTful API、阿里云 OSS/ECS 等），而非固定编写接口。适合需要多工具协同、逐步推理、或把已有业务 API 快速封装给模型使用的场景。注意 MCP 只能在智能体/工作流应用中使用，不能在直接调用千问 API 时接入；且会因工具返回内容进上下文而增加 token 消耗。 - -## 技术选型参考 - -1. **先明确诉求边界**：是"迁移/复用现有 OpenAI 代码"（走兼容接口族）、"用框架拼装 RAG/[智能体应用](../concepts/agent-application.md)"（走开源框架），还是"让运行时智能体动态调用外部工具"（走 MCP）。三者并非互斥，常组合使用——例如用兼容接口族做模型调用，同时用 MCP 扩展工具能力。 -2. **地域与鉴权一致性**：兼容接口族需按[业务空间](../concepts/workspace.md)专属域名拼装 `base_url`，弗吉尼亚地域使用固定域名且不带 `{WorkspaceId}`；Spring AI Alibaba 应用集成与知识库检索对 API Key 变量名约定不同（`DASHSCOPE_API_KEY` vs `AI_DASHSCOPE_API_KEY`），关键是 `application.yml` 占位符与实际变量名一致；子业务空间一律需要业务空间 ID。 -3. **能力限制与成本**：LlamaIndex 云端方案不支持自定义切分/嵌入；Completions 仅限北京地域；MCP 单智能体最多 5 个服务、工作流每节点单工具、自定义服务托管在 FC 无固定出口公网 IP（访问云资源需配白名单或打通 VPC）。批量推理与 Batch Chat 可享 50% 费用优惠；MCP 联网搜索有免费额度与 QPS 限制，自定义服务按响应速度分基础/极速两种计费。 -4. **协议演进**：MCP 已从旧版 SSE 升级为 Streamable HTTP，已开通用户需"取消开通"后重新"立即开通"完成升级；Responses API 旧版路径即将停用，应使用 `/compatible-mode/v1/responses`，且 `previous_response_id` 传顶层 `id`（UUID，有效期 7 天）。 - -## 被对比主题页 - -- [frameworks](../api/frameworks.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [model context protocol](../guides/model-context-protocol.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md deleted file mode 100644 index 58583e15..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md +++ /dev/null @@ -1,73 +0,0 @@ -# 模型微调、压缩与部署对比 - -在百炼平台上，把一个模型从「原始能力」推向「专属生产服务」通常经过三个环节：**模型微调 → 模型压缩（可选）→ 模型部署**。三者构成一条完整的自定义模型生产链路，但目标、输入输出、计费方式和适用场景各不相同。本页横向对比这三个环节的关键差异，帮助开发者理清「先做什么、要不要做、怎么上线」的技术选型问题。 - -需要先明确三者的关系： - -- **模型微调**解决「模型能不能做这件事」——把领域知识、指令遵循、人类偏好或特定音色写入参数。 -- **模型压缩**解决「部署贵不贵」——把全精度微调模型量化为低精度版本，降低部署所需的 MU 规格与推理成本，是链路中的**可选**环节。 -- **模型部署**解决「怎么对外提供服务」——为预置模型或自定义模型分配资源专享的推理服务，满足高并发、低延迟需求。 - -> 三者均**仅在华北2（北京）地域可用**，且需使用该地域的 API Key；子账号（RAM 用户）需预先获得相应授权。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（量化） | 模型部署（Deployment） | -| --- | --- | --- | --- | -| 核心目标 | 定制模型能力（知识/指令/偏好/音色） | 降低部署规格与推理成本 | 提供资源专享的推理服务 | -| 链路位置 | 第一步（必选） | 中间步骤（可选） | 最后一步（上线必选） | -| 输入 | 训练/验证数据集（JSONL、ZIP、OSS 挂载） | 上游全精度微调模型 + 可选校准数据 | 预置模型或自定义（含压缩后）模型 | -| 输出 | 微调后模型（`finetuned_output`） | 低精度量化模型 | 专属推理服务（`deployed_model`） | -| 支持模型范围 | 千问文本/VL、万相图像/视频、CosyVoice 等多模态 | 仅平台微调产出的自定义模型（如 qwen3.5-flash） | 部分预置模型 + 所有调优后模型 | -| 主要方式 | CPT / SFT（全参、LoRA）/ DPO | 量化（不含剪枝、蒸馏） | PTU / 模型单元(MU) / 按 Token | -| API 端点 | `POST /api/v1/fine-tunes`（配合 `/files`） | 无公开 API，仅控制台创建压缩任务 | `POST/GET/DELETE /api/v1/deployments` | -| 控制台路径 | 模型调优页面 | 模型 → 模型训练 → 模型压缩 | 我的模型 / 部署页面 | -| 计费方式 | API 仅按 Token；训练单元须走控制台 | 压缩任务限时免费，成本体现在部署阶段 | PTU（按 TPM）/ MU（按时长×单元）/ 按 Token | -| 可逆性 | 可继续微调、可组合多阶段 | **不可逆**，不支持继续微调或二次压缩 | 可下线重建；计费方式创建后不可改 | -| 典型场景 | 效果不达标、需深度领域/风格定制 | 微调模型上线成本高、需降本 | 高并发/低延迟生产、专属推理服务 | - -## 各环节适用场景建议 - -### 模型微调：什么时候做 - -- 当 Prompt 工程、插件调用等手段**仍无法满足效果**时，才引入微调。 -- 文本生成推荐按 `CPT（可选）→ SFT → DPO（可选）` 递进组合：补领域知识用 CPT（千万级 Token 无标签文本），学会遵循指令用 SFT（1000+ 条 ChatML 问答对），对齐人类偏好用 DPO（100+ 组 chosen/rejected）。 -- 训练模式在模型支持全参时**优先选全参**（费用相同、效果更好）；数据集小或对时间/成本敏感则用 LoRA 高效训练。 -- 多模态定制（视觉理解、图像/视频、语音合成）走各自独立的超参与数据格式，其中 CosyVoice 当前**只能通过 API 发起**。 -- 入门建议用控制台零代码流程；批量/自动化建议走 API 四步流程（上传数据 → 创建任务 → 轮询状态 → 部署调用），注意 **API 创建的任务仅支持按 Token 计费**。 - -### 模型压缩：要不要做 - -- 仅当已有**平台微调产出的自定义模型**且**部署成本偏高**时才考虑，属于可选优化环节。 -- 收益示例：qwen3.5-flash 微调模型从 MU1*2（108 元/小时）压缩到 MU8*1（47 元/小时），成本节省约 56%。 -- 权衡精度与成本：量化模板中 MU 编号越大，部署规格越小、成本越低，但精度损失可能越大；用与推理场景语义相近的校准数据可提升量化精度。 -- **注意不可逆**：压缩后不能继续微调、不能二次压缩，调整需从上游全精度模型重新压缩。 -- 压缩任务当前**限时免费**，建议在免费期内对同一模型尝试多个量化模板，用业务测试集验证后再选最优方案上线。 - -### 模型部署：怎么上线 - -按流量特征选择互斥的计费方式（创建后不可更改，切换须先下线再重部署）： - -- **预置吞吐（PTU）**：预留资源保障 TPM，额度内不限速，TPS 通常提升 1.5～2.0 倍，支持长输入（部分模型达 200K）与前缀缓存折扣。适合流量可预估的高负载生产（智能客服、实时内容审核）。超额或超长输入自动转按量计费，业务不中断。 -- **模型单元（MU）**：按时长×单元数计费，资源独占、性能可自定义，支持 PD 分离降低首 Token 延迟。适合需要独占资源、性能可控的生产场景，也是压缩后模型的落地方式。 -- **按 Token 使用量**：不使用不计费，仅支持基础模型 SFT 高效训练后的自定义模型，主要用于**调优效果验证**；扩缩容需控制台人工审核。 - -LoRA 模型导入需满足约束：仅支持 LoRA（不支持全参微调）、rank 为 8/16/32/64、不得修改 vocab 与 chat_template、VL 模型须冻结 VIT。API 部署通过 `plan` 字段区分计费方式（`ptu` / `mu` / `lora`），流程为创建部署 → 轮询至 `RUNNING` → 调用推理 → 用完 DELETE 下线停止计费。 - -## 技术选型参考 - -1. **只想验证微调效果**：微调（SFT 高效训练）→ 直接用「按 Token」部署验证，成本最低、无需长期占用资源。 -2. **要正式上线且流量可预估**：微调 →（评估成本后可选压缩）→ PTU 部署，享受不限速吞吐与缓存折扣。 -3. **要独占资源、性能可控**：微调 → 压缩降本 → MU 部署，用量化模板在成本与精度间取平衡。 -4. **本地已训练 LoRA**：跳过平台微调，直接经「我的模型」导入（满足 rank 与配置约束）→ 部署。 -5. **成本敏感的高负载场景**：优先评估「压缩 + PTU/MU」组合，先在压缩免费期内多模板试验，用业务测试集选出精度/成本最优版本再上线。 - -总体决策顺序：先用微调确保能力达标，再判断是否需要压缩降本，最后按流量与性能诉求选择部署计费方式。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md deleted file mode 100644 index 025bfd66..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型微调与模型压缩对比 - -[模型微调（Fine-tuning）](../concepts/fine-tuning.md)与模型压缩（量化）是百炼平台模型生产链路中两个相邻但目标截然不同的环节。微调解决的是「模型能不能做好某项任务」的**能力问题**，通过训练修改模型参数来提升特定行业/业务表现；模型压缩解决的是「模型部署贵不贵」的**成本问题**，通过量化把全精度微调模型转为低精度版本，从而降低部署所需的 MU 规格。二者在完整链路中的位置为：**模型调优 →（可选）模型压缩 → 模型部署**——压缩的输入正是微调的产出。本文从技术选型角度对比两者的关键差异，帮助开发者判断在什么阶段该用哪个能力。 - -## 关键维度对比 - -| 维度 | [模型微调（Fine-tuning）](../concepts/fine-tuning.md) | 模型压缩（量化） | -| --- | --- | --- | -| 核心目的 | 提升模型在特定任务/领域的能力 | 降低部署规格与推理成本 | -| 处理对象 | 基础模型（千问系列、VL、图像/视频、语音等） | 仅限百炼平台微调产出的自定义模型 | -| 输入 | 训练数据集（ChatML / 纯文本 / ZIP 音视频等） | 上游全精度微调模型 + 可选校准数据集 | -| 输出 | 新的自定义微调模型 | 低精度（量化）版本的自定义模型 | -| 技术手段 | CPT / SFT / DPO（全参或 LoRA 高效训练） | 量化（不含结构剪枝、知识蒸馏） | -| 使用方式 | 控制台可视化 或 API/命令行（DashScope HTTP） | 控制台：模型 → 模型训练 → 模型压缩 → 创建压缩任务 | -| API 端点 | `POST /api/v1/files`、`/api/v1/fine-tunes`、`/api/v1/deployments` | 以控制台操作为主（未提供公开 API 流程） | -| 关键配置 | learning_rate、n_epochs、max_length、lora_rank、training_type 等 | 量化模板（MU 编号）、量化产出后缀、校准数据 | -| 计费方式 | 按训练 Token 计费（CosyVoice 0.2 元/千 Tokens；部署另计） | 压缩任务本身限时免费；压缩后模型按部署 MU 规格计费 | -| 地域限制 | 仅华北2（北京），须用该地域 API Key | 仅华北2（北京） | -| 可逆性/复用 | 微调模型可继续被压缩、部署 | 不可逆；压缩后不支持继续微调或二次压缩 | -| 典型收益 | 指令遵循、领域知识、偏好对齐、风格定制 | 部署成本下降（示例约节省 56%） | -| 典型场景 | 通用模型无法满足行业需求、需注入专业知识或对齐偏好 | 微调模型已达标、需在保持能力前提下压低上线成本 | - -## 适用场景建议 - -### 优先选择模型微调 - -- Prompt 工程、插件调用等方法已用尽，仍无法满足业务对准确率、专业性或风格的要求。 -- 需要向模型注入领域词汇与事实（CPT）、教会模型遵循特定指令或执行任务（SFT）、或对齐人类偏好并抑制幻觉（DPO）。这三种方式可按 `CPT（可选）→ SFT → DPO（可选）` 递进组合。 -- 涉及多模态定制：视觉理解（VL，仅 SFT）、图像/视频生成（SFT-LoRA，需触发词）、语音合成（CosyVoice，仅 API 发起）。 -- 建议：若模型支持全参训练则优先全参（效果更好且与高效训练计费相同）；对成本/时间敏感或数据集较小时选 LoRA 高效训练。 - -### 优先选择模型压缩 - -- 已经拥有一个训练达标的全精度微调模型，主要痛点是**部署/推理成本偏高**。 -- 希望在尽量保持模型能力的前提下降低部署 MU 规格（MU 编号越大规格越小、成本越低，但精度损失可能越大）。 -- 建议：利用压缩任务限时免费的窗口，对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选出成本与精度平衡最优的方案再正式上线；若面向特定场景（如客服问答），校准数据应选语义相近的数据集以提升量化精度。 - -### 二者结合的典型链路 - -对绝大多数生产落地而言，两者不是「二选一」而是「先后使用」：先用**微调**把模型能力打磨到位并验证效果，再用**压缩**在上线前压低部署成本。需特别注意压缩不可逆——若后续还想继续微调或调整，必须回到上游全精度微调模型重新压缩，因此应先冻结微调版本、确认效果达标后再进入压缩环节。 - -## 技术选型速查 - -- 目标是「让模型更会做事」→ 微调。 -- 目标是「让模型更省钱部署」→ 压缩。 -- 需要通过 API/命令行自动化全流程（上传数据 → 训练 → 部署）→ 微调具备完整 HTTP 接口；压缩目前以控制台操作为主。 -- 数据集较小或预算有限 → 微调选 LoRA 高效训练；压缩选较大 MU 编号模板并做好精度验证。 -- 两者均受**华北2（北京）地域**约束，须使用该地域 API Key 与工作空间。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md deleted file mode 100644 index b77b31c6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型微调与模型压缩对比 - -在百炼平台的模型生产链路中，模型微调（Fine-tuning）与模型压缩（量化）是两个位置相邻但目标截然不同的环节。完整链路为：**模型调优 → 模型压缩（可选）→ 模型部署**。微调解决的是「模型效果不够好」的问题，通过修改模型参数让模型在特定行业/业务上表现更佳；模型压缩解决的是「模型部署成本太高」的问题，通过降低参数精度（量化）在尽量保持能力的前提下缩小部署所需的 MU 规格。两者并非替代关系，而是先后衔接：通常先微调得到高精度自定义模型，再对其进行压缩以降低推理成本。本文面向开发者，对两者的关键维度做对比，供技术选型参考。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（量化） | -| --- | --- | --- | -| 核心目标 | 提升模型在特定任务/领域的效果 | 降低部署 MU 规格，减少推理成本 | -| 技术手段 | CPT / SFT / DPO（全参或 LoRA） | 量化（不含结构剪枝、知识蒸馏） | -| 链路位置 | 生产链路起点 | 位于微调与部署之间（可选环节） | -| 输入对象 | 基础模型 + 训练数据集 | 百炼平台微调产出的自定义模型 | -| 输入数据格式 | ChatML（SFT/DPO）、纯文本（CPT）、ZIP/OSS（[多模态](../concepts/multimodal.md)）等 | 源模型 +（条件选填）最多 5 个校准数据集 | -| 输出产物 | 微调后的自定义模型（可部署、可再微调） | 低精度自定义模型（不可再微调、不可二次压缩） | -| 支持模型 | 千问文本/VL、Wan 图像/视频、CosyVoice 语音等[多模态](../concepts/multimodal.md) | 以控制台展示为准，如 qwen3.5-flash-2026-02-23 微调模型 | -| 使用方式 | 控制台（可视化）或 API/命令行（DashScope HTTP） | 控制台操作：模型训练 → 模型压缩 → 创建压缩任务 | -| 关键 API 端点 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`、`GET /api/v1/fine-tunes/`、`POST /api/v1/deployments` | 暂无公开 API，通过控制台创建压缩任务 | -| 关键配置项 | learning_rate、n_epochs、max_length、lora_rank、training_type 等 | 任务名称、源模型、量化模板（MU 编号）、量化后缀、校准数据 | -| 任务状态 | PENDING → RUNNING → SUCCEEDED | PENDING → QUEUING → RUNNING → SUCCEEDED / FAILED / CANCELED | -| 计费方式 | 按训练 Token 计费；CosyVoice 0.2 元/千 Tokens + 部署时长 | 压缩任务限时免费，压缩后模型按部署 MU 规格计费 | -| 可逆性 | 微调模型可继续训练、再压缩 | **不可逆**，不支持继续微调或二次压缩 | -| 地域限制 | 仅华北2（北京），须用该地域 API Key | 仅华北2（北京） | -| 典型场景 | 注入领域知识、学会指令遵循、对齐人类偏好、定制风格 | 高精度模型上线前的成本优化 | - -## 各方案的适用场景建议 - -### 优先使用模型微调 - -- **Prompt 工程/插件调用仍无法满足需求**：需要模型掌握专业词汇、事实性知识（CPT），或稳定地遵循特定指令与任务格式（SFT）。 -- **需要对齐人类偏好、抑制幻觉**：在 SFT 之上叠加 DPO，用「更好-更差」回答对进一步优化。 -- **[多模态](../concepts/multimodal.md)定制**：图像/视频生成的风格定制（Wan LoRA + 触发词）、视觉理解任务、语音合成音色克隆（CosyVoice，仅 API）。 -- **效果为先且能接受较高部署规格**：官方推荐若模型支持全参训练则优先全参（效果更好且与高效训练计费相同）。 - -### 适合使用模型压缩 - -- **已有微调完成的高精度自定义模型，且部署成本偏高**：例如需要长期在线服务、对每小时 MU 费用敏感的场景。以 qwen3.5-flash 微调模型为例，压缩后可从 MU1\*2（108 元/小时）降至 MU8\*1（47 元/小时），成本节省约 56%。 -- **业务对少量精度损失可接受**：MU 编号越大部署规格越小、成本越低，但精度损失可能越大，需按业务权衡。 -- **有语义相近的校准数据**：客服问答等场景应选择贴近实际推理语义的校准数据集，以提升量化精度。 - -## 面向开发者的技术选型参考 - -1. **先明确瓶颈**：效果不达标 → 微调；效果已达标但推理太贵 → 压缩。二者不冲突，常规链路是「先微调，再按需压缩」。 -2. **顺序与不可逆性**：压缩必须以微调产出的自定义模型为输入，且压缩不可逆。若后续还想继续微调或调整训练，务必保留上游全精度微调模型；压缩模型无法二次微调或二次压缩。 -3. **成本估算**：微调按训练 Token 计费（CosyVoice 另计），部署按 MU 规格计费；压缩任务本身限时免费，收益体现在部署阶段更低的 MU 规格。建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证效果，选最优方案上线。 -4. **接入方式差异**：微调支持控制台与 API/命令行（可编排到 CI/CD），压缩目前以控制台操作为主。 -5. **共同约束**：两者均仅在华北2（北京）可用，须使用该地域 API Key；压缩仅支持百炼平台微调产出的自定义模型，不支持基础模型或第三方模型。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md new file mode 100644 index 00000000..f349420c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md @@ -0,0 +1,60 @@ +# 多模态生成 API 对比：图像、视频与3D生成 + +本文档面向百炼平台开发者，旨在系统性对比图像生成、视频生成与3D生成三类核心多模态生成 API 的能力边界、技术特性与工程实践差异。随着AIGC应用向高维内容（2D→3D→时序）纵深演进，准确理解各模态API在输入约束、输出形态、调用范式、计费逻辑及适用场景上的异同，是构建稳定、高效、可扩展的AI原生应用的关键前提。本对比基于当前（2026年Q2）百炼平台正式发布的生产级API能力，所有信息均来自官方文档与实际接口行为验证。 + +## 关键维度对比 + +| 维度 | 图像生成 API | 视频生成 API | 3D生成 API | +|------|--------------|----------------|--------------| +| **核心能力定位** | 文生图、图生图、局部编辑、风格迁移、背景/人物/商品等垂直场景生成 | 文生视频、图生视频（首帧/首尾帧/续写）、参考生视频、数字人播报、口型驱动、视频编辑与风格重绘 | 文生3D、单图生3D、多视角图生3D；输出带PBR材质或无贴图的GLB模型及渲染预览图 | +| **主流输入格式** | - 文本 [prompt](../guides/prompt.md)（中英文，≤512 token）
- 参考图 URL（最多14张，HTTPS公网可访问）
- `messages` 结构（支持图文混排） | - 文本 [prompt](../guides/prompt.md)
- 单图/首帧/首尾帧/参考图 URL（`media` 数组，类型明确标注）
- 音频 URL（数字人场景）
- 多媒体混合输入（如图+音频） | - 文本 [prompt](../guides/prompt.md)（≤1024字符）
- 单张图像 URL（JPEG/PNG，20–6000px，≤20MB）
- 四视角图像数组（固定顺序：前/左/后/右，长度必须为4，空位用 `{}` 占位） | +| **主流输出格式** | PNG/JPEG 图片 URL（同步直出）或 `output.results[].url`（异步）；支持水印开关 | MP4 视频 URL（`output.results[].url`）；部分模型额外返回关键帧图、音频轨道等；URL有效期24小时 | GLB 模型 URL（`pbr_model_url` 或 `base_model_url`） + 渲染预览图 URL（`rendered_image_url`）；所有URL有效期仅**2小时** | +| **支持模型（代表性）** | `qwen-image-3.0-pro`, `wan2.7-image-pro`, `kling/kling-v3-omni-image-generation`, `vidu/vidu-image_reference2image`, `facechain-portrait-generation`, `outfitanyone` | `wan2.7-t2v-*`, `pixverse/pixverse-c1-t2v`, `vidu/t2v`, `kling/kling-video`, `liveportrait`, `emo`, `video-style-transform`, `pixverse-upscale` | `Tripo/Tripo-H3.1`（高精度，≤200万面），`Tripo/Tripo-P1.0`（专业级，≤2万面） | +| **API 端点（典型）** | 同步：`POST /api/v1/services/aigc/multimodal-generation/generation`
异步：`POST /api/v1/services/aigc/xxx/generation` → `GET /api/v1/tasks/{task_id}` | 统一异步端点：
`POST /api/v1/services/aigc/video-generation/video-synthesis` → `GET /api/v1/tasks/{task_id}` | 强地域限定端点（仅华北2）：
`POST /api/v1/services/aigc/video-generation/3d-generation` → `GET /api/v1/tasks/{task_id}` | +| **调用模式** | **混合模式**：
- 同步直出（推荐）：`qwen-image-3.0-pro`, `wan2.7-image-pro`, `z-image-turbo` 等
- 异步轮询（兼容）：`kling`, `vidu`, `wanx` 系列旧模型 | **强制异步**：
所有模型均需 `X-DashScope-Async: enable`，任务创建后轮询；`task_id` 有效期24小时 | **强制异步**：
必须携带 `X-DashScope-Async: enable`；`task_id` 有效期24小时；**不支持同步调用** | +| **地域与密钥约束** | **严格隔离**：华北2（北京）、新加坡、美国（弗吉尼亚）地域的 API Key 与 Workspace ID 完全独立，不可混用 | **严格隔离**：Key、Endpoint、Workspace ID 必须属同一地域；跨地域调用直接鉴权失败 | **强地域锁定**：**仅支持华北2（北京）地域**；其他地域 Key 或 URL 均无效 | +| **计费方式** | - 免费额度：多数模型提供 500 张/90天（主账号与RAM子账号共享）
- 计费粒度：按“生成张数”计费（如 `wanx-v1`: 0.16元/张）
- 限时免费模型（如 `wanx-x-painting`）额度耗尽即停用，不可续费 | - 免费额度：按“任务次数”或“视频秒数”提供（如 `wan2.7` 系列约 30 秒/90天）
- 计费粒度：按“任务成功执行次数”或“输出视频时长×分辨率系数”计费（如 `pixverse-upscale` 按4K超分帧数计费）
- 数字人模型常按“音频时长+图像分辨率”复合计费 | - 免费额度：开通 Tripo 服务后赠送初始额度（具体以控制台为准）
- 计费粒度：按“成功生成的3D模型任务次数”计费；`H3.1`（高面数）单价高于 `P1.0`（专业级）
- **无按面数/贴图质量细分计费，仅按任务成功与否结算** | +| **典型响应时效（平均）** | 同步：3–8 秒
异步：PENDING→SUCCEEDED 通常 10–60 秒（复杂提示/高分辨率可能达2分钟） | PENDING→SUCCEEDED 通常 60–300 秒（3–5秒视频）；超清/长时长/多动作模型可达5–10分钟 | PENDING→SUCCEEDED 通常 2–8 分钟（文生3D较慢，单图/多图相对快）；`H3.1-ultra` 模式可能超10分钟 | +| **关键限制与注意事项** | - 输入图必须公网HTTPS可访问，无中文路径
- `size`/`aspect_ratio` 参数因模型而异，需查对应文档
- 部分模型（如 `wanx` 系列）已停止维护，官方推荐迁移至 `qwen-image` 或 `wan2.7` | - 所有请求必须含 `X-DashScope-Async: enable`
- 数字人模型需先调用 `detect` 模型校验输入合规性
- 旧版 `wan2.1`–`wan2.6` endpoint 已废弃，新版统一使用 `/video-synthesis` | - `prompt`/`image`/`images` **三者严格互斥**，共存即报错
- 多图输入必须为长度4数组，顺序不可变
- 输出URL有效期仅**2小时**，必须及时下载保存
- 不支持任何同步调用尝试 | + +## 各方案适用场景建议 + +| 场景类型 | 推荐方案 | 理由说明 | +|----------|-----------|-----------| +| **高频、轻量、实时反馈型应用**
（如电商详情页实时换背景、设计工具内嵌草图转图、社交App滤镜式生成） | ✅ **图像生成 API（同步直出模型）**
如 `qwen-image-3.0-pro`、`wan2.7-image-pro` | 同步模式毫秒级响应，低延迟体验佳；支持自由分辨率与批量生成（`n=1–9`），契合前端即时交互需求；免费额度充足，成本可控。 | +| **叙事性、时序性、动态表达型内容生产**
（如营销短视频自动生成、教育课件动画、游戏过场预演、数字人直播开场） | ✅ **视频生成 API（`wan2.7` 或 `pixverse` 系列）**
优先选用支持首尾帧/续写的模型 | `wan2.7` 提供最完整的图生视频控制能力（首帧启动、首尾帧约束、视频续写），保障叙事连贯性；`pixverse` 在对口型、动作模仿上表现突出；异步模式天然适配后台任务队列。 | +| **产品可视化、工业设计、虚拟空间构建**
（如电商3D商品展示、AR试穿底层建模、游戏资产快速原型、建筑可视化） | ✅ **3D生成 API（`Tripo/Tripo-P1.0` 或 `H3.1`）** | 唯一提供标准GLB输出的官方API，直接对接Unity/Unreal/WebGL渲染管线；`P1.0` 平衡速度与质量，适合批量生成；`H3.1-ultra` 满足高精度工业级需求；多视角输入显著提升几何准确性。 | +| **需要强语义控制与精细编辑的创意工作流** | ✅ **图像生成 API（垂直专用模型组合）**
如 `wanx-x-painting`（局部重绘） + `wanx-style-repaint-v1`（人像风格） + `image-out-painting`（扩展） | 垂直模型参数精简、效果确定性强；可通过链式调用（上一输出作为下一输入）构建非破坏性编辑流水线，远超通用模型的可控性。 | +| **数字人驱动与音视频融合场景** | ✅ **视频生成 API（人物驱动类）**
如 `liveportrait`（灵动人像）、`emo`（悦动人像）、`wan2.2-s2v`（数字人播报） | 专为肖像动画优化，内置人脸检测、关键点追踪、表情/唇动解耦模块；支持音频驱动，输出自然流畅；需配合 `detect` 模型前置校验，确保输入质量。 | +| **低成本快速验证与MVP开发** | ⚠️ **谨慎选择**：优先用图像API（免费额度高、调试快）
避免早期重度依赖视频/3D API | 视频与3D生成任务耗时长、失败率略高、URL有效期短，调试周期长；图像API可快速验证提示词工程、风格偏好与基础流程，是更高效的前期验证手段。 | + +## 技术选型参考指南（致开发者） + +1. **从调用范式开始决策**： + 若业务要求**毫秒级响应**（如Web应用内联生成），图像API的同步直出是唯一选择；若可接受**秒级到分钟级延迟**（如后台任务、邮件通知式交付），则视频与3D API 的异步模式完全适用，且更利于资源调度与错误重试。 + +2. **地域与基础设施先行**： + 务必在编码前确认目标地域——**3D API 锁死华北2**，视频/图像API虽多地可用，但Key、Workspace、Endpoint必须严格匹配。建议在CI/CD中注入地域变量，避免硬编码。 + +3. **输入准备是成败关键**： + - 图像/视频API：所有外部URL必须是**公网HTTPS、无中文路径、开启公共读**（OSS需设ACL为public-read）；本地文件请先上传至对象存储再传URL。 + - 3D API：多图输入务必按「前/左/后/右」顺序填充4元素数组，缺失视角用 `{}`，不可省略或错位。 + +4. **输出持久化策略**： + - 图像/视频URL有效期24小时，**建议收到后立即下载并存入自有存储**； + - **3D模型URL仅2小时有效！** 必须在轮询到 `SUCCEEDED` 后**立刻并发下载** `pbr_model_url` 和 `rendered_image_url`，否则任务结果将不可恢复。 + +5. **错误处理标准化**： + 所有API均返回 `request_id`，它是阿里云工单排查的唯一凭证；常见错误应主动捕获： + - `BadRequest.InputDownloadFailed` → 检查图片/视频URL可访问性； + - `InvalidParameter` → 核对输入字段互斥性（尤其3D的`prompt`/`image`/`images`）； + - `Forbidden.AccessDenied` → 确认地域Key与Endpoint匹配； + - `ServiceUnavailable.TooManyRequests` → 视频/3D轮询接口RPS限20，改用[异步回调](https://help.aliyun.com/zh + +## 被对比主题页 + +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) +- [3d generation](../api/3d-generation.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md deleted file mode 100644 index 8b952a86..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md +++ /dev/null @@ -1,54 +0,0 @@ -# 图像、视频与3D生成对比 - -百炼平台把图像、视频、3D 三类生成能力统一收口在模型推理网关下，使用相同的 `Authorization: Bearer ` 鉴权与 JSON 承载的输入输出协议，但三者在输入模态、输出形态、调用模式、耗时、计费维度与典型场景上差异显著。本文面向需要做多模态生成技术选型的开发者，横向对比三类能力的关键维度，帮助快速锁定适合业务的接口族。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 输入格式 | 文本 [prompt](../guides/prompt.md)；参考图/蒙版（URL 或 Base64） | 文本；首帧/首尾帧图像；参考多图；视频；音频 | 文本 [prompt](../guides/prompt.md)（≤1024 字符）；单图 URL；多图（前/左/后/右 4 视角，固定数组） | -| 输出格式 | 单张或多张静态图像（临时 URL） | 视频文件（临时 URL） | GLB 模型（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| 支持模型族 | 通义千问图像、万相（Wan）、Z-Image、可灵、创意工具系列 | 万相（HappyHorse/Wan/wanx）、爱诗 PixVerse、Vidu、可灵 | Tripo（`Tripo/Tripo-H3.1`、`Tripo/Tripo-P1.0`） | -| API 端点 | OpenAI 兼容 `/compatible-mode/v1/images/generations`；或 DashScope 原生 `/services/aigc/text2image/image-synthesis` | `/services/aigc/video-generation/video-synthesis`（万相2.7 等）；或 `/services/aigc/image2video/video-synthesis`（动作/换人/数字人/旧版首尾帧） | `/services/aigc/video-generation/3d-generation` | -| 调用模式 | 同步（兼容模式）或异步轮询（万相/创意工具，返回 task_id） | 全部异步：创建任务得 task_id → 轮询 `GET /tasks/{task_id}` | 全部异步：创建任务得 task_id → 轮询 `GET /tasks/{task_id}` | -| 必要请求头 | `Authorization`；异步任务需 `X-DashScope-Async: enable` | `Authorization`、`Content-Type: application/json`、`X-DashScope-Async: enable`（缺异步头报 `does not support synchronous calls`） | `Authorization`、`X-DashScope-Async: enable` | -| 典型耗时 | 秒级（同步）到数十秒（异步） | 1–5 分钟；视频编辑 5–10 分钟 | 较长，轮询建议间隔约 15 秒 | -| task_id 有效期 | 异步任务 24 小时 | 24 小时 | 24 小时；超时返回 `UNKNOWN` | -| 产物下载链接有效期 | 临时 URL，需及时下载/转存 OSS | 临时 URL | 2 小时 | -| 地域约束 | 各地域通用，按模型开通 | 模型/Endpoint/API Key 须同地域；PixVerse、Vidu 仅华北2（北京） | 仅华北2（北京），且须用北京 API Key | -| 内容安全 | 内置审核，违规返回 `DataInspectionFailed` | 内置审核 | 内置审核，失败返回 `code`/`message` | -| 计费方式 | 按张/按次（视模型） | 按任务/时长 | 按任务类型（`text-to-3d` / `image-to-3d` / `multi-image-to-3d`）计数，`usage` 含生成数量 | -| 典型场景 | 文生图、图像编辑、电商/营销垂类、人像玩法、艺术文字 | 文生视频、图生视频、视频编辑、数字人、肖像动态、风格重绘 | 文生 3D、单图生 3D、多图生 3D，产出可二次加工的 GLB 资产 | - -## 调用模式差异 - -图像生成是三类中唯一支持**同步调用**的：千问-文生图等可走 OpenAI 兼容 `images/generations` 直接拿结果。万相与创意工具则多用 DashScope 原生异步协议，提交后拿 `task_id` 轮询。 - -视频与 3D 一律异步，且都强制 `X-DashScope-Async: enable` 头。视频接口明确要求"同地域"约束（模型、Endpoint、API Key 必须同地域），3D 则更严格——仅华北2（北京）可用。两者都强调"请勿重复创建任务，直接轮询"。 - -## 输入模态对比 - -- **图像生成**：以文本 [prompt](../guides/prompt.md) 为主，部分编辑接口接受参考图与蒙版，输入图支持公网 URL 或 Base64。 -- **视频生成**：多模态输入最丰富，万相2.7 支持文本/图像/音频/视频混合输入，参考生视频可保持角色与音色一致性；首尾帧、参考多图等模式扩展了可控性。 -- **3D 生成**：`prompt`、`image`、`images` 三者互斥。多图模式视角顺序固定为前/左/后/右，数组长度固定 4，不需要的视角传空对象 `{}`，这是 3D 独有的约束。 - -## 输出与产物处理 - -图像与视频输出都是临时 URL，文档建议及时下载或转存 OSS。3D 输出更结构化：根据 `pbr`/`texture` 参数组合返回 `pbr_model_url`（PBR 材质 GLB）或 `base_model_url`（无贴图基础模型），并附 `rendered_image_url` 预览图，链接有效期仅 2 小时，短于图像/视频的临时 URL 生命周期。 - -## 选型建议 - -- **静态视觉物料（海报、商品图、人像）**：选图像生成。中文语义优先千问-文生图或万相-文生图 V2；按指令改图用万相-图像生成与编辑 2.7；电商/营销垂类用虚拟模特、AI 试衣、创意海报；高美感/艺术风格用 Z-Image 或可灵。 -- **动态视频内容（短剧、营销视频、数字人播报）**：选视频生成。通用文生/图生优先万相2.7（`wan2.7-t2v`/`wan2.7-i2v`）；多镜头叙事用万相2.6 `shot_type: multi`；数字人/换人用 `wan2.2-s2v`、`wan2.2-animate-mix`；北京地域可按需选 PixVerse、Vidu。 -- **可复用 3D 资产（游戏、电商 3D 展示、工业建模）**：选 3D 生成，仅限北京地域。高精度需求用 `Tripo/Tripo-H3.1`（最高 200 万面，`geometry_quality: ultra`）；追求速度用 `Tripo/Tripo-P1.0`（最高 2 万面）。需要 PBR 材质保留默认 `pbr: true`，仅需白模则同时关 `texture` 与 `pbr`。 -- **跨模态组合**：可先用图像生成产出关键帧，再喂给视频生成的图生/首尾帧接口生成动态内容；3D 则更适合独立资产管线，与图像/视频管线并行而非串行。 - -新接入一律优先最新版本（万相 2.7、通用图像编辑 2.5、wan2.7、Tripo-H3.1/P1.0），旧版接口保留兼容但不再增强。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md deleted file mode 100644 index f6d09469..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md +++ /dev/null @@ -1,60 +0,0 @@ -# 入门路径对比：开始使用与模型快速上手 - -阿里云百炼同时面向"应用构建者"和"模型调用开发者"两类用户，提供了两条不同的入门路径。`start-using` 侧重于在控制台零代码搭建基于私有知识的问答应用（智能体应用、工作流应用、知识库等）；`get-started-with-models` 则侧重于通过兼容 OpenAI 的 API 直接调用大模型完成第一次推理。本页从目标用户、输入形式、输出形式、支持模型、API 端点、计费方式、典型场景等维度对比两条路径，帮助开发者根据需求做出技术选型。 - -## 关键维度对比 - -| 维度 | [start using](../guides/start-using.md)（应用构建路径） | [get started with models](../guides/get-started-with-models.md)（模型调用路径） | -| --- | --- | --- | -| 主要目标 | 零代码/低代码搭建端到端的私有知识问答应用 | 通过 API 直接调用大模型完成文本/多模态推理 | -| 目标用户 | 业务人员、应用开发者、希望快速上线 RAG 应用的团队 | 开发者、需要将模型推理集成到自有代码或后端服务的工程师 | -| 上手时长 | 约 5 分钟完成第一个智能体应用（含知识库构建） | 几行代码即可完成首次调用，开通账号后即可跑通 | -| 输入形式 | 控制台可视化配置：System Prompt、知识库、技能、MCP 工具、工作流节点 | HTTP 请求 / OpenAI 兼容 SDK / DashScope SDK，传入 `messages`、`model`、参数 | -| 输出形式 | 发布后的应用（Web、微信、钉钉、音视频实时互动）+ Responses API 同步/异步调用 | 模型推理结果（文本、图像、音频、视频、向量、重排序结果） | -| 应用类型 | 智能体应用（Agent 2.0）、工作流应用、高代码应用、MCP 服务 | 直接调用模型 API；可叠加模型调优（SFT/CPT/DPO）、模型部署、模型评测 | -| 支持模型 | 千问系列（推荐 qwen3.7-max 用于问答）、QwQ 推理系列、DeepSeek 系列、视觉模型 qwen-vl-plus/max、嵌入 text-embedding-v3/v4 | 文本生成（qwen3.7-max/plus、qwen3.6-flash、deepseek-v4-pro/flash、kimi-k2.7-code、glm-5.2 等）、图像/视频/3D/音频/全模态/向量/重排序全谱系 | -| 知识库 | 核心能力：文档/数据/图片/音视频知识库、智能切分、检索调优、图文检索、监控 API | 不直接提供知识库；如需 RAG 需自行搭建或与百炼应用/知识库集成 | -| API 端点 | Responses API（同步 + `background=true` 异步），调用时需传入自定义参数；通过应用 ID 调用 | OpenAI 兼容 `/compatible-mode/v1`、Anthropic 兼容地址、DashScope SDK 端点；按计费方案选择 Base URL | -| 接入域名 | 应用调用走百炼统一接入，发布渠道含微信/钉钉/音视频 SDK | 业务空间专属 `{WorkspaceId}.{region}.maas.aliyuncs.com`（生产推荐）、`dashscope.aliyuncs.com`（存量）、`trial.{region}.maas.aliyuncs.com`（验证）、Token/Coding Plan 专属域名 | -| 凭证与鉴权 | 控制台操作为主；API 调用时使用应用相关凭证 | API Key（按业务空间隔离，不可跨地域混用）；建议写入环境变量 `DASHSCOPE_API_KEY` | -| 地域选择 | 应用开发与批量推理、模型调优仅在北京与新加坡支持 | 多地域可选（北京、新加坡、法兰克福、东京、弗吉尼亚），地域决定接入点与数据存储；服务部署范围可限定中国内地/国际/全球 | -| 计费方式 | 大模型调用计费 + 知识库规格费用 + 知识库模型调用费用（2026-01-04 起正式计费）；提供限时免费额度 | 按量付费（Dashscope 域名 / 业务空间专属 / 试用域名）、Token Plan（交互式，不可用于后端）、Coding Plan（AI 编码套餐）；不同方案对应不同 Base URL | -| 调用模式 | 同步调用（实时交互，可复用 OpenAI 代码库）、异步调用（`background=true` 返回 Task ID） | 同步、流式（SSE）、批量推理、异步任务；请求超时最高 3600 秒（业务空间专属域名） | -| 可观测性 | 应用观测（端到端流程）、应用评测（智能体/工作流/自定义评测集）、长期记忆与用户画像 | 模型告警（仅北京/新加坡）、模型评测、模型调优 | -| 典型交付物 | 可对外发布的问答应用（含欢迎语、预设问题、发布渠道、音视频互动） | 一次模型推理调用或集成了模型能力的后端服务 | - -## 适用场景建议 - -### 选择 [start using](../guides/start-using.md)（应用构建路径）当： - -- 需要在 5 分钟内零代码搭建一个能回答私有领域问题的问答应用。 -- 业务方或非工程师角色希望可视化配置 System Prompt、知识库、技能、MCP 工具。 -- 需要完整 RAG 流程（知识库构建、切分策略、检索调优、图文检索）且不想自行实现。 -- 需要发布到微信、钉钉、H5/APP 等渠道，或需要音视频实时互动。 -- 需要端到端应用观测、评测、长期记忆与用户画像管理。 - -### 选择 [get started with models](../guides/get-started-with-models.md)（模型调用路径）当： - -- 开发者需要将大模型推理直接集成到自有代码、后端服务或 AI 工作流中。 -- 需要使用 OpenAI 兼容 SDK 或 DashScope SDK，复用现有 OpenAI 代码库。 -- 需要访问文本、图像、视频、3D、音频、全模态、向量/重排序等完整模型谱系。 -- 对地域、服务部署范围、接入域名、并发上限、SLA 有精细控制需求。 -- 需要使用 Token Plan / Coding Plan 等专项计费方案，或需要进行模型调优（SFT/CPT/DPO）、模型部署、模型评测。 - -### 两条路径结合使用： - -实际项目中两条路径常常互补——先用 `get-started-with-models` 跑通模型调用、选定合适模型与地域，再用 `start-using` 将模型能力封装为带知识库、技能、发布渠道的完整应用；反过来，已发布的应用也可通过 Responses API 被后端服务以 OpenAI 兼容方式调用。开发者可先明确"我要的是模型推理能力还是端到端应用"，再据此选择起点。 - -## 来源主题页 - -- [start using](../guides/start-using.md)（guides/start-using.md） -- [get started with models](../guides/get-started-with-models.md)（guides/get-started-with-models.md） - -## 被对比主题页 - -- [start using](../guides/start-using.md) -- [get started with models](../guides/get-started-with-models.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md deleted file mode 100644 index c57c9b93..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md +++ /dev/null @@ -1,78 +0,0 @@ -# 图像生成 vs 视频生成 vs 3D生成 - -百炼平台同时提供图像生成、视频生成和 3D 模型生成三大视觉内容创作能力。三者在输入输出格式、模型生态、调用方式、计费模式和适用场景上存在显著差异。本文从开发者技术选型角度，对三类能力进行系统对比，帮助开发者根据业务需求选择最合适的方案。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| **输入格式** | 文本([prompt](../guides/prompt.md))、参考图片、涂鸦草图 | 文本([prompt](../guides/prompt.md))、首帧图片、参考图/视频、音频 | 文本([prompt](../guides/prompt.md))、单图、多图(4视角) | -| **输出格式** | PNG 图片(512x512 至 4K) | MP4 视频(5-10秒，1280x720等) | GLB 模型(带PBR材质贴图或无贴图) | -| **核心模型** | 千问(Qwen-Image)、万相(Wan)、Z-Image、可灵(Kling) | 万相(Wan 2.7)、HappyHorse、Pixverse、Vidu、Kling | Tripo-H3.1、Tripo-P1.0 | -| **调用方式** | 同步/异步均支持 | [异步调用](../concepts/async-invocation.md)(创建任务+轮询获取) | [异步调用](../concepts/async-invocation.md)(创建任务+轮询获取) | -| **典型响应时间** | 秒级至十秒级 | 1-5 分钟 | 分钟级(需轮询，建议间隔15秒) | -| **API端点** | [DashScope SDK](../concepts/dashscope-sdk.md) / HTTP | `/api/v1/services/aigc/video-generation/video-synthesis` | `/api/v1/services/aigc/video-generation/3d-generation` | -| **支持地域** | 多地域(部分模型仅北京) | 多地域(北京、新加坡等) | 仅华北2(北京) | -| **SDK兼容性** | [DashScope SDK](../concepts/dashscope-sdk.md)、HTTP | OpenAI兼容SDK、[DashScope SDK](../concepts/dashscope-sdk.md)、HTTP | 仅HTTP | -| **产物有效期** | 即时返回，无时效限制 | task_id 24小时有效 | 下载链接2小时有效，task_id 24小时有效 | -| **模型数量** | 20+ 模型(含创意工具) | 10+ 模型系列 | 2 个模型 | -| **编辑能力** | 局部重绘、风格迁移、扩图、超分 | 视频重绘、风格转换、口型替换 | 无编辑能力 | - -## 输入输出能力详细对比 - -| 能力 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| 文生内容 | 支持(文生图) | 支持(文生视频) | 支持(文生3D) | -| 图生内容 | 支持(图像编辑、参考生图) | 支持(图生视频、参考生视频) | 支持(单图/多图生3D) | -| 多模态输入 | 多图参考、涂鸦 | 关键帧序列、音频驱动 | 4视角多图(前左后右) | -| 批量生成 | 单次1-9张 | 单次1条视频 | 单次1个模型 | -| 最大输出分辨率 | 4K(wan2.7-image-pro) | 1920x1080 | 最高200万面(H3.1) | - -## 计费与商业化对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| 计费单位 | 按张计费 | 按任务/时长计费 | 按任务计费 | -| 免费体验 | 部分模型有免费额度 | 部分模型有免费额度 | 需开通Tripo服务 | -| 商业化程度 | 大部分已商业化 | 主力模型已商业化 | 已商业化 | - -## 适用场景建议 - -**选择图像生成的场景**： - -- 电商商品图、营销海报、社交媒体配图等静态视觉内容制作 -- 需要精确文字渲染(如广告文案嵌入图片) -- 图像局部编辑、风格转换、AI试衣等垂直场景 -- 对响应速度要求高(秒级出图) -- 需要批量生成多张候选图供筛选 - -**选择视频生成的场景**： - -- 短视频内容创作、广告视频制作 -- 数字人驱动(音频/文本驱动说话、唱歌) -- 服装展示、舞蹈动作等动态展示 -- 需要关键帧精确控制镜头运动 -- 已有视频的风格转换或口型替换 - -**选择3D生成的场景**： - -- 游戏资产、AR/VR场景中的3D物体快速原型 -- 电商3D商品展示(可旋转查看) -- 建筑/工业设计的概念验证模型 -- 需要标准PBR材质的可渲染模型 - -## 技术选型决策参考 - -1. **内容维度**：静态画面选图像生成；需要动态表现选视频生成；需要空间立体展示选3D生成。 -2. **时效要求**：图像生成响应最快(秒级)，适合实时交互；视频和3D均为分钟级异步任务，适合离线批处理。 -3. **生态成熟度**：图像生成模型最丰富、功能最全面；视频生成处于快速发展期，模型迭代频繁；3D生成目前模型较少，但输出质量已达可用水平。 -4. **地域限制**：3D生成仅限北京地域，部分图像创意工具同样限北京；视频生成地域覆盖较广。 -5. **集成复杂度**：图像生成支持同步调用，集成最简单；视频和3D需要实现异步轮询逻辑，建议封装任务状态管理层。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md deleted file mode 100644 index bfc49c97..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md +++ /dev/null @@ -1,52 +0,0 @@ -# 图像、视频与 3D 生成对比 - -阿里云百炼平台在 DashScope 网关上提供了图像、视频与 3D 三大类视觉内容生成能力。三者虽同属「生成式媒体」范畴、共用同一套鉴权与任务模型，但在输入输出格式、可用模型、调用协议、地域限制与产物形态上差异明显。本文面向开发者，横向梳理三类方案的关键维度，帮助在技术选型时快速判断该用哪一类 API。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 主要能力 | 文生图、图像编辑、图像翻译、垂直创意工具（虚拟模特、扩图、擦除补全、海报等） | 文生视频、图生视频（首帧/首尾帧）、参考生视频、视频编辑、数字人、人像驱动、超清、对口型 | 文生 3D、单图生 3D、多图生 3D | -| 典型输入 | `prompt` / `negative_prompt`；编辑类传 `image_url`/`images`/`mask_image_url`；新版用 `messages` 多模态结构 | 文生用 `prompt`；图生/参考/编辑用 `media` 数组或 `image_url`/`video_url`/`audio_url`/`first_frame`/`last_frame` 等 | `prompt`（≤1024 字符）/ `image`（单图）/ `images`（4 元数组：前左后右），三者互斥 | -| 输出格式 | 图像 URL（有效期 24 小时），分辨率随模型而异（如 512×512~2048×2048、1K/2K/4K） | 视频 URL，分辨率 480P/540P/720P/1080P、可设 `duration` | 带贴图 PBR 材质 GLB（`pbr_model_url`）或无贴图基础模型（`base_model_url`）+ 预览渲染图，下载链接**仅 2 小时** | -| 代表模型 | Qwen-Image、万相 Wan/WanX（t2i/imageedit）、Z-Image、可灵、Vidu 等 | 万相 Wan 2.1-2.7、PixVerse、Vidu、可灵 Kling、HappyHorse、EMO/LivePortrait 等 | `Tripo/Tripo-H3.1`（最高 200 万面）、`Tripo/Tripo-P1.0`（最高 2 万面，速度更快） | -| 调用协议 | 异步（主流）+ **HTTP 同步**（仅 wan2.6/2.7、z-image 等新版） | **全部异步**（创建任务 → 轮询），无同步 | **全部异步**（创建任务 → 轮询），无同步 | -| 典型端点 | `text2image/image-synthesis`、`image2image/image-synthesis`、`multimodal-generation/generation`、`virtualmodel/generation` 等多种 | `video-generation/video-synthesis`（主）、部分数字人用 `image2video/video-synthesis` | `video-generation/3d-generation`（单一端点） | -| 生成耗时 | 通常 1-2 分钟 | 通常 1-5 分钟（统一编辑约 5-10 分钟） | 较长，建议轮询间隔约 15 秒 | -| 地域可用性 | 华北2（北京）为主，部分模型也支持新加坡/美国；大量创意工具**仅北京** | 华北2（北京）为主，第三方模型多**仅北京**，万相等部分支持新加坡/美国/德国 | **仅华北2（北京）** | -| 计费方式 | 仅对成功输出图片计费，含免费额度（通常 500 张/90 天），主子账号共享 | 按成功任务计费，随模型与分辨率/时长而异 | 仅对成功结果计数，`usage` 记录任务类型/数量/质量 | -| 关键请求头 | `Authorization`、`Content-Type`；异步需 `X-DashScope-Async: enable` | 同左，异步必带 `X-DashScope-Async: enable` | 同左，异步必带 `X-DashScope-Async: enable` | -| 任务状态 | `PENDING`/`RUNNING`/`SUSPENDED`/`SUCCEEDED`/`FAILED` | 同类异步枚举，过期返回 `UNKNOWN` | `PENDING`/`RUNNING`/`SUCCEEDED`/`FAILED`/`CANCELED`/`UNKNOWN` | - -## 共性与差异要点 - -**共性**：三者都基于 DashScope 网关，共用 `Authorization: Bearer $DASHSCOPE_API_KEY` 鉴权、`model`/`input`/`parameters` 的请求体结构，异步模式均为「创建任务拿 `task_id` → 轮询查询」，且 `task_id` 有效期统一为 24 小时、创建异步任务必须携带 `X-DashScope-Async: enable`。地域隔离规则也一致：模型、Endpoint 与 API Key 必须同地域，跨地域会鉴权失败。 - -**核心差异**： - -- **协议丰富度**：只有图像生成的部分新版模型（wan2.6/2.7、z-image）支持一次请求返回结果的 HTTP 同步调用；视频与 3D **全部只能异步**。 -- **地域自由度**：图像与视频在北京之外还有一定跨地域支持，而 3D 生成**只有华北2（北京）**可用，选型时需特别注意。 -- **产物时效**：图像/视频 URL 有效期 24 小时，而 3D 模型下载链接**仅 2 小时**，需在生成后尽快下载并转存。 -- **输入约束**：3D 的 `prompt`/`image`/`images` 三者互斥，多图必须是固定 4 元数组（前左后右）；图像与视频则允许更灵活的多模态、多图输入组合。 - -## 适用场景建议 - -- **图像生成**：适合海报、电商主图、创意配图、虚拟模特试衣、图像翻译/编辑等静态视觉需求。追求低延迟、希望一次请求出结果时，优先选支持 HTTP 同步的新版模型（wan2.6/2.7、z-image-turbo）；需要精细编辑（改文字、局部重绘、扩图、去水印）则用千问/万相编辑系列。 -- **视频生成**：适合短视频、广告片、数字人播报、口播/对口型、人像驱动等动态内容。需明确任务类型（文生/图生/首尾帧/参考/编辑）选择对应模型，注意多镜头控制方式在不同模型间不一致（万相 2.7、PixVerse-c1 用自然语言 `prompt`，旧版万相 2.6 需显式 `shot_type: "multi"` + `prompt_extend: true`）。 -- **3D 生成**：适合游戏/电商/XR 场景的 3D 资产快速建模。需要高精度、高面数选 `Tripo/Tripo-H3.1`（可用 `geometry_quality: ultra` 达 200 万面）；追求速度、面数需求不高选 `Tripo/Tripo-P1.0`。务必在北京地域开通 Tripo 服务并及时下载 2 小时时效的产物。 - -## 技术选型参考 - -1. **先按产物形态定类别**：要静态图片 → 图像；要动态视频 → 视频；要可交互 3D 模型（GLB） → 3D。 -2. **再评估延迟要求**：对响应速度敏感的图像场景可用同步协议；视频与 3D 必须做好异步轮询与任务状态处理（含 `FAILED`/`UNKNOWN`）。 -3. **确认地域与开通**：3D 与多数第三方视频/图像创意模型只在北京可用，需保证模型、Endpoint、API Key 同地域，并提前在控制台开通授权（如 Tripo）。 -4. **规划产物存储**：所有产出均为限时 URL，建议生成后立即转存至 OSS，其中 3D 仅 2 小时窗口最需注意。 -5. **统一工程实现**：三类 API 共用鉴权、请求头与异步模型，可复用同一套任务提交/轮询/重试封装，仅按 `model` 与端点差异做分支。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-apis-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-apis-comparison.md new file mode 100644 index 00000000..aa141286 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-apis-comparison.md @@ -0,0 +1,67 @@ +# 知识管理相关能力对比：Knowledge API、知识库与向量检索 + +## 背景与目的 +在百炼平台构建 RAG（[检索增强生成](../concepts/rag.md)）应用时，开发者常面临多种知识管理能力的选择：是直接调用封装好的业务级接口？还是基于底层能力自主编排？亦或需精细控制向量化与排序环节？本页旨在系统对比 **Knowledge API（应用网关层）**、**知识库（产品功能层）** 和 **向量与排序能力（模型原子能力层）** 三类核心能力，从技术定位、使用边界、集成成本与可控性等维度提供清晰的选型依据，帮助开发者根据实际场景快速决策，避免能力误用、计费冗余或架构耦合风险。 + +--- + +## 关键维度对比表 + +| 维度 | Knowledge API | 知识库（产品功能） | 向量与排序能力（Vector & Sort） | +|------|----------------|---------------------|------------------------------| +| **定位层级** | 应用网关层（面向业务场景的 RESTful 封装） | 产品功能层（可视化 + 工作流 + API 的全栈 RAG 解决方案） | 模型原子能力层（基础 AI 原语：embedding / multimodal embedding / rerank） | +| **输入格式** | - 检索：`query`（字符串）
- 问答：`messages`（标准 Chat 格式数组）
- 可选 `knowledgeIds`（字符串数组） | - 控制台：上传 PDF/DOCX/TXT/图片/音视频等文件
- API：`CreateIndex` + `Retrieve` 接口支持结构化文档元数据与文本切片 | - 向量：`input`（string/array/file URL）
- 多模态：`contents` 数组（含 `text`/`image`/`video`/`multi_images` 对象）
- 排序：`query` + `documents`（文本或跨模态对象数组） | +| **输出格式** | - 检索：JSON 数组，含 `chunks`（含 `content`, `score`, `metadata`）
- 问答：SSE 流式事件（`planning`/`retrieving`/`generating`），最终为完整答案 JSON | - 控制台：可视化召回结果、溯源高亮、日志分析看板
- 工作流节点：`result` 变量（结构化 chunk 列表）
- API：`RetrieveResponse`（含 `chunks`, `rerank_scores`, `trace_id`） | - 向量：`data.embedding`（float 数组）+ `usage`（token 计数）
- 排序：`results`（按 score 排序的 `index`/`relevance_score` 数组） | +| **支持模型** | ❌ **不支持指定模型**：
- 检索：平台统一调度语义检索引擎
- 问答：底层 LLM 固定调度（非用户可选） | ✅ **支持广泛模型协同**：
- 预置：Qwen3/Qwen2.5/Qwen2/Long/Max/Plus/Turbo/VL-Max/OCR、DeepSeek-R1、Llama3.1、Yi-Large 等
- 自定义：百炼调优后的千问系列模型（以控制台实际可选为准） | ✅ **细粒度模型选择**：
- 向量：`qwen3.7-text-embedding`, `text-embedding-v4`, `qwen3-vl-embedding`, `tongyi-embedding-vision-plus-2026-03-06` 等
- 排序：`qwen3-rerank`, `qwen3-vl-rerank`, `gte-rerank-v2`（即将下线） | +| **API 端点** | - 检索：`POST /api/v1/indices/knowledge/search`
- 问答：`POST /api/v2/apps/knowledge/chat`
（需 workspaceId + Bearer [Token](../concepts/token.md)） | - 控制台操作无端点
- 工作流节点为内部调度
- 底层 OpenAPI：
 `POST /api/v1/services/indices/create`
 `POST /api/v1/services/indices/{index_id}/retrieve` | - 文本向量：`POST /compatible-mode/v1/embeddings`（OpenAI 兼容）或 `/api/v1/services/embeddings/text-embedding/...`
- 多模态向量：`POST /api/v1/services/embeddings/multimodal-embedding/...`
- 排序：`POST /compatible-api/v1/reranks`（纯文本）或 `/api/v1/services/rerank/...`（跨模态） | +| **计费方式** | ✅ 按调用次数计费（QPS 限流 25）
❌ **不单独计向量/Rerank 费用** —— 平台内包处理，费用隐含在 `knowledge` 接口单价中 | ✅ **双轨计费**：
- **规格费**：标准版（0.03 元/小时）、旗舰版（RCU）
- **模型费**：向量化、Rerank、路由、问答生成均按 token 单独计费（Rerank 费 = 初步召回总切片数 × avg_token × 单价） | ✅ **按模型调用计费**：
- 向量：按输入 token 数 × 模型单价
- 排序：按 `query` + `documents` 总 token 数 × 模型单价
- 异步批处理：按行计费（每行 ≤ 2048 token） | +| **典型场景** | - 快速上线客服问答机器人（无需关注底层细节）
- 内部[工具集成](../concepts/tool-integration.md)轻量级知识检索（如工单系统查 SOP）
- 需要 SSE 流式响应、中断控制、开箱即用的对话体验 | - 构建企业级智能知识中枢（多源异构文档 + 多轮对话 + 权限隔离）
- 需精细化配置：相似度阈值、TopK、Meta 抽取、标签过滤、拒答策略
- 要求 SLS 日志审计、用量监控与性能告警 | - 自研 RAG 框架（如 LangChain/LlamaIndex 集成）
- 构建跨模态搜索（图文混合检索、视频关键帧召回）
- 替换默认 Rerank 模型以优化特定领域排序效果
- 批量预计算向量入库（百万级文档离线向量化） | +| **开发控制力** | ⚠️ **低**：不可替换模型、不可跳过 Rerank、不可自定义切片逻辑、不可干预检索流程阶段 | ⚠️ **中**：可通过工作流节点组合、提示词工程、参数调节（如关闭 Rerank）实现部分定制，但知识库类型/Meta 配置创建后不可变 | ✅ **高**：完全掌控输入/输出、模型选择、参数调优（`dimensions`, `instruct`, `enable_fusion`, `top_n` 等）、调用链路（同步/异步/兼容模式） | + +--- + +## 适用场景建议（面向开发者） + +| 场景描述 | 推荐方案 | 理由说明 | +|----------|-----------|-----------| +| **MVP 快速验证**：2 小时内上线一个支持 PDF 文档问答的内部工具，无复杂配置需求，接受平台默认模型与流程 | ✅ Knowledge API | 仅需构造简单 HTTP 请求 + SSE 解析，零知识库创建、零模型选型、零向量管理，最小接入成本。 | +| **生产级知识中心**：需支持 50+ 部门上传不同格式文档（含扫描件 OCR）、设置部门级标签权限、启用多轮对话改写、对接 SLS 做 QA 质量分析 | ✅ 知识库（旗舰版） | 提供完整的生命周期管理（上传/解析/发布/下线）、可视化调试界面、工作流节点编排、SLS 日志投递、以及企业级配额与安全控制，是“开箱即用”的 RAG 生产解决方案。 | +| **自研 RAG 框架集成**：已基于 LangChain 构建服务，需替换默认 embedding 模型为 `qwen3.7-text-embedding`，并用 `qwen3-vl-rerank` 对图文混合结果重排序 | ✅ 向量与排序能力 | 提供 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)与 SDK，可无缝注入现有框架；支持任意模型组合与参数微调，满足算法团队对召回质量的精细要求。 | +| **批量离线处理**：需将 10 万份合同 PDF 提取文本并生成向量，存入自有向量数据库（如 Milvus） | ✅ 向量与排序能力（异步批处理） | `text-embedding-async-v2` 支持单次 10 万行，输入为 OSS URL，自动分片并发处理，比循环调用同步接口效率提升百倍，且费用更优。 | +| **混合检索增强**：在知识库检索基础上，额外叠加关键词 BM25 结果，并用 `qwen3-rerank` 统一融合排序 | ⚠️ 知识库 + 向量与排序能力 | 知识库原生支持混合检索（向量+关键词），但若需自定义融合策略（如加权、规则过滤），应通过 `Retrieve` API 获取原始结果后，调用 `qwen3-rerank` 二次排序。 | + +--- + +## 技术选型参考指南 + +### ✅ 优先选择 Knowledge API 当： +- 项目周期紧张，需“API 即服务”； +- 不关心底层模型、向量维度、Rerank 策略等技术细节； +- 场景明确为“检索”或“问答”，且接受平台统一调度； +- 客户端需强流式体验（SSE 中断/增量渲染）。 + +### ✅ 优先选择知识库当： +- 需长期运营知识资产（版本管理、权限分级、审计日志）； +- 要求低代码/无代码交付（业务人员可维护）； +- 场景复杂：多轮对话、文件预解析、引用溯源、拒答控制； +- 需与百炼工作流、智能体深度集成。 + +### ✅ 优先选择向量与排序能力当： +- 已有成熟 RAG 架构，仅需替换/增强某环节（如升级 embedding 模型）； +- 需跨模态（图+文+视频）联合检索； +- 要求极致性能与成本控制（如关闭 Rerank、自定义 TopK）； +- 需批量异步处理（日志向量化、历史文档入库）。 + +> **重要提醒**： +> - **地域限制**：知识库与 Knowledge API 均**仅支持华北2（北京）地域**；向量与排序能力全球可用（需确认具体模型地域支持）。 +> - **模型演进**：`gte-rerank-v2` 已标记为下线（2026-05-30），新项目请直接选用 `qwen3-rerank` 或 `qwen3-vl-rerank`。 +> - **权限隔离**：Knowledge API 使用 `API Key` 鉴权；知识库 API 需 `AliyunBailianDataFullAccess` 权限；向量/排序 API 使用 `DASHSCOPE_API_KEY`（兼容 OpenAI）。 +> - **调试建议**：复杂问题请分层验证——先用 Knowledge API 快速验证效果，再用知识库控制台查看召回详情，最后调用向量/排序 API 检查各环节输出，定位瓶颈。 + +## 被对比主题页 + +- [knowledge](../api/knowledge.md) +- [knowledge base](../guides/knowledge-base.md) +- [vector and sort](../api/vector-and-sort.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md deleted file mode 100644 index c96ffa81..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md +++ /dev/null @@ -1,63 +0,0 @@ -# 知识库与记忆库对比 - -知识库（Knowledge Base）与记忆库（Memory Library）都是百炼平台为大模型补充外部信息、突破上下文限制的能力，但二者解决的问题截然不同：知识库面向**领域知识的检索增强（RAG）**，把企业私有文档、结构化数据变成可被大模型检索的语料；记忆库面向**跨会话的用户记忆**，把对话中提取的关键信息和用户画像持久化，让智能体"记住"用户。开发者在做技术选型时，常会混淆两者，本文从多个维度对比，帮助你判断何时用哪一个、以及如何组合使用。 - -## 核心定位差异 - -- **知识库**：解决"大模型不知道我的专有知识/最新信息"的问题。数据来源是**静态文档语料**（手册、Excel、图片、音视频），检索的是与问题语义相关的知识切片。 -- **记忆库**：解决"大模型跨会话记不住用户"的问题。数据来源是**动态对话流**，自动提取记忆片段与用户画像，检索的是与当前对话相关的历史记忆。 - -## 关键维度对比 - -| 维度 | 知识库（Knowledge Base） | 记忆库（Memory Library） | -| --- | --- | --- | -| 解决的核心问题 | 领域知识补充、RAG 检索增强 | 跨会话上下文记忆、用户个性化 | -| 数据来源 | 静态文档、Excel/CSV、图片、音视频 | 对话消息流，或 `custom_content` 直写 | -| 输入格式 | pdf/docx/ppt/txt/markdown/html、图片、音视频文件 | `messages` 对话数组 或 `custom_content` 文本 | -| 存储内容 | 向量化后的文档切片 | 记忆片段 + 结构化用户画像 | -| 输出/召回 | 语义相关的知识切片（供大模型引用） | 语义相关的记忆条目（注入 Prompt） | -| 检索机制 | 向量+关键词混合检索 + Rerank 重排 | 语义检索（`top_k` 控制条数） | -| API 端点 | 阿里云百炼 SDK（ApplyFileUploadLease → AddFile → CreateIndex → SubmitIndexJob …） | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*`（AddMemory / SearchMemory） | -| 鉴权方式 | 子账号需 AliyunBailianDataFullAccess 策略 | 环境变量 `DASHSCOPE_API_KEY`（`sk-` 开头） | -| 控制台接入 | 三步创建 → 关联智能体/工作流/外部应用 | 可视化管理记忆库与记忆规则 | -| 零侵入接入 | 工作流节点 / 应用挂载 | OpenClaw 记忆插件（自动捕获/自动召回） | -| 数据隔离 | 按知识库、业务空间划分 | 按 `user_id` 命名空间隔离 | -| 有效期 | 索引长期有效 | 记忆片段默认 180 天（可配 7/30/180 天或永不过期） | -| 计费方式 | 规格费用（运行时长）+ 模型调用费（向量化+Rerank）；2026-01-04 起计费 | 随长期记忆 API 调用（提炼/向量化/检索由服务端完成） | -| 地域限制 | 仅华北2（北京）可用 | 通过 DashScope API 接入，不受知识库地域限制 | -| 典型场景 | 企业文档问答、产品手册检索、表格查询、图文/音视频问答 | 用户偏好记忆、长期助理、跨会话个性化 Agent | - -## 参数与调优侧重点 - -- **知识库**：调优围绕检索质量，关键参数有相似度阈值（0.01~1.0）、初步检索 TopK（1~100）、最大召回数量（1~20）、Rerank 排序模型、Meta 信息与标签过滤。注意 Rerank 费用取决于初步召回切片总数，降低 TopK 可显著省钱。 -- **记忆库**：调优围绕记忆写入与召回，关键参数有 `top_k`（建议 3~10）、`minScore` 相似度阈值、`profile_schema`（画像模板）、`project_id`（记忆片段规则），以及 OpenClaw 插件的 `autoCapture` / `autoRecall` 开关。 - -## 适用场景建议 - -**优先选知识库**，当你需要： -- 让大模型基于企业内部文档、产品手册、FAQ 等**专有静态知识**回答问题； -- 检索结构化表格数据、图片内容或音视频剧情； -- 对答案准确性、引用来源、拒答/防泄漏有强要求（知识问答服务的极速/多轮智能模式）。 - -**优先选记忆库**，当你需要： -- 让智能体**跨会话记住用户偏好、历史信息与画像**，实现个性化交互； -- 为长期助理、客服 Agent 补充"上一次聊了什么"的连续性； -- 通过 OpenClaw 插件对现有 Agent 做**零侵入**的记忆增强。 - -## 组合使用 - -二者并不互斥，在完整的智能体架构中往往协同：**知识库**提供"专业知识大脑"（知道领域事实），**记忆库**提供"个人记忆大脑"（记得当前用户）。例如一个企业客服 Agent，可用知识库检索产品条款保证回答准确，同时用记忆库记住该用户的历史工单和偏好，做到既专业又贴心。 - -## 技术选型小结 - -- 判断信息**是否随用户/会话变化**：不变的领域知识 → 知识库；随对话演进的用户信息 → 记忆库。 -- 判断数据**来源形态**：文件/表格/多媒体 → 知识库；对话流 → 记忆库。 -- 判断接入约束：需要中国站华北2地域且走百炼 SDK → 知识库；只需 DashScope API Key 快速接入或 OpenClaw 插件 → 记忆库。 -- 追求最佳体验时，二者组合使用，分别承担"知识准确性"与"个性化连续性"两个正交目标。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md deleted file mode 100644 index bb4bf118..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md +++ /dev/null @@ -1,78 +0,0 @@ -# 知识库、记忆库与数据接入对比 - -百炼平台提供了三种互补的数据管理机制：**知识库**（Knowledge Base）、**记忆库**（Memory Library）和**数据连接**（Data Connection）。三者分别面向不同的数据形态和业务需求，开发者在构建[智能体应用](../concepts/agent-application.md)时往往需要根据数据特征、实时性要求和集成复杂度做出技术选型。本文从核心定位、数据来源、检索方式、[计费](../concepts/billing.md)模式等关键维度对三者进行系统对比，帮助开发者快速找到最适合自身场景的方案。 - -## 核心定位对比 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 核心技术 | RAG（检索增强生成） | 长期记忆 API（语义提取 + 持久化） | 数据源连接器（实时访问） | -| 解决的问题 | 为大模型补充私有文档和最新信息 | 解决跨会话上下文丢失问题 | 统一管理和访问企业外部数据源 | -| 数据形态 | 非结构化文档、结构化表格、图片、音视频 | 对话中的关键事件、用户画像属性 | 数据库、文档系统、对象存储 | -| 数据生命周期 | 持久存储，手动管理 | 可配置有效期（7/30/180 天或永不过期） | 实时连接，数据留在原系统 | -| 数据所有权 | 平台托管（上传后由平台管理） | 平台托管（自动提取并存储） | 数据留在原处，平台仅建立连接 | - -## 数据输入与支持格式 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 输入方式 | 本地上传、OSS 导入 | 对话消息自动提取或 custom_content 直写 | 连接器配置（数据库凭证、OSS Bucket、[Token](../concepts/token.md)） | -| 支持格式 | PDF/DOCX/DOC/PPTX/TXT/MD/HTML/XLSX/XLS/PNG/JPG/BMP/GIF/音视频 | 对话 messages（JSON）或自定义文本 | MySQL/PostgreSQL/PolarDB-X 2.0/语雀/OSS/文件/表格 | -| 单文件限制 | 文档最大 150 MB（1000 页）；文本最大 10 MB；图片最大 20 MB；音视频最大 512 MB | 无文件概念，按对话轮次写入 | 文件连接器：平台存储最多 100,000 个文件、1 TB；表格连接器：1 TB 免费额度 | -| 数据解析 | 电子文档/文档智能/大模型/Qwen VL/音视频 五种解析方式 | 系统自动从对话中提炼关键信息 | 文件连接器支持与知识库相同的五种解析方式 | - -## 检索与集成方式 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 检索机制 | 语义向量检索 + 关键词混合检索 + Rerank 排序 | 语义检索（基于 user_id 隔离） | SQL 查询（流处理类）或向量检索（OSS 连接器） | -| API 端点 | 百炼 SDK 知识检索/知识问答接口 | `dashscope.aliyuncs.com/api/v2/apps/memory/*` | 通过应用内工作流节点或内置工具调用 | -| 集成入口 | [智能体应用](../concepts/agent-application.md)、工作流应用、外部 API | API 直连、OpenClaw 插件（零侵入） | 应用内数据连接器节点 | -| 多源联合 | 支持最多 15 个知识库联合检索 | 按 user_id 自动聚合同一用户记忆 | 支持同时配置多个连接器 | -| 实时性 | 需重新导入和索引后才能检索新内容 | 对话结束后即时写入，下次对话可召回 | 流处理类实时访问源数据库最新数据 | - -## [计费](../concepts/billing.md)与配额 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 规格费用 | 标准版 0.03 元/知识库/小时；旗舰版 0.2 元/RCU/小时 | 以长期记忆 API 调用[计费](../concepts/billing.md) | 平台存储限时免费；自有 OSS 按 OSS 费用 | -| 模型调用费用 | 向量化 + Rerank 排序（按 [Token](../concepts/token.md) 计费） | 记忆提取和检索的模型调用费用 | 大模型文档解析按模型调用计费 | -| 免费额度 | 新用户 720 小时标准版（开通后 30 天有效） | 参见长期记忆 API 计费说明 | 文件/表格平台存储限时免费（1 TB） | -| 并发限制 | 标准版 1 QPS；旗舰版 50-10,000 QPS | 参见 API 限流说明 | 取决于源数据库和网络配置 | -| 地域限制 | 仅华北2（北京） | 参见 DashScope 服务地域 | 流处理类需网络可达；PolarDB-X 仅支持私网 | - -## 适用场景建议 - -**选择知识库**适用于： -- 企业有大量私有文档（产品手册、技术规范、FAQ 等），需要让大模型基于这些文档进行准确问答 -- 需要图文并茂的回复、复杂 PDF/图表理解、音视频内容检索等[多模态](../concepts/multimodal.md)场景 -- 对检索精度有较高要求，需要向量检索 + Rerank + 多轮对话改写等完整 RAG 流水线 -- 数据更新频率中等（可接受重新导入和索引的延迟） - -**选择记忆库**适用于： -- 智能体需要记住用户的历史偏好、习惯和关键事件，实现个性化持续服务 -- 需要跨会话保持上下文连贯性（如客服机器人记住用户之前反馈的问题） -- 希望零侵入接入（通过 OpenClaw 插件自动捕获和召回，无需改业务代码） -- 数据以对话形式产生，不需要事先准备结构化文档 - -**选择数据连接**适用于： -- 企业数据存储在 MySQL、PostgreSQL 等关系数据库中，需要智能体实时查询最新数据 -- 数据不适合或不允许复制到平台，要求数据留在原系统 -- 需要对接语雀文档、OSS 存储等已有数据系统 -- 业务场景需要执行 SQL 查询获取精确的结构化数据（如订单查询、库存查询） - -## 组合使用建议 - -三种方案并非互斥，在复杂业务场景中推荐组合使用： - -- **知识库 + 记忆库**：知识库提供专业领域知识，记忆库记住用户偏好和历史交互，二者结合实现既专业又个性化的智能服务。 -- **知识库 + 数据连接**：知识库提供静态文档知识，数据连接提供实时业务数据，适合需要同时参考文档和查询数据库的场景（如技术支持 + 工单系统）。 -- **三者组合**：在全场景智能助手中，知识库负责专业知识问答，数据连接负责实时数据查询，记忆库负责用户画像和交互记忆，共同构建完整的智能体数据底座。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md deleted file mode 100644 index 818cb053..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md +++ /dev/null @@ -1,69 +0,0 @@ -# 知识库与长期记忆对比 - -阿里云百炼平台针对「让大模型使用外部信息」提供了两条不同路径：**知识库（RAG）** 负责为模型补充**私有文档与最新事实**，**长期记忆（Memory Library / 长期记忆 API）** 负责跨会话持久化**用户个性化信息**。二者常被混淆，但设计目标、数据形态、检索机制与计费方式都截然不同。本文面向开发者，从技术选型角度对二者做逐维度对比，帮助你在智能体、问答、Agent 等场景中选对能力。 - -其中长期记忆在平台上有两个入口：控制台侧的**记忆库（Memory Library Overview）** 与开放的 **长期记忆（新）RESTful API**，二者底层为同一套服务，下文将它们合并为「长期记忆」一列，必要处再区分入口差异。 - -## 一句话定位 - -- **知识库**：把企业文档/结构化数据/多模态素材建成可检索索引，回答前先检索、再增强生成，解决「模型不知道我的私有资料」。 -- **长期记忆**：从多轮对话中自动提炼关键事件与用户画像并持久化，后续对话语义召回后注入 Prompt，解决「模型记不住这个用户」。 - -## 关键维度对比 - -| 维度 | 知识库（RAG） | 长期记忆（Memory Library / 长期记忆 API） | -| --- | --- | --- | -| 解决的问题 | 为模型补充私有数据与最新信息，提升回答准确性 | 跨会话保留用户偏好与历史，实现个性化上下文 | -| 数据来源 | 预先上传的文档、Excel/CSV、图片、音视频等素材 | 从对话消息自动提取，或 `custom_content` 直接写入 | -| 内容形态 | 切片（chunk）+ 向量索引 + Meta 信息 | 记忆片段（关键事件）+ 用户画像（结构化属性） | -| 写入时机 | 建库/导入阶段离线建索引（AddFile→CreateIndex→SubmitIndexJob） | 对话过程中/结束后实时写入（AddMemory） | -| 输入格式 | 文件（pdf/docx/ppt ≤150MB、txt/md/html ≤10MB、图片 ≤20MB、音视频 ≤512MB） | `messages`（对话，最多 50 条）或 `custom_content`（≤512 字符）+ `user_id` | -| 输出/召回 | 召回相关切片（单次最多 20），供大模型增强生成 | 召回相关记忆片段/画像，注入 Prompt | -| 检索机制 | Query 改写 → 向量+关键词混合检索 → Rerank 精排 → 加权返回 | 语义检索，可选 rerank / query 重写 / 意图判别 | -| 隔离维度 | 按知识库 ID / 业务空间 | 按 `user_id`（记忆空间），可再按 `memory_library_id` | -| API 端点 | `bailian.cn-beijing.aliyuncs.com`（阿里云 OpenAPI 风格） | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` | -| 认证方式 | 子账号策略 `AliyunBailianDataFullAccess` | Header `Authorization: Bearer $DASHSCOPE_API_KEY` | -| 核心接口 | AddFile / CreateIndex / SubmitIndexJob / 知识检索 / 知识问答 | AddMemory / SearchMemory / ListMemory / Update / Delete / ProfileSchema 系列 | -| 关键参数 | 相似度阈值、召回片段数 TopK(1–20)、权重、Meta 抽取、智能切分 | `top_k`(1–100)、`min_score`(默认 0.3)、`enable_rerank/rewrite/judge` | -| 地域限制 | 仅中国站**华北2（北京）**可开通使用 | 通过 DashScope 全局域名接入，无北京地域限制 | -| 数据有效期 | 长期保存，删除即永久清除且停止计费 | API 侧「暂无失效日期」；控制台默认 180 天，可配 7/30/180 天或永不过期 | -| 计费方式 | 规格费用（标准版 0.03 元/库/小时；旗舰版 0.2 元/RCU/小时）+ 模型调用 Token 费；2026-01-04 起计费 | 按接口调用限流管理（全部 ≤3000 QPM，add 120 QPM，search 300 QPM） | -| 零侵入接入 | 关联到智能体/工作流应用（知识库节点） | OpenClaw 记忆插件（`before_agent_start` 召回 + `agent_end` 捕获） | -| 典型场景 | 企业文档问答、产品手册、客服、结构化数据查询、多模态搜索 | 个人助理记住用户习惯、智能体长期陪伴、跨会话偏好延续 | - -## 适用场景建议 - -### 选择知识库（RAG）当 - -- 你有**成体量的私有资料**（文档、手册、报表、图片、音视频）需要模型准确引用。 -- 要求回答**可溯源**（展示引用来源）、可做效果评测与持续优化。 -- 数据是**面向所有用户共享**的事实性知识，而非某个用户的个人信息。 -- 可接受仅在**华北2（北京）**地域使用，并规划好规格费用与 Token 成本。 - -### 选择长期记忆当 - -- 你要让智能体**记住单个用户**的偏好、习惯、历史事件，实现个性化。 -- 数据来自**实时对话**、随时间增长且需自动去重/更新。 -- 需要按 `user_id` 做**多用户记忆隔离**，或用画像模板维护结构化属性。 -- 希望**零侵入接入**（OpenClaw 插件）或用轻量 HTTP API 快速集成。 - -### 组合使用（推荐） - -二者并不互斥，常见的高质量智能体会同时使用：**知识库**提供权威事实与私有知识，**长期记忆**提供该用户的个性化上下文。典型编排为——对话前用 `SearchMemory` 召回用户记忆注入 Prompt，同时用知识检索召回相关文档切片；对话后用 `AddMemory` 沉淀新的用户信息。这样既「答得准」又「记得住」。 - -## 技术选型速查 - -- 「模型不知道我的资料」→ 知识库。 -- 「模型记不住这个用户」→ 长期记忆。 -- 需要**引用溯源 / 结构化数据查询 / 多模态素材** → 知识库。 -- 需要**跨会话个性化 / 用户画像 / 对话自动沉淀** → 长期记忆。 -- 受限于**北京地域**或对**规格费用**敏感 → 优先评估知识库成本；长期记忆走 DashScope 全局接入且以调用限流计。 -- 想**最快接入 Agent** → 长期记忆 OpenClaw 插件（配 `apiKey` + `userId` 即可自动捕获/召回）。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [long term memory new](../api/long-term-memory-new.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md deleted file mode 100644 index 9ff18a9a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md +++ /dev/null @@ -1,59 +0,0 @@ -# 知识库 vs 记忆库 vs 数据接入对比 - -百炼平台提供三种核心数据管理能力：知识库（RAG）、记忆库（长期记忆）和数据连接。三者均用于为大模型补充外部信息，但在数据组织方式、检索机制和适用场景上存在本质差异。本文从开发者技术选型角度，对三者的能力边界和关键特征进行系统对比。 - -## 关键维度对比 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 核心定位 | 基于 RAG 的私有文档检索增强 | 跨会话长期记忆持久化与召回 | 外部数据源统一接入与实时访问 | -| 输入格式 | PDF、DOCX、TXT、Markdown、HTML、XLSX、图片、音视频等非结构化文件 | 对话消息（messages）或自定义内容（custom_content） | 数据库连接串、OSS Bucket、语雀 [Token](../concepts/token.md)、本地文件上传 | -| 数据存储方式 | 平台托管（向量化切片存储） | 平台托管（记忆片段 + 用户画像） | 平台托管或流处理（原数据源实时访问） | -| 检索机制 | 向量检索 + 关键词检索 + Rerank 精排 | 语义检索（基于 user_id 隔离） | SQL 查询（流处理类）或向量检索（OSS 连接器） | -| API 端点 | 知识检索服务 / 知识问答服务（百炼 SDK） | `dashscope.aliyuncs.com/api/v2/apps/memory/*` | 通过应用工作流节点或工具调用 | -| 数据粒度 | 文档切片（最大 6,000 [Token](../concepts/token.md)/片） | 记忆片段（事件级）或用户画像（属性级） | 原始数据行/文件/文档 | -| 数据更新方式 | 手动上传 / OSS 导入，需重新索引 | 自动从对话提取，支持去重和动态更新 | 实时连接原数据源，数据变更即时生效 | -| 多用户隔离 | 无内置用户隔离（按知识库粒度管理） | 原生 user_id 级别隔离 | 按连接器实例隔离 | -| 并发规格 | 标准版 1 QPS / 旗舰版 50-10,000 QPS | 未公开 QPS 限制 | 取决于底层数据源能力 | -| [计费](../concepts/billing.md)方式 | 标准版 0.03 元/库/小时；旗舰版 0.2 元/RCU/小时 + 检索/排序 [Token](../concepts/token.md) 费 | 包含在百炼平台使用中（按 API 调用） | 文件/表格平台存储限时免费；流处理按底层数据源[计费](../concepts/billing.md) | -| 数据有效期 | 永久（手动删除） | 默认 180 天（可配置 7/30/180 天或永不过期） | 永久（随原数据源生命周期） | -| 集成方式 | 智能体应用 / 工作流节点 / SDK API | API 直连 / OpenClaw 插件零侵入接入 | 工作流节点 / 应用工具调用 | - -## 适用场景建议 - -### 选择知识库 - -- 企业有大量非结构化文档（产品手册、FAQ、技术文档）需要语义检索 -- 需要高精度的文档问答能力，支持[多模态](../concepts/multimodal.md)（图文、音视频） -- 对检索并发有明确要求（旗舰版支持万级 QPS） -- 需要精细控制检索质量（切片策略、Rerank、标签过滤等） - -### 选择记忆库 - -- 智能体需要跨会话记住用户偏好、历史交互和个性化信息 -- 需要按用户维度隔离记忆空间 -- 希望零侵入集成（通过 OpenClaw 插件自动捕获/召回） -- 数据来源是对话本身，而非预置文档 - -### 选择数据连接 - -- 需要实时查询企业数据库中的结构化数据（MySQL、PostgreSQL、PolarDB-X） -- 数据存储在外部系统（语雀、OSS）且需保持同步 -- 应用需要执行 SQL 查询获取精确结果 -- 数据更新频繁，不适合定期导入知识库 - -## 组合使用建议 - -三种能力并非互斥，实际应用中常组合使用： - -- **知识库 + 记忆库**：知识库提供通用文档检索，记忆库补充用户个性化上下文，实现"懂业务 + 懂用户"的智能体 -- **知识库 + 数据连接**：非结构化文档走知识库语义检索，结构化数据走数据连接 SQL 精确查询 -- **三者结合**：在工作流中编排知识库节点、数据连接工具和记忆注入，构建具备完整数据感知能力的复杂应用 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md deleted file mode 100644 index 83e098a7..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md +++ /dev/null @@ -1,63 +0,0 @@ -# 知识库与记忆库对比 - -阿里云百炼平台同时提供**知识库（Knowledge Base）**和**记忆库（Memory Library）**两种数据增强能力，二者分别面向不同的信息管理需求。知识库基于 RAG 技术，用于将企业私有文档、结构化数据等外部知识注入大模型，提升特定领域问答的准确性；记忆库则面向跨会话场景，自动从对话中提取和持久化关键信息，使智能体能够在多轮交互中保持对用户偏好与历史上下文的理解。本文从核心定位、数据模型、接入方式、检索机制等维度进行系统对比，帮助开发者根据业务场景做出合理的技术选型。 - -## 关键维度对比 - -| 维度 | 知识库（Knowledge Base） | 记忆库（Memory Library） | -| --- | --- | --- | -| **核心定位** | 基于 RAG 的外部知识检索增强，为大模型补充私有数据与最新信息 | 跨会话长期记忆，自动提取并持久化对话中的关键信息 | -| **数据来源** | 用户主动导入的文档（PDF/Word/Markdown/HTML/Excel/CSV）、图片、音视频等 | 从对话历史中自动提取，或通过 `custom_content` 直接写入 | -| **数据类型** | 文档搜索、数据查询（NL2SQL）、图片问答、音视频搜索四类，创建时选定不可更改 | 记忆片段（事件和信息）和用户画像（结构化属性），可独立或组合使用 | -| **存储粒度** | 切片（Chunk）级别，按智能切分或自定义规则对文档分片，单切片上限 6,000 Token | 记忆片段级别，系统自动提炼为精简的事实描述；用户画像按属性字段存储 | -| **数据持久性** | 持久存储，无过期机制 | 记忆片段可配置有效期（7/30/180 天或永不过期），控制台默认 180 天 | -| **检索机制** | 向量 + 关键词混合检索 + Rerank 排序，支持相似度阈值、TopK、权重等精细调控 | 基于语义的记忆检索（SearchMemory），通过 `top_k` 控制召回数量 | -| **向量模型** | `text-embedding-v4`/`text-embedding-v3`（512 维）；视觉理解用 `qwen3-vl-embedding`；图片问答用 `multimodal-embedding-v1`（1024 维） | 由平台内部自动处理向量化，用户无需选择向量模型 | -| **排序模型** | 支持 `qwen3-rerank`、`qwen3-rerank(hybrid)`、`qwen3-vl-rerank` | 无独立排序模型，由记忆检索服务内部排序 | -| **API 端点** | `bailian.cn-beijing.aliyuncs.com`，通过阿里云 SDK（`alibabacloud_bailian20231229`）调用 | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*`，通过 DashScope API Key 认证 | -| **认证方式** | 阿里云 AccessKey（`ALIBABA_CLOUD_ACCESS_KEY_ID` / `SECRET`）+ WorkspaceId | DashScope API Key（`DASHSCOPE_API_KEY`，以 `sk-` 开头） | -| **接入方式** | 控制台可视化 + 开放 API（仅文档搜索类） | API 直连 + OpenClaw 记忆插件（零侵入自动捕获/召回） | -| **数据隔离** | 按知识库实例隔离，单次检索可绑定最多 15 个知识库 | 按 `user_id` 隔离，同一 `user_id` 共享记忆空间 | -| **可挂载模型** | 千问全系列（QwQ/Long/Max/Plus/Turbo/Coder 等）及第三方模型（DeepSeek/Llama/Yi 等） | 不直接挂载模型，作为独立记忆服务由应用侧在 Prompt 中注入 | -| **计费模式** | 标准版 0.03 元/小时；旗舰版 0.2 元/RCU/小时（1 RCU = 50 QPS），另计 SLS 日志存储费用 | 随 DashScope API 调用计费，无独立规格费用 | -| **地域限制** | 仅支持中国站华北2（北京）地域 | 通过 DashScope 全局端点访问 | -| **监控能力** | 内置日志投递至 SLS，支持调用审计、问题排查、用量统计与告警 | 通过控制台查看记忆库统计信息，无独立日志服务集成 | - -## 适用场景建议 - -### 适合选择知识库的场景 - -- **企业知识问答**：需要基于产品手册、内部文档、技术规范等大量非结构化文档进行精准问答。 -- **结构化数据查询**：通过自然语言查询 Excel/CSV 中的结构化数据（NL2SQL）。 -- **[多模态](../concepts/multimodal.md)检索**：需要对图片、音视频内容进行搜索和问答。 -- **高并发生产环境**：旗舰版支持最高 10,000 QPS，适合对吞吐量有要求的业务系统。 -- **精细化检索调优**：需要通过多种向量模型、排序策略、切片方式、相似度阈值等参数精确控制召回质量。 - -### 适合选择记忆库的场景 - -- **个性化智能助手**：需要记住用户偏好、习惯和历史交互信息，提供个性化服务。 -- **跨会话上下文保持**：用户多次对话之间需要保持连贯性，避免重复提供相同信息。 -- **用户画像构建**：需要从对话中自动提取和维护用户的结构化属性（年龄、职业、偏好等）。 -- **轻量级集成**：通过 OpenClaw 插件零侵入接入现有 Agent，无需改造应用代码。 -- **对话式应用**：客服机器人、个人助理等需要"记住"用户的长期交互场景。 - -### 组合使用 - -在实际业务中，知识库与记忆库可以组合使用以实现最佳效果。例如，一个智能客服系统可以同时挂载知识库获取产品文档中的专业知识，又通过记忆库记住每位用户的历史问题和偏好，从而在准确回答专业问题的同时提供个性化的服务体验。 - -## 技术选型参考 - -选择知识库还是记忆库，核心取决于数据的来源和用途： - -- 如果数据是**预先准备好的静态文档**，需要检索后辅助回答 -- 选择**知识库**。 -- 如果数据是**从对话中动态产生**的，需要跨会话持久化 -- 选择**记忆库**。 -- 如果两种需求并存，建议**同时接入**，各司其职。 - -从工程复杂度看，记忆库的接入成本更低（尤其是 OpenClaw 插件方式），而知识库提供了更丰富的检索调优手段和更高的生产环境承载能力。开发者应根据数据规模、并发需求、检索精度要求和集成方式综合评估。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-guide-vs-api.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-guide-vs-api.md deleted file mode 100644 index d5b6d709..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-guide-vs-api.md +++ /dev/null @@ -1,56 +0,0 @@ -# 托管智能体：指南与 API 对比 - -百炼平台的 Managed Agents（托管智能体）提供了两套面向不同使用者的文档视角：**使用指南**侧重在控制台向导中完成端到端配置，适合快速上手与业务验证；**API 参考**则给出完整的 REST 资源模型与端点定义，适合工程化集成与自动化。二者描述的是同一套托管运行时——平台在服务端托管会话状态、沙箱环境与工具执行——但抽象层级、覆盖范围与目标读者不同。本页从关键维度做对比，帮助开发者按阶段选择合适的入口。 - -## 关键维度对比 - -| 维度 | 使用指南（guides/managed-agents） | API 参考（api/managed-agents-api） | -| --- | --- | --- | -| 定位 | 概念讲解 + 控制台向导操作流程 | REST 资源模型 + 端点契约 | -| 目标读者 | 初次接入、业务验证、快速体验 | 后端集成、SDK 调用、自动化编排 | -| 核心对象 | Agent、Environment、Session、Event（4 类） | Agent、Environment、Session、Event、Skill、File（6 类） | -| 操作方式 | 控制台向导 + 预览调试标签页 | 直接调用 REST API / SDK | -| 认证方式 | 控制台登录（隐式） | HTTP Header 携带 `Authorization: Bearer ` | -| API 基地址 | 未强调 | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio` | -| 支持地域 | 未强调 | 当前仅 `cn-beijing` | -| Agent 端点 | `POST /api/v1/agentstudio/agents`（创建） | 创建/获取/列出/更新/归档/列出版本 全套 | -| 版本管理 | 未展开 | 每次更新递增 `version`，会话锁定当时快照 | -| 事件流 | SSE 事件流（`GET /sessions/{id}/events/stream`） | SSE 长连接 + 事件历史分页（`GET /sessions/{id}/events`） | -| Skill 支持 | 提及可挂载预置工具组合 | 完整生命周期：上传/安全扫描/版本/下载/删除 | -| File 支持 | 上传挂载，路径 `/mnt/session/uploads` | 独立资源：上传/查询/列出/删除，审核状态机 | -| 文件大小上限 | 单文件 10 MB（上传挂载语境） | 单文件 20 MB，工作空间总量 100 GB，保留 30 天 | -| 状态机 | 会话状态、中断续接、工具审批（概念性描述） | `idle` → `running` → `idle` / `terminated`（显式定义） | -| 示例模型 | `qwen3-max`（向导示例另见 `qwen3.7-plus`，以下拉列表为准） | `qwen3-max` | -| 典型场景 | 多步工具调用、代码执行、文件处理等长时任务的快速搭建 | 生产环境集成、批量会话编排、CI/自动化流水线 | - -## 关键差异说明 - -- **抽象层级**：指南把流程压缩成"配置智能体 → 配置环境 → 发起会话 → 发送事件"四步，隐藏了大量端点细节；API 参考把每类资源的增删改查、版本、归档、软/硬删除逐一列清，是精确的契约。 -- **资源覆盖**：指南聚焦运行链路的 4 个核心对象（Agent / Environment / Session / Event），而 API 额外把 **Skill** 与 **File** 提升为一等资源，并明确了它们的安全扫描/审核状态机与挂载版本锁定规则。 -- **文件配额不一致**：指南上下文提到单文件 10 MB，API 参考给出 20 MB、工作空间 100 GB、保留 30 天。以 API 参考的配额为准，并注意实际以控制台/接口返回为准。 -- **端点方法差异**：API 总览页把部分更新端点标注为 `PATCH`，而各资源详情页标注为 `POST`（Agent / Environment / Session 均如此）。集成时以各资源详情页为准。 -- **版本与快照语义**：只有 API 参考明确了"会话创建时锁定 Agent 版本、Skill 挂载锁定具体 version、更新不影响已有会话"的重要约束，这对生产环境的稳定性至关重要。 - -## 适用场景建议 - -- **选择「使用指南」入口**： - - 第一次接触 Managed Agents，想快速理解 Agent / Environment / Session / Event 之间的关系。 - - 需要在控制台向导中拖拽配置、用预览调试标签页按事件类型（User、Agent、Tool、Tool_output 等）观察执行过程。 - - 做业务原型、Demo 或轻量验证，暂不需要写代码。 - -- **选择「API 参考」入口**： - - 需要把托管智能体嵌入后端服务、批量创建/复用 Agent 与 Environment。 - - 关注鉴权、地域、版本锁定、状态机、配额等工程化细节。 - - 需要管理 Skill 上传/安全扫描/版本挂载，或对 File 做上传/审核/挂载的自动化处理。 - - 构建 CI/自动化流水线，依赖稳定的端点契约与 `x-request-id` 排障能力。 - -## 技术选型参考 - -推荐路径是"**先指南、后 API**"：用指南在控制台跑通端到端流程、确认业务可行性，再切换到 API 参考完成工程化集成。两者并非二选一，而是覆盖同一运行时的不同生命周期阶段。集成落地时，务必以 API 参考为准处理认证（Bearer API Key）、地域限制（`cn-beijing`）、版本快照语义与端点方法差异，并按 API 侧的 20 MB / 100 GB / 30 天配额规划文件资源。 - -## 被对比主题页 - -- [managed agents](../guides/managed-agents.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md deleted file mode 100644 index a5f55a41..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md +++ /dev/null @@ -1,58 +0,0 @@ -# 托管智能体与 LLM 应用对比 - -百炼平台既提供面向"应用构建"的 LLM 应用体系（智能体 / 工作流 / 高代码），也提供面向"长时自主任务"的 Managed Agents 托管运行时。二者虽然都以大模型为核心、都能接入知识库与 MCP 工具，但在运行模式、状态管理、执行环境和目标场景上定位截然不同。本文从技术选型视角梳理二者差异，帮助开发者判断"我该用哪一个"。 - -## 概念定位 - -- **托管智能体（Managed Agents）**：平台在服务端托管的智能体运行时，为多步工具调用、代码执行、文件处理等长时任务提供独立云端沙箱容器。会话状态、事件历史由服务端持久化，支持中断与续接，智能体在沙箱内自主执行命令、读写文件、安装依赖。 -- **LLM 应用**：面向应用构建的一整套模式，包含智能体（Agent，含 2.0 / 1.0）、工作流（Workflow）、高代码应用三种类型，覆盖从零代码配置到 Python 编码的不同开发门槛，用于快速搭建可发布、可被 API 调用的 AI 应用。 - -## 关键维度对比 - -| 维度 | Managed Agents（托管智能体） | LLM 应用（智能体/工作流/高代码） | -| --- | --- | --- | -| 核心定位 | 服务端托管的长时任务运行时 | 应用构建平台，多种应用形态 | -| 运行模式 | 服务端维护会话状态，支持中断与续接 | 智能体应用多为无状态调用，应用侧维护上下文 | -| 执行环境 | 独立沙箱、云端容器 | 共享运行时（高代码可选 Serverless/K8s 部署） | -| 事件模型 | 会话级 SSE 事件流，事件历史持久化 | 响应级[流式输出](../concepts/streaming.md) | -| 开发方式 | API 编排（配置 Agent/Environment/Session） | 零代码配置 / 可视化编排 / Python 编码 | -| 主要 API 端点 | `/api/v1/agentstudio/agents`、`/environments`、`/sessions`、`/sessions/{id}/events` | 应用发布后经"发布渠道"提供的调用 API | -| 工具能力 | 内置 7 个工具（bash/read/write/edit/glob/grep/download_file）+ MCP + Skill | 内置沙箱工具、知识库、MCP、插件（按应用类型） | -| 文件处理 | 上传挂载到 `/mnt/session/uploads`，单文件 ≤ 10MB | 文件问答（全文引用/RAG/自定义），单会话 ≤ 10 文件、单文件 ≤ 10MB | -| 记忆/上下文 | 服务端持久化会话事件，可挂载资源复用 | 短期记忆 0-30 轮；长期记忆暂未支持 | -| 发布与集成 | 通过 Agent/Environment/Session API 直接编排 | 需先"发布"，再经发布渠道 API/第三方平台调用 | -| 计费方式 | 按模型 Token 用量 + 沙箱运行资源 | 模型 Token + 知识库召回 + MCP/插件 + 高代码函数计算/网关/存储 | -| 典型场景 | 多步工具调用、代码执行、文件批处理等长时自主任务 | 问答对话、固定流程自动化、企业级后端服务 | - -## 适用场景建议 - -**优先选择 Managed Agents 的场景：** - -- 任务需要**多步自主决策 + 代码执行**，如自动化数据处理、脚本编写与运行、文件批量转换。 -- 需要**长时运行**且要在中途中断、审批、续接的任务，依赖服务端持久化的事件历史。 -- 需要**隔离的沙箱环境**安装依赖（apt/pip）、运行 shell 命令，且不希望自行维护会话上下文。 - -**优先选择 LLM 应用的场景：** - -- **智能体应用（Agent 2.0）**：意图开放、需模型自主规划调用知识库/MCP 的问答与对话类应用，业务人员即可零代码配置。 -- **工作流应用**：流程固定、需精确控制执行链路的多步骤自动化，适合 IT 运维与业务分析师用可视化编排。 -- **高代码应用**：对性能、可观测性、企业级运维有要求的生产级 AI 后端，由工程师用 Python 编码并部署到 Serverless/K8s。 -- **文件问答**：面向文档总结、长文检索、图片/视频分析等，直接在智能体应用中上传文件即可。 - -## 技术选型参考 - -- **控制粒度维度**：从"AI 自主"到"人工精确控制"依次为——Managed Agents / Agent 2.0（模型自主规划）> 工作流（预定义流程）> 高代码（完全代码控制）。需要确定性流程选工作流或高代码；需要自主探索选 Managed Agents 或 Agent 2.0。 -- **状态需求维度**：需要跨轮次持久化会话、支持中断续接的长时任务，只有 Managed Agents 原生支持；LLM 智能体应用多为无状态，需应用侧自行维护上下文。 -- **交付形态维度**：要"对外发布 + 多渠道集成（钉钉/公众号）"选 LLM 应用；要"程序化编排自主任务"直接对接 Managed Agents 的 agentstudio API。 -- **开发门槛维度**：业务人员/产品经理→Agent 2.0；运维/分析师→工作流；AI 工程师→高代码或 Managed Agents API。 - -> 提示：二者并非互斥。可在 LLM 应用中通过 MCP/知识库快速搭建对话入口，同时把重型的多步执行任务下沉到 Managed Agents 沙箱，形成"轻交互层 + 重执行层"的组合架构。 - -## 被对比主题页 - -- [managed agents](../guides/managed-agents.md) -- [llm application](../guides/llm-application.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md deleted file mode 100644 index 12f72f00..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md +++ /dev/null @@ -1,48 +0,0 @@ -# 图像、视频与3D生成对比 - -阿里云百炼平台提供了覆盖多模态内容生成的三大能力：**图像生成**、**视频生成**与**3D 资产生成**。三者都通过 DashScope 网关对外提供服务，共享相似的异步调用范式与鉴权体系，但在支持模型、输入输出格式、接口路径、地域可用性、计费方式和典型应用场景上存在显著差异。本文从技术选型角度对三者做横向对比，帮助开发者根据业务需求快速定位合适的能力。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 支持模型 | 千问 Qwen-Image、万相 Wan/Wanx、Z-Image、可灵 Kling、Vidu 等多家族 | 万相 Wan、HappyHorse、PixVerse、Vidu、可灵 Kling 及多种人像驱动模型 | 仅 Tripo（`Tripo/Tripo-H3.1`、`Tripo/Tripo-P1.0`） | -| 核心功能 | 文生图、图生图、图像编辑、局部重绘、扩图、虚拟模特、AI 试衣、创意海报等 | 文生视频、图生视频（首帧/首尾帧/续写）、参考生视频、视频编辑、数字人/口型/舞蹈等 | 文生 3D、单图生 3D、多图生 3D | -| 输入格式 | `prompt` / `messages` / `images`（JPG/PNG/JPEG/BMP/WEBP，[512,4096] 像素，≤10MB） | `input.prompt` + `input.media`（`first_frame`/`last_frame`/`image_url`/`video` 等） | `prompt`（≤1024 字符）/ `image` / `images`（4 元素数组，三者互斥；JPEG/PNG，[20,6000] 像素，≤20MB） | -| 输出格式 | 图像 URL（有效期 24 小时） | 视频 URL（异步返回） | GLB 模型（`pbr_model_url` / `base_model_url`）+ 预览渲染图（下载链接有效期 2 小时） | -| 主要 API 端点 | 多路径：`.../text2image/image-synthesis`、`.../aigc/multimodal-generation/generation`、`.../image-generation/generation` 等 | `POST .../aigc/video-generation/video-synthesis`（部分模型走 `.../aigc/image2video/video-synthesis`） | `POST .../aigc/video-generation/3d-generation` | -| 调用方式 | 以异步为主（`X-DashScope-Async: enable` + 轮询）；新模型支持 HTTP 同步 | 全部异步（创建任务 → 轮询查询） | 全部异步（创建任务 → 轮询，建议 15 秒间隔） | -| 任务耗时 | 通常 1-2 分钟 | 通常 1-5 分钟 | 较长（异步） | -| 地域可用性 | 华北2（北京）、新加坡、美国（弗吉尼亚）等多地域，独立 Key 不可混用 | 万相/HappyHorse 多地域；PixVerse/Vidu/Kling/数字人/人像模型仅北京 | 仅华北2（北京），需该地域 API Key | -| 计费方式 | 仅对成功生成的输出图片计费，含 90 天免费额度 | 多为后付费按视频时长（元/秒）或按张计费 | 按任务成功结果计数（`text-to-3d`/`image-to-3d`/`multi-image-to-3d`） | -| 并发限制 | 主/子账号共享 QPS 与处理中任务数 | 同时处理中任务通常限 1（排队执行） | 查询接口默认 RPS 20 | -| 典型场景 | 电商海报、虚拟模特、AI 试衣、创意插画、图文混排 | 短视频、广告、数字人播报、动画、口型替换 | 游戏/AR/VR 资产、3D 建模、数字孪生 | - -## 共性特征 - -- **统一网关与异步范式**：三者均通过 DashScope 网关调用，且核心流程都是「创建任务获取 `task_id` → 轮询 `GET .../api/v1/tasks/{task_id}`」。创建任务时必须携带 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。 -- **`task_id` 有效期均为 24 小时**，切勿重复创建任务，轮询获取即可。 -- **地域隔离**：不同地域拥有独立的 API Key 与请求地址，跨地域调用会导致鉴权失败。百炼推荐迁移到业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）。 -- **请求体结构**：普遍由 `model` / `input` / `parameters` 三部分组成。 - -## 各方案适用场景建议 - -- **图像生成**：适合需要静态视觉内容的场景，模型选择最丰富、地域覆盖最广、计费门槛最低（仅计成功输出图并有免费额度），是多模态生成中最成熟、门槛最低的入口。电商与创意工具（虚拟模特、AI 试衣、创意海报）尤为突出，但需注意部分创意工具模型仅提供免费体验、用尽后不可付费。 -- **视频生成**：适合需要动态内容的场景，功能维度最复杂（文生/图生/参考生/编辑/数字人）。选型时优先选用走新版协议、功能最全的万相 wan2.7 系列；若使用 PixVerse、Vidu、Kling 或人像/数字人模型需注意仅限北京地域。计费按时长且并发通常限 1，需评估吞吐与排队成本。 -- **3D 生成**：能力最聚焦，仅基于 Tripo 模型，产出带 PBR 材质的 GLB 模型。仅限北京地域、下载链接仅 2 小时有效期，需及时下载与转存。适合游戏、AR/VR、工业设计等对 3D 资产有需求的场景；`Tripo-H3.1` 追求高精度（最高 200 万面），`Tripo-P1.0` 追求速度。 - -## 技术选型参考 - -1. **地域约束优先评估**：3D 生成及大量第三方视频模型仅支持北京地域，若业务部署在新加坡/海外，应优先确认图像生成或万相视频系列的多地域可用性。 -2. **同步 vs 异步**：仅图像生成的部分新模型（如 `wan2.6-image`、`z-image-turbo`）支持 HTTP 同步一次返回，对低延迟场景友好；视频与 3D 必须异步轮询，需在客户端实现任务状态管理。 -3. **输出有效期差异**：图像/视频结果与 `task_id` 有效期 24 小时，而 3D 产物下载链接仅 2 小时，集成 3D 时务必在回调或轮询成功后立即下载转存。 -4. **计费模型差异**：图像按成功输出图计费且有免费额度、门槛最低；视频按时长计费、并发受限、成本更高；3D 按成功任务计数。批量或高并发场景需据此估算成本与吞吐。 -5. **接口路径不统一**：三大能力乃至同一能力内不同模型的接口路径都可能不同（尤其图像与视频），接入前务必以对应模型的官方文档为准。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md deleted file mode 100644 index aa5c3e48..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md +++ /dev/null @@ -1,72 +0,0 @@ -# 图像生成 vs 视频生成 vs 3D生成对比 - -百炼平台提供图像生成、视频生成和 3D 模型生成三大多媒体内容创作能力。三者在输入输出形态、调用模式、模型生态和适用场景上存在显著差异。本文从开发者技术选型角度，系统对比三类生成 API 的核心维度，帮助快速定位最适合业务需求的能力。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -| --- | --- | --- | --- | -| **输入格式** | 文本 [prompt](../guides/prompt.md)、参考图（URL/Base64）、蒙版、涂鸦草图 | 文本 [prompt](../guides/prompt.md)、首帧/首尾帧图像、参考图、音频、源视频 | 文本 [prompt](../guides/prompt.md)、单图（URL）、多图（4视角，前左后右） | -| **输出格式** | 图片 URL（JPEG/PNG），需及时下载 | 视频 URL（MP4），需及时下载 | GLB 模型文件 URL（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| **调用模式** | 同步（OpenAI 兼容）或异步（DashScope 原生） | 仅异步（创建任务 + 轮询 task_id） | 仅异步（创建任务 + 轮询 task_id） | -| **典型耗时** | 秒级（同步）~ 十几秒（异步） | 1-10 分钟 | 数分钟（建议 15 秒间隔轮询） | -| **API 端点** | `/compatible-mode/v1/images/generations` 或 `/api/v1/services/aigc/text2image/image-synthesis` | `/api/v1/services/aigc/video-generation/video-synthesis` 或 `/image2video/video-synthesis` | `/api/v1/services/aigc/video-generation/3d-generation` | -| **支持模型** | 通义千问图像、万相 V1/V2/2.6/2.7、Z-Image、可灵、创意工具（10+） | 万相 HappyHorse/wan2.7/2.6/2.5/2.2/2.1、爱诗 PixVerse、Vidu、可灵 | Tripo-H3.1（高精度）、Tripo-P1.0（专业快速） | -| **模型数量** | 20+ 模型/接口 | 30+ 模型变体 | 2 个模型 | -| **地域限制** | 无特殊限制 | 部分模型仅华北2（北京） | 仅华北2（北京） | -| **[计费](../concepts/billing.md)方式** | 按图片张数/分辨率[计费](../concepts/billing.md) | 按视频时长/分辨率[计费](../concepts/billing.md) | 按任务次数计费 | -| **协议支持** | OpenAI 兼容 + DashScope 原生 | DashScope 原生（异步必填 X-DashScope-Async 头） | DashScope 原生（异步必填 X-DashScope-Async 头） | -| **task_id 有效期** | 异步任务 24 小时 | 24 小时 | 24 小时 | -| **内容安全** | 内置审核，违规拒绝 | 内置审核，违规拒绝 | 遵循平台统一内容安全策略 | - -## 功能丰富度对比 - -| 能力类别 | 图像生成 | 视频生成 | 3D生成 | -| --- | --- | --- | --- | -| 文本生成 | 文生图（多模型可选） | 文生视频 | 文生 3D | -| 图像驱动 | 图像编辑、局部重绘、涂鸦作画 | 图生视频（首帧/首尾帧）、参考生视频 | 单图生 3D、多图生 3D | -| 风格控制 | 风格参数、负面词、艺术风格模型 | 视频风格重绘 | 贴图质量、几何精度 | -| 垂类工具 | 虚拟模特、AI试衣、创意海报、背景生成等 10+ 工具 | 视频换人、数字人、肖像动态（唱演/播报/口型替换） | 无 | -| 后处理 | 擦除补全、画面扩展、人物分割 | 视频编辑（指令+参考图） | 无 | - -## 适用场景建议 - -### 图像生成 - -- 电商商品图批量制作（虚拟模特、背景替换、创意海报） -- 营销素材快速迭代（文生图 + 风格控制） -- 内容创作中的插图与配图需求 -- 人像娱乐玩法（风格重绘、写真） -- 需要低延迟同步返回结果的场景 - -### 视频生成 - -- 短视频/广告创意自动化生产 -- 数字人播报与虚拟主播 -- 电商商品动态展示视频 -- IP 角色动画与多镜头叙事 -- 视频内容二次编辑与换人 - -### 3D生成 - -- 游戏/XR 场景中的 3D 资产快速原型 -- 电商商品 3D 展示与 AR 试用 -- 建筑/工业设计概念验证 -- 需要 PBR 材质的高精度渲染场景 - -## 技术选型建议 - -1. **追求响应速度**：图像生成支持同步调用，秒级返回；视频和 3D 均为异步，分钟级等待不可避免。 -2. **模型生态丰富度**：图像生成 > 视频生成 > 3D生成。图像生成模型和垂类工具最多，3D 目前仅有 Tripo 系列。 -3. **地域部署灵活性**：图像生成地域限制最少；视频和 3D 部分模型仅限华北2（北京）。 -4. **业务垂类覆盖**：电商/营销场景，图像生成的垂类工具链最完善；数字人/播报场景选视频生成；3D 资产生产选 3D 生成。 -5. **[多模态](../concepts/multimodal.md)组合**：可先用图像生成产出关键帧，再送入视频生成做动态化；或先用图像/多视角图生成 3D 模型，形成完整的内容生产流水线。 -6. **成本控制**：图像生成单次成本最低，3D 生成单次成本最高但产出资产复用价值大。建议根据产出资产的复用频次评估 ROI。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md deleted file mode 100644 index ce1262d9..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md +++ /dev/null @@ -1,69 +0,0 @@ -# 长期记忆、记忆库与[知识库](../concepts/knowledge-base.md)对比 - -百炼平台为开发者提供了三类用于"补充模型上下文"的能力：**长期记忆（新）API**、**记忆库（Memory Library）**、**[知识库](../concepts/knowledge-base.md)（Knowledge Base）**。三者定位不同：长期记忆 API 是底层的 RESTful 接口集合；记忆库是建立在长期记忆 API 之上、面向"跨会话用户记忆"的产品化封装（含控制台管理、OpenClaw 插件零侵入接入）；[知识库](../concepts/knowledge-base.md)则基于 RAG 技术，面向"私有文档/数据的语义检索"。本页通过关键维度对比，帮助开发者根据业务诉求（用户画像 vs 文档问答 vs 零侵入记忆）做出技术选型。 - -## 关键维度对比 - -| 维度 | 长期记忆（新）API | 记忆库（Memory Library） | 知识库（Knowledge Base） | -| --- | --- | --- | --- | -| 定位 | 底层 RESTful API，存储/检索/更新/删除用户记忆片段与画像 | 长期记忆 API 的产品化封装，含控制台与 OpenClaw 插件 | 基于 RAG 的私有数据语义检索，为模型补充领域知识 | -| 输入格式 | `messages`（对话）或 `custom_content`（自定义文本，≤512 字符） | 同长期记忆 API；OpenClaw 插件自动捕获对话 | 文件（pdf/docx/xlsx/图片/音视频等）、数据表、OSS 导入 | -| 输出格式 | 记忆片段（`memory_nodes`）、用户画像（结构化属性） | 记忆片段 + 用户画像，可注入 Prompt | 召回的文本切片（≤20 个/次），拼装后注入 Prompt | -| API 端点 | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` | 复用长期记忆 API；OpenClaw 插件通过 Gateway 钩子调用 | 阿里云百炼 SDK / 检索 API（需 AliyunBailianDataFullAccess 权限） | -| 支持模型 | 不直接绑定模型；记忆提取与画像由服务端完成 | 同长期记忆 API；OpenClaw 插件不支持 Coding Plan [API Key](../concepts/api-key.md) | 预置千问系列、DeepSeek-R1/V3.1、abab6.5s、Llama3.1、Yi-Large 及自定义模型 | -| 数据存储 | 记忆片段与画像，按 `user_id` 隔离，按 `memory_library_id` 分库 | 同长期记忆 API；默认库预置"默认有效期 180 天"规则 | 向量索引（text-embedding-v3/v4 512 维；multimodal-embedding-v1 1024 维） | -| 计费方式 | 按 API 调用计费（具体见平台说明） | 同长期记忆 API | 规格费用（按小时）+ 模型调用费用（[Token](../concepts/token.md)）；标准版 0.03 元/小时，旗舰版 0.2 元/RCU/小时 | -| 并发/限流 | 全部接口 ≤3000 QPM；add 120 QPM；search 300 QPM | 同长期记忆 API | 标准版 1 QPS（固定）；旗舰版 50–10,000 QPS（1–200 RCU） | -| 有效期 | API 直写：暂无失效日期 | 控制台规则：7/30/180 天或永不过期（默认 180 天） | 持久存储，无失效概念 | -| 接入方式 | 直接调用 RESTful API；Python 可用 `agentscope-runtime` | API 直连 / 百炼控制台 / OpenClaw 插件（零侵入） | 控制台创建 + SDK 集成；可挂载到[智能体应用](../concepts/agent-application.md)、工作流应用 | -| 典型场景 | 跨会话个性化、用户偏好持久化、自动提取关键事件 | 用户长期记忆、画像维护、Agent 零侵入记忆接入 | 私有文档问答、领域知识检索、图文/音视频内容搜索 | - -## 各方案适用场景建议 - -### 长期记忆（新）API - -适合需要**精细控制记忆生命周期**的开发者：自行管理写入（AddMemory）、语义检索（SearchMemory）、更新与删除，并通过画像模板（Profile Schema）维护结构化用户属性。当业务需要将记忆能力嵌入自有应用、对 `user_id` 与 `memory_library_id` 做多租户隔离、或希望用 `custom_content` 直接写入指定记忆（绕过对话提炼）时，优先选择此 API。 - -### 记忆库（Memory Library） - -适合希望**以最低接入成本获得跨会话记忆**的场景： - -- **OpenClaw 插件方式**：通过 `before_agent_start`（自动召回）与 `agent_end`（自动捕获）两个 Gateway 钩子实现零侵入记忆，所有提炼、向量化、语义检索由百炼服务端完成。适合基于 OpenClaw 构建的 Agent，无需改动业务代码。 -- **控制台方式**：在百炼控制台可视化管理记忆库与规则，支持配置记忆片段有效期（7/30/180 天或永不过期），适合非技术运营人员参与记忆策略管理。 -- **API 直连方式**：与长期记忆 API 一致，适合自定义接入。 - -注意：OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置。 - -### 知识库（Knowledge Base） - -适合需要**让模型基于私有数据回答问题**的 RAG 场景： - -- 文档问答（pdf/docx/markdown/图片等，单文件最大 150MB） -- 数据查询（xlsx 数据表，最大 10 万行） -- 图片问答（multimodal-embedding-v1） -- 音视频搜索（最大 512MB） - -知识库仅在**中国站华北2（北京）**地域可用，提供标准版（1 QPS、≤100 GB）与旗舰版（50–10,000 QPS、≤9,999 GB）两档规格。创建后知识库类型、metadata 抽取与切片策略不可更改，需一次性规划。适合企业知识库、产品手册问答、领域知识检索等"知识供给"型应用。 - -## 技术选型参考 - -| 选型问题 | 推荐方案 | -| --- | --- | -| 需要记住用户偏好、历史事件，实现跨会话个性化 | 长期记忆 API 或记忆库 | -| 基于 OpenClaw 构建 Agent，希望零侵入接入记忆 | 记忆库（OpenClaw 插件） | -| 需要运营人员在控制台管理记忆规则与有效期 | 记忆库（控制台） | -| 需要精细控制记忆的增删改查与画像模板 | 长期记忆（新）API | -| 需要基于私有文档/数据做语义检索问答 | 知识库 | -| 需要处理图片/音视频等多模态内容检索 | 知识库（图片问答/音视频搜索） | -| 高并发检索（>1 QPS） | 知识库旗舰版（最多 10,000 QPS） | -| 同时需要"用户记忆"和"文档知识" | 记忆库 + 知识库组合使用，二者互补 | - -**一句话总结**：长期记忆 API 与记忆库解决"记住用户是谁、做过什么"的问题（个性化上下文），知识库解决"模型不知道的领域知识"的问题（RAG 检索）。二者并不互斥，可在同一应用中组合使用——用记忆库维持用户画像与历史，用知识库供给领域文档，共同提升大模型在特定业务中的表现。 - -## 被对比主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) -- [knowledge base](../guides/knowledge-base.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md deleted file mode 100644 index a5963c3b..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md +++ /dev/null @@ -1,59 +0,0 @@ -# 记忆库与长期记忆对比 - -百炼平台为解决大模型跨会话上下文丢失的问题，提供了围绕"长期记忆"的完整能力。在文档体系中，这一能力以两种视角呈现： - -- **记忆库（Memory Library）**：偏产品与方案视角，强调接入方式（控制台 / HTTP API / OpenClaw 插件）、记忆片段与用户画像两类内容形态，以及记忆规则与有效期等可运营配置。 -- **长期记忆（新）API**：偏接口与实现视角，给出 RESTful 端点、请求/响应字段、限流策略和画像模板（Profile Schema）的完整 CRUD。 - -二者底层共用同一套 `https://dashscope.aliyuncs.com/api/v2/apps/memory/` 服务，记忆片段与用户画像的数据模型一致；区别在于"封装层"和"可控粒度"。本文从开发者技术选型角度对两者进行对比。 - -## 关键维度对比 - -| 维度 | 记忆库（Memory Library） | 长期记忆（新）API | -| --- | --- | --- | -| 定位 | 上层方案与产品形态：跨会话记忆的整体接入 | 底层 RESTful 接口参考：程序化记忆管理 | -| 文档位置 | 应用使用指南（guides） | 应用 API 参考（api） | -| 接入方式 | 控制台可视化管理 + HTTP API + OpenClaw 插件零侵入 | 直接 HTTPS 调用 `Authorization: Bearer $DASHSCOPE_API_KEY` | -| 记忆内容 | 记忆片段 + 用户画像（两类可独立或组合使用） | 记忆片段 + 用户画像（同一数据模型） | -| 写入输入 | `messages`（自动提取）或 `custom_content`（直写，最大 512 字符），二选一 | 同左，`messages` 最多 50 条对话 | -| 检索能力 | `SearchMemory`，主要参数 `top_k`（建议 3–10） | `SearchMemory`，含 `top_k`（1–100，默认 10）、`min_score`（默认 0.3）、`enable_rerank`/`enable_judge`/`enable_rewrite`、`project_ids` 多规则混合检索 | -| 管理 API | 重点呈现 `AddMemory` / `SearchMemory`（写入与召回） | `AddMemory` / `SearchMemory` / `ListMemory` / `UpdateMemory` / `DeleteMemory` + 画像模板 CRUD + `GetUserProfile` | -| 画像管理 | 通过 `profile_schema` 参数提取画像，模板在记忆库详情页获取 | 提供 `CreateProfileSchema` / `ListProfileSchemas` / `UpdateProfileSchema` / `DeleteProfileSchema` / `GetProfileSchema` 全套接口 | -| 记忆有效期 | 控制台默认规则 180 天，可配 7/30/180 天或永不过期；按规则可编辑 | API 文档标注"生成的记忆片段与用户画像暂无失效日期"（以控制台记忆规则配置为准） | -| 记忆规则 | 每账号自带默认记忆库 + 默认规则（不可删除），可创建新记忆库与规则 | 通过 `memory_library_id`、`project_id` 参数指定记忆库与规则；不传使用默认 | -| 用户隔离 | `user_id` 命名空间隔离，OpenClaw 插件所有 Agent 共享同一记忆 | `user_id` 最大 64 字符，用于标识记忆归属 | -| 限流 | 未单独列出，复用底层 API 限额 | 全部接口合计 ≤ 3000 QPM；`add` 120 QPM；`search` 300 QPM | -| 客户端 SDK | Python `agentscope-runtime`（`AddMemory` / `SearchMemory` 等封装，需 `close()`） | 以 cURL / REST 为主，参数与字段为权威定义 | -| 插件支持 | OpenClaw `modelstudio-memory-for-openclaw`：`before_agent_start` 自动召回 + `agent_end` 自动捕获 | 不直接涉及，插件内部回调此 API | -| 典型场景 | 跨会话个性化、Agent 偏好记忆、控制台运营记忆规则、OpenClaw 零侵入接入 | 程序化记忆 CRUD、画像模板生命周期管理、批量召回与重排序、自建记忆编排 | - -## 适用场景建议 - -### 选择"记忆库（Memory Library）"视角当 - -- 需要在控制台可视化地创建记忆库、配置记忆片段规则与有效期（7/30/180 天或永不过期）。 -- 希望以"产品方案"形式接入，例如通过 OpenClaw 插件实现 `autoCapture` / `autoRecall` 的零侵入跨会话记忆，而不愿手写每轮的写入与检索调用。 -- 业务侧关注的是"用户偏好持续化""Agent 跨会话理解"等整体能力，而非单个接口字段。 - -### 选择"长期记忆（新）API"视角当 - -- 需要在自研应用中精细控制每一步记忆操作：写入、搜索、列表、更新、删除，以及对画像模板做完整的增删改查。 -- 需要使用高级检索参数（`min_score` 阈值、`enable_rerank` 重排序、`enable_judge` 意图判别、`enable_rewrite` query 重写、`project_ids` 多规则混合检索）来调优召回质量。 -- 需要依据明确的限流（3000 QPM 总量、add 120 QPM、search 300 QPM）做容量规划与重试策略。 -- 需要程序化维护用户画像模板的字段定义，或对接已有用户体系做批量画像写入与读取。 - -## 技术选型小结 - -记忆库与长期记忆（新）API 并非二选一的两套系统，而是**同一能力的产品层与接口层**： - -- 做方案设计与运营配置时，以"记忆库"文档为准（接入方式、记忆规则、有效期、OpenClaw 插件配置）。 -- 做接口对接与字段实现时，以"长期记忆（新）API"文档为准（端点、参数、返回结构、限流、画像模板 CRUD）。 - -实践建议：先用记忆库视角确定接入形态（控制台 / API / 插件）与记忆规则，再在长期记忆（新）API 中查证具体端点与字段；两者配合即可覆盖从产品方案到代码实现的完整链路。注意记忆有效期以控制台记忆规则配置为准——API 文档的"暂无失效日期"指 API 直写且不指定 `project_id` 时使用默认规则的情形。 - -## 被对比主题页 - -- [memory library overview](../guides/memory-library-overview.md) -- [long term memory new](../api/long-term-memory-new.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md deleted file mode 100644 index e6d71fca..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md +++ /dev/null @@ -1,43 +0,0 @@ -# 记忆能力对比（长期记忆 vs 记忆库） - -百炼平台提供两种面向"跨会话上下文持久化"的记忆能力描述入口：一是 **长期记忆（新）**（API 参考视角，`api/long-term-memory-new.md`），二是 **记忆库（Memory Library）**（用户指南视角，`guides/memory-library-overview.md`）。两者本质指向同一套底层长期记忆 API（`https://dashscope.aliyuncs.com/api/v2/apps/memory/*`），但在文档定位、接入方式、管理入口和适用对象上存在差异。本页从技术选型角度对二者进行对比，帮助开发者快速判断应参考哪一份文档、采用哪种接入路径。 - -## 关键维度对比 - -| 维度 | 长期记忆（新）（API 参考） | 记忆库（Memory Library）（用户指南） | -| --- | --- | --- | -| 文档定位 | RESTful API 接口参考，逐接口说明请求/响应字段 | 能力总览与接入指南，含概念、控制台管理与插件接入 | -| 受众 | 直接调用 HTTP API 的后端/服务端开发者 | 需要端到端方案选型、含控制台与零侵入接入的开发者 | -| Base URL | `https://dashscope.aliyuncs.com/api/v2/apps/memory/` | 同上（`https://dashscope.aliyuncs.com/api/v2/apps/memory/*`） | -| 认证方式 | Header `Authorization: Bearer $DASHSCOPE_API_KEY` | 环境变量 `DASHSCOPE_API_KEY`，同样以 `Bearer` 方式携带 | -| 接口覆盖 | AddMemory / SearchMemory / ListMemory / DeleteMemory / UpdateMemory / CreateProfileSchema / ListProfileSchemas / DeleteProfileSchema / UpdateProfileSchema / GetProfileSchema / GetUserProfile 共 11 个 | 重点讲 AddMemory / SearchMemory，并补充 ListMemory、CreateProfileSchema、GetUserProfile 等封装类 | -| SDK 支持 | 以 cURL 示例为主 | 额外提供 Python `agentscope-runtime` 封装类，需在 `finally` 调 `close()` | -| 零侵入接入 | 未涉及 | 提供 OpenClaw 记忆插件，`before_agent_start`/`agent_end` 钩子自动召回/捕获 | -| 控制台管理 | 未涉及 | 支持控制台可视化管理记忆库、记忆规则、默认有效期 | -| 记忆有效期 | 明确"生成的记忆片段与用户画像暂无失效日期" | 控制台默认规则 180 天，可配置 7/30/180 天或永不过期；以控制台规则为准 | -| 限流说明 | 给出账号级 QPM：全局 3000、add 120、search 300 | 未在总览中给出 QPM 数字 | -| 检索增强参数 | 列出 `top_k`/`min_score`/`enable_rerank`/`enable_judge`/`enable_rewrite` | 仅点出 `top_k`、`minScore`（插件）等关键参数 | -| 画像能力 | 完整的 Profile Schema CRUD 与 GetUserProfile 接口 | 介绍画像模板概念与 `profileSchema` 配置项，接口细节指向 API 参考 | -| [计费](../concepts/billing.md)方式 | 未在本页说明 | 未在本页说明（统一走 DashScope [计费](../concepts/billing.md)） | -| 典型场景 | 需要精细控制记忆 CRUD、画像模板、检索召回参数的服务端集成 | 需要快速接入、可视化运维或让 OpenClaw Agent 自动具备记忆 | - -## 适用场景建议 - -- **参考"长期记忆（新）"文档的情况**：你需要直接对接 HTTP API，关心每个接口的请求体、响应字段、`event` 事件类型（ADD/UPDATE/DELETE）、画像模板的完整 CRUD，或需要按 `enable_rerank`/`enable_judge`/`enable_rewrite` 等参数精细调优检索召回；适合自研 Agent 后端、需要严格接口契约的服务端开发者。 -- **参考"记忆库"文档的情况**：你希望先从业务视角理解"记忆片段 vs 用户画像"两类持久化内容的差异与组合用法，或希望通过控制台创建/编辑记忆库与记忆规则、配置有效期，又或者你的 Agent 运行在 OpenClaw 之上，希望以插件方式零侵入获得"自动捕获/自动召回"能力；适合做整体方案选型与低代码运维的团队。 -- **二者结合使用**：多数生产落地建议先读"记忆库"理解概念与管理入口，再用"长期记忆（新）"对照接口字段落地代码；OpenClaw 用户可仅依赖插件配置项即可跑通，深定制时再回查 API 参考。 - -## 技术选型参考 - -1. 接入路径：纯后端 HTTP 调用 → 选 API 直连（两份文档均适用，接口细节以"长期记忆（新）"为准）；OpenClaw Agent → 选记忆插件（仅"记忆库"文档覆盖）。 -2. 有效期策略：若需记忆按天失效，务必以控制台记忆规则配置为准（默认 180 天）；API 直写且不指定 `project_id` 时走默认规则，"暂无失效日期"的说法仅适用于不经过规则提炼的直写场景。 -3. 命名空间隔离：两份文档均强调 `user_id` 为记忆空间隔离维度，不同 `user_id` 完全隔离；OpenClaw 插件目前所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置，也不支持百炼 Coding Plan 的 API Key。 -4. 检索质量：需要重排序、意图判别、query 重写等高级召回能力时，参考"长期记忆（新）"的 SearchMemory 参数；插件场景受 `topK`/`minScore` 配置项约束。 -5. 画像存储：需要固定结构化属性（年龄、职业、偏好等）持久化时使用用户画像，字段命名应清晰具体、避免同义并存；接口细节走"长期记忆（新）"的 Profile Schema 系列。 - -## 被对比主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md deleted file mode 100644 index 61ec6040..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md +++ /dev/null @@ -1,43 +0,0 @@ -# 模型直调与应用调用对比 - -百炼平台对外提供两类 API 调用路径：**模型直调**（直接调用 Qwen 等文本生成模型）与**应用调用**（调用在控制台预先编排好的智能体、工作流、Agent 2.0 应用）。两者底层都走 DashScope 网关并复用同一套 [API Key](../concepts/api-key.md)，但在调用对象、输入格式、能力范围与典型场景上有显著差异。本页面向做技术选型的开发者，按关键维度对比两种方案，帮助快速判断应走哪条路径。 - -## 关键维度对比 - -| 维度 | 模型直调（Qwen API） | 应用调用（Application API） | -| --- | --- | --- | -| 调用对象 | 单个文本生成模型（`model` 字段指定 Qwen 系列模型名） | 控制台已编排好的应用（智能体 / 工作流 / Agent 2.0），由 `APP_ID` 标识 | -| 核心入口 | OpenAI 兼容 Chat Completions / Responses、Anthropic 兼容 Messages、DashScope 原生 | OpenAI 兼容 Responses API（`/apps/agent/{APP_ID}/compatible-mode/v1/responses`）、DashScope 原生 API（`/apps/{APP_ID}/completion`） | -| 前置准备 | [API Key](../concepts/api-key.md)（`DASHSCOPE_API_KEY`） | [API Key](../concepts/api-key.md) + 应用 ID（`APP_ID`，控制台手动获取；子[业务空间](../concepts/workspace.md)还需 `Workspace ID`） | -| 输入格式 | `messages` 数组（OpenAI/Anthropic 兼容）或 Responses 的 `input`；需自行维护对话历史（Responses 接口除外） | `input.prompt` 字符串 / `input` 消息数组（OpenAI 兼容）；`input.prompt` + `parameters` + `biz_params`（DashScope 原生） | -| 对话历史 | 仅 OpenAI 兼容 Responses 由平台自动管理；其余接口调用方自行拼接 `messages` | 支持 `session_id`（云端托管，1 小时有效、最多 50 轮）或自行管理 `messages`；同时传两者时以 `messages` 为准 | -| 内置工具能力 | 联网搜索、代码解释器、网页内容提取仅 Responses 接口内置；其他接口需自行定义工具 | 应用编排阶段在控制台绑定插件 / 节点，调用时通过 `biz_params.user_defined_params` 透传业务参数，工具由应用内部装配 | -| 支持模型 | Qwen 全系列（含 VL 多模态），按 `model` 字段切换 | 由应用编排时所选模型决定（如智能体做图像输入需选 Qwen-VL 并设「自定义处理」） | -| [流式输出](../concepts/streaming-output.md) | 各接口均支持 `stream` 参数 | 同步调用支持 `stream=True`；[异步调用](../concepts/async-invocation.md)暂不支持流式；工作流需在输出节点启用「[流式输出](../concepts/streaming-output.md)」并重新发布 | -| 异步执行 | 由调用方自行实现 | OpenAI 兼容 Responses 支持 `background=True`，返回任务 ID 后轮询 `retrieve` 至终态 | -| 多模态 | 各接口按协议支持文本 / 图像 / 文件（`input_image` / `input_file`） | `input` 消息数组支持 `input_text` / `input_image` / `input_file`（`input_file` 仅[智能体应用](../concepts/agent-application.md)支持） | -| 计费方式 | 按 token 用量计费（输入 + 输出） | 按 token 用量计费；应用编排内部多步调用累计计费 | -| 功能完整度 | DashScope 原生接口参数最全；兼容接口为保证协议一致可能不暴露全部原生参数 | DashScope 原生 `/completion` 功能更全、性能更高；[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)便于复用 OpenAI 生态 | -| 典型场景 | 单模型推理、文本生成、对话补全、从 OpenAI/Anthropic 平迁、需要精细采样参数 | 复用已编排的多步骤智能体 / 工作流、内置 RAG 与插件、长耗时异步任务、业务系统集成 | - -## 选型建议 - -- **选模型直调**：当你只需要一个模型做单轮或多轮文本生成、补全、对话，且希望直接复用 OpenAI / Anthropic SDK 与既有代码库，或需要最全的采样参数与原生能力时。从外部平台迁入时优先评估兼容接口，需要极致功能时再切到 DashScope 原生接口。 -- **选应用调用**：当业务逻辑已被编排成智能体或工作流（含 RAG、插件、多节点流程），希望把整套能力一次性集成进业务系统，而不是在调用方重新实现编排逻辑时。长耗时任务（报告生成、多步工具调用）选异步 `background=True`；需要复用 OpenAI 工具链选 Responses API，需要更全功能与更高性能选 DashScope 原生 `/completion`。 -- **混合使用**：两者共用同一 `DASHSCOPE_API_KEY`，可在同一业务系统中并存——轻量推理走模型直调，复杂编排走应用调用，按场景而非按模型选路径。 - -## 注意事项 - -- **凭证管理**：API Key 推荐写入 `DASHSCOPE_API_KEY` 环境变量，不要在生产环境硬编码。 -- **应用 ID 获取**：`APP_ID` 与 `Workspace ID` 目前只能在控制台手动复制，不支持 API / CLI 查询；RAM 子账号默认只能查看其已加入的[业务空间](../concepts/workspace.md)。 -- **地域差异**：上述 Endpoint 默认适用于华北2（北京）；德国（法兰克福）、新加坡、日本（东京）等地域或调用子[业务空间](../concepts/workspace.md)下应用时，请求须包含 `Workspace ID`，且该 ID 是对应地域 Base URL 的组成部分。 -- **兼容接口的功能取舍**：OpenAI / Anthropic 兼容接口为保证协议一致性，可能不暴露百炼原生全部参数；如需最全参数、插件或业务字段，应改用 DashScope 原生接口。 -- **迁移评估**：从 OpenAI / Anthropic 迁移到模型直调时，先确认目标 Qwen 模型在对应兼容接口下是否支持所需参数（`temperature`、`tools`、`stream` 等）；迁移到应用调用时，注意异步不支持流式、多轮历史需自行管理（除非用 `session_id`）。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md deleted file mode 100644 index af15d86e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md +++ /dev/null @@ -1,64 +0,0 @@ -# 模型微调、压缩与部署对比 - -百炼平台围绕“把通用大模型变成业务专属模型”提供了一条递进式链路：**模型微调（Fine Tuning）→ 模型压缩（Model Compression）→ [模型部署](../concepts/model-deployment.md)（Model Deployment）**。三者并非并列的替代方案，而是同一交付链上前后衔接的环节——微调负责把业务/场景知识写入模型参数，压缩负责把全精度微调模型量化为低精度版本以降低部署门槛，部署负责把模型上线为可调用的推理服务。本文从输入格式、输出产物、支持模型、操作入口、计费方式、典型场景等维度做横向对比，帮助开发者明确各环节的边界与衔接关系，避免在选型时混淆“训练 / 压缩 / 部署”三件事。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine Tuning） | 模型压缩（Model Compression） | [模型部署](../concepts/model-deployment.md)（Model Deployment） | -| --- | --- | --- | --- | -| 核心目标 | 将业务/场景知识、风格、偏好写入模型参数 | 将全精度微调模型量化为低精度版本，降低显存与部署成本 | 将预置或调优后的模型上线为独立、资源专享的推理服务 | -| 在链路中的位置 | 链路起点，产出可被压缩/部署的微调模型 | 微调之后、部署之前的可选优化环节 | 链路终点，对外提供推理 API | -| 输入格式 | CPT：1000 万+ Token 无标签领域文本；SFT：1000+ 条问-答对；DPO：100+ 组偏好对；万相图像/视频：`.zip` 内含 `data.jsonl` 与素材；CosyVoice：多条/数小时录音 | 当前工作空间内已有的自定义微调模型；量化模板要求时还需校准数据（内部上传且已发布的数据集） | 模型 ID（预置模型或调优/压缩产出模型）；API/命令行调用还需 API Key 与业务空间 | -| 输出产物 | 自定义微调模型（全参或 LoRA），可用于压缩或直接部署 | 量化后的低精度模型，落盘到「我的模型」，可直接部署 | 在线推理服务（`deployed_model` 唯一 ID，`status=RUNNING`） | -| 支持模型 | 文本：Qwen3.x / Qwen2.5 系列等；视觉：Qwen3-VL / Qwen2.5-VL；图像：wan2.7；视频：wan2.2/wan2.5；语音：cosyvoice-v3-flash | 仅「支持压缩的模型」列表中的自定义微调模型（如 qwen3.5-flash-2026-02-23 对应的微调版）；已量化模型不可二次压缩 | 预置模型、调优产出模型、压缩产出模型、OSS 导入的 LoRA 模型（仅 LoRA，不支持全参微调模型导入） | -| 操作入口 | 控制台「模型调优」页面；或 DashScope API `POST /api/v1/fine-tunes` | 控制台「模型训练 > 模型压缩」页面；或 OpenAPI | 控制台「[模型部署](../concepts/model-deployment.md)」页面（北京）；或 `https://dashscope.aliyuncs.com/api/v1/deployments` | -| 计费方式 | 按 Token 计费（API 创建的任务仅支持按 Token，不支持训练单元）；控制台任务可用训练单元（预付费/后付费）；计费与 `n_epochs`/`batch_size`/`max_length` 等超参相关 | 压缩功能限时免费；产出模型部署后按所选部署单元规格计费 | 预置吞吐（PTU）、模型单元（MU）、按 Token 用量三种；PTU 支持长输入阶梯系数与前缀缓存折扣；服务创建后无法切换计费方式 | -| 可逆性 | 可重新训练、迭代 | 不可逆：产出模型不支持继续训练，也不支持二次压缩；如需迭代须回到上游全精度模型重新训练 | 可下线重新部署以切换计费方式；按 Token 用量模式一个月不使用自动释放 | -| 任务状态流转 | PENDING → SUCCEEDED（轮询 `GET /api/v1/fine-tunes/{job_id}`） | 待开始 / 排队中 / 运行中 / 停止中 / 压缩成功 / 压缩失败 / 已取消；仅排队中/运行中可取消 | PENDING → RUNNING（轮询 `GET /api/v1/deployments/{deployed_model}`） | -| 关键约束 | `-Base` 后缀模型不可直接用于调优对话；CosyVoice 调优仅支持 API 发起、无法扩展基础模型不支持的语种 | 校准数据控制台 UI 暂不支持 OSS 挂载类型；模板名称决定压缩后可部署规格（如 MU5、MU8） | 仅华北二（北京）地域；导入 LoRA 的 rank 须为 8/16/32/64 且各层一致；VL 模型须冻结 VIT | - -## 各方案适用场景建议 - -### 模型微调 - -适合“Prompt 工程已到上限、需要把知识或风格沉淀进模型参数”的场景： - -- **CPT（继续预训练）**：业务有大量无标注领域语料（专业词汇、行业事实），需要先做领域适应。 -- **SFT（监督微调）**：需要模型遵循特定对话格式或任务执行规范，有 1000+ 条高质量问-答对。 -- **DPO（直接偏好优化）**：SFT 之后想进一步对齐人类偏好、抑制坏答案。 -- **LoRA 高效训练**：数据集较小、需快速验证、追求低成本快迭代；全参训练则用于追求全局效果最优。 -- 多模态场景：文生图/图生图训练人物 IP 或风格 LoRA（万相 wan2.7）、图生视频训练首帧/首尾帧 LoRA（wan2.2/wan2.5）、语音合成训练专属音色（CosyVoice）。 - -### 模型压缩 - -适合“已有全精度微调模型，但希望部署到更小规格单元、降低成本并提升吞吐”的场景： - -- 微调产物部署成本偏高，想通过量化降到 MU5/MU8 等更小规格。 -- 对推理显存和吞吐有明确优化诉求，且不打算继续迭代模型参数（压缩不可逆）。 -- 不适合：尚未完成微调的模型、已量化过的模型、需要继续训练迭代的模型。 - -### 模型部署 - -适合“模型已就绪（无论预置、微调还是压缩产出），需要对外提供高并发、低延迟推理服务”的场景： - -- **PTU（预置吞吐）**：高负载生产环境、稳定吞吐、流量可预估，追求 TPS 提升（约 1.5~2.0 倍于按 Token）。 -- **模型单元（MU）**：资源独占、需自定义性能指标、长时任务；支持 PD 分离以降低首 Token 延迟。 -- **按 Token 用量**：调优后效果验证、性价比优先、对并发延迟要求不高；仅支持部分 LoRA 调优模型，一个月不使用自动释放。 -- **自定义 LoRA 导入**：本地训练的 LoRA 模型（千问3/千问2.5 系列及 VL 版本）需从 OSS 导入后再部署；注意推理参数建议参照 vLLM 默认值调整。 - -## 技术选型参考 - -1. **先判断是否需要动参数**：如果 Prompt 工程能解决，就不必微调；只有当知识/风格/偏好需要沉淀进模型时才进入微调环节。 -2. **微调模式选择**：快速验证用 LoRA，全局最优用全参；多模态按模型类型选 SFT-LoRA；语音合成走 CosyVoice SFT。 -3. **是否插入压缩环节**：微调产物部署成本敏感时，在部署前加一步压缩；但需注意压缩不可逆、不支持二次压缩，迭代须回到全精度模型重训。 -4. **部署计费方式选择**：生产高负载选 PTU（兼顾长输入与前缀缓存优惠）；资源独占与 PD 分离选模型单元；仅做效果验证选按 Token 用量。计费方式创建后不可切换，需下线重部署。 -5. **链路衔接**：微调 →（可选）压缩 → 部署 是单向链路；压缩产出的模型不可继续训练，所以“想压缩又想继续迭代”的需求要把全精度模型作为迭代基线，每次迭代后重新压缩。 -6. **入口选择**：零代码/合规场景走控制台；需要自动化编排（如 CI/CD）走 API/命令行；CosyVoice 调优目前仅支持 API，控制台不可用。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md deleted file mode 100644 index cca2e06a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md +++ /dev/null @@ -1,43 +0,0 @@ -# 微调、压缩与部署方案对比 - -百炼平台围绕「让模型贴合业务」提供了三条相互衔接的路径：**模型调优（Fine Tuning）**把业务/场景知识写进模型参数；**模型压缩（Model Compression）**用量化算法降低微调模型的精度与显存占用；**模型部署（Model Deployment）**把平台预置模型或调优后的模型部署为在线推理服务。三者构成 `调优 → （可选）压缩 → 部署 → 调用` 的链路，本文从输入产出、支持模型、操作入口、计费方式、典型场景等维度做横向对比，帮助开发者根据「是否需要写知识」「是否要降本」「并发与延迟要求」做技术选型。 - -## 关键维度对比 - -| 维度 | 模型调优（Fine Tuning） | 模型压缩（Model Compression） | 模型部署（Model Deployment） | -| --- | --- | --- | --- | -| 一句话定位 | 将业务/场景知识写入模型参数，压低延迟、抑制幻觉、对齐偏好 | 将全精度微调模型量化为低精度版本，降低显存与部署成本 | 把预置或调优后的模型部署为资源专享的在线推理服务 | -| 输入 | 训练数据集：CPT 需 1000 万+ [Token](../concepts/token.md) 无标签文本；SFT 需 1000+ 条问答对；DPO 需 100+ 组偏好对；万相/CosyVoice 需 `.zip`（含 `data.jsonl` + 媒体素材） | 当前工作空间内符合条件的**自定义微调模型**（全精度）+ 可选校准数据集 | 平台预置模型、调优产出模型、OSS 导入的 LoRA 模型；部署参数（`plan`、`capacity`、`ptu_capacity` 等） | -| 输出 | 微调后的自定义模型（全参或 LoRA 产物），可继续部署或压缩 | 量化后的低精度模型，可直接用于部署 | 在线推理服务（`deployed_model`），返回 `status` 为 `RUNNING` 后即可调用 | -| 主要调优/处理方式 | CPT、SFT（全参/高效 LoRA）、DPO（全参/LoRA）；万相/CosyVoice 仅 `efficient_sft` | 量化模板（如 MU5、MU8），模板名决定压缩后可部署规格 | 预置吞吐（PTU）、模型单元（MU）、[Token](../concepts/token.md) 用量（lora）三种计费方式 | -| 支持模型 | 千问3.x/2.5 文本与 VL 系列、万相 wan2.7/wan2.2/wan2.5、CosyVoice-v3-flash 等 | 千问系列对应自定义微调模型（如 `qwen3.5-flash-2026-02-23`），以控制台列表为准 | 平台预置模型 + 部分调优模型；OSS 导入仅支持 LoRA，rank 须为 8/16/32/64 | -| 操作入口 | 控制台「模型调优」页面（零代码）或 DashScope API（`/api/v1/fine-tunes`） | 控制台「模型训练 > 模型压缩」页面或 OpenAPI | 控制台（北京）或 API/命令行（`/api/v1/deployments`） | -| 关键 API/参数 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`（`training_type`、`hyper_parameters`）、`GET /api/v1/fine-tunes/{job_id}`、`POST /api/v1/deployments` | 任务名称、源模型、量化产出后缀、量化模板、`custom_calibration_file_ids` | `POST /api/v1/deployments`（`plan`=`ptu`/`mu`/`lora`）、`GET /api/v1/deployments/{deployed_model}`、`DELETE /api/v1/deployments/{deployed_model}` | -| 计费方式 | 控制台任务支持模型训练单元（预付费/后付费）；API 任务仅按 [Token](../concepts/token.md) 计费，不支持训练单元 | 当前**限时免费**；产出模型部署后按所选部署单元标准规格计费 | PTU（时长×预置吞吐）、模型单元（时长×数量×单价，可包月）、Token 用量（输入/输出 Token×单价，不使用不计费） | -| 扩缩容 | 不适用（训练任务级） | 不适用（压缩任务级） | PTU/模型单元自助增减；Token 用量需控制台提交申请人工审核 | -| 是否可逆 | 可基于基础模型重新训练迭代 | **不可逆**：产出模型不支持继续训练或二次压缩，需回到上游全精度模型重训 | 可随时 `DELETE` 下线服务，但删除后不可恢复 | -| 典型场景 | 注入领域知识、复刻特定风格/IP、对齐人类偏好、专属音色训练 | 降低微调模型部署成本，部署到更小规格单元、提升吞吐 | 高并发低延迟生产环境、稳定吞吐、调优后效果验证、长时任务 | - -## 各方案适用场景建议 - -- **模型调优（Fine Tuning）**：当 Prompt 工程已无法满足延迟、幻觉或风格复刻要求，需要把业务/场景知识直接写进模型参数时选用。数据量充足（CPT 千万级 Token、SFT 千条级问答、DPO 百组级偏好对）、且希望模型在特定任务上达到全局最优时，优先考虑全参训练；数据量小、需快速验证或仅复刻风格（万相 LoRA、CosyVoice 专属音色）时，优先用 LoRA 高效训练。注意 API 任务仅按 Token 计费，需用训练单元请走控制台。 -- **模型压缩（Model Compression）**：当已有全精度微调模型、希望降低显存占用并部署到更小规格单元时选用。压缩功能当前限时免费，是把调优成果低成本上线的有效手段。务必注意产出模型**不可继续训练、不可二次压缩**，迭代需回到上游全精度模型；模板名（如 MU5、MU8）直接决定后续可部署规格，应在压缩阶段就规划好部署目标。 -- **模型部署（Model Deployment）**：当需要资源专享、高并发、低延迟或稳定吞吐的生产级推理服务时选用。计费方式一旦创建不可更改，需按业务特征提前选定：高负载、流量可预估、追求低延迟选 PTU（TPS 约为按 Token 的 1.5~2.0 倍）；资源独占、长时任务、需 PD 分离选模型单元；调优后效果验证、对并发延迟要求不高、追求高性价比选 Token 用量（仅部分 LoRA 模型支持，一个月不用自动释放）。 - -## 选型链路建议 - -三者并非互斥，而是上下游关系：先用**调优**把知识写进模型；若部署成本敏感，再用**压缩**把全精度产物量化到更小规格；最后用**部署**把模型（无论是否压缩）上线为推理服务。典型组合： - -1. **轻量风格定制**：SFT-LoRA（万相/CosyVoice）→ 部署（Token 用量，快速验证）。 -2. **生产级领域模型**：CPT → SFT（全参）→ 压缩（量化到 MU8）→ 部署（PTU 或模型单元，高并发低延迟）。 -3. **偏好对齐优化**：SFT → DPO → 压缩 → 部署（模型单元，PD 分离降首 Token 延迟）。 - -选型核心判断：**要写知识选调优，要降成本选压缩，要上线选部署**；若同时追求极致效果与极致成本，按 `调优（全参）→ 压缩 → 部署（PTU/MU）` 串联即可。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md new file mode 100644 index 00000000..a6d48a64 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md @@ -0,0 +1,69 @@ +# 模型部署方案对比：Model Production、High-Speed Inference 与 Fine-tuning + +本文旨在帮助开发者清晰区分百炼平台中三类核心模型服务能力——**Model Production（模型生产）**、**High-Speed Inference（高性能推理）** 与 **Fine-tuning（微调）**，明确其定位、能力边界、技术约束及适用阶段。三者并非互斥替代关系，而是构成「训练 → 优化 → 部署 → 加速」全链路的关键环节： +- **Fine-tuning** 解决「模型好不好用」——通过业务数据定制模型能力； +- **Model Production** 解决「模型能不能上线」——将训练成果封装为稳定、可扩缩的生产级服务； +- **High-Speed Inference** 解决「上线后快不快、稳不稳」——在服务已就绪前提下，保障高并发下的确定性吞吐与低延迟体验。 +正确理解三者差异，是避免资源错配、计费异常与架构返工的技术选型前提。 + +## 关键维度对比表 + +| 维度 | Model Production | High-Speed Inference | Fine-tuning | +|------|------------------|------------------------|-------------| +| **核心目标** | 将训练完成/导入的模型发布为可管理、可扩缩的在线推理服务 | 在已有模型服务基础上，提升吞吐量（TPM/TPS）或降低端到端延迟 | 基于自有数据对基座模型进行参数级优化，提升领域适配性与任务表现 | +| **输入格式** | 已训练完成的模型 ID（如 `ft-xxx`）、部署配置（`instance_type`, `deployment_name` 等） | 标准 API 请求（含 `model` 参数），需匹配预留 model code 或快速模式专属域名 | 训练数据集（JSONL / ZIP 包）、超参配置（`learning_rate`, `n_epochs`, `lora_rank` 等）、基础模型标识 | +| **输出格式** | 部署成功后返回 `endpoint_url`，调用该地址返回标准 OpenAI 兼容响应（如 `/v1/chat/completions`） | 同标准 API 响应结构；快速模式额外返回 `delta.reasoning_content` 字段；TPM 预留无结构变化 | 训练完成后生成新模型 ID（如 `ft-qwen3-8b-20240520-abc123`），**不直接提供推理接口**，需经 Model Production 部署后方可调用 | +| **支持模型** | 所有已完成微调（Fine-tuning）或手动导入的模型（ID 以 `ft-` 或 `import-` 开头）；**不支持原生基座模型直连部署** | TPM 预留：千问、GLM、DeepSeek、Kimi 等主流基座模型（含 `glm-5.2`）；
快速模式：**仅 `glm-5.2-fast-preview`**（Preview 阶段） | 文本（Qwen3/3.5）、视觉（万相 Wan2.7）、语音（CosyVoice-v3-flash）、视频（Wan2.7-i2v）、强化学习（Qwen3.5-9B）等多模态模型，覆盖 SFT、DPO、CPT、RL 等范式 | +| **API 端点** | `POST /v1/deployments`（创建部署）
`GET /v1/deployments/{name}`（查询状态）
`DELETE /v1/deployments/{name}`（下线） | TPM 预留：复用标准 API 域名（如 `https://dashscope.aliyuncs.com/...`），仅需替换 `model` 参数为预留 code；
快速模式：**必须使用专属域名**（如 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`） | `POST /api/v1/fine-tunes`（提交任务）
`GET /api/v1/fine-tunes/{job_id}`（轮询状态）
`POST /api/v1/files`（上传数据） | +| **计费方式** | **按实例规格 + 运行时长计费**（GPU 实例小时单价 × 实际运行秒数），部署期间持续计费；
微调作业单独计费（按 MTU 或 GPU 小时） | TPM 预留：**预付费**，按天购买 kTPM 容量（输入/输出独立计价），超量部分自动溢出至按量计费；
快速模式：**纯按量计费**，按实际输入/输出 token 计费（缓存命中享折扣） | **按训练资源消耗计费**：
- SFT/DPO/CPT：按 GPU 小时或 MTU（Model Training Unit）计费；
- RL：**强制使用 MTU 计费**；
微调作业超时（72h）自动终止且不计费 | +| **典型场景** | - 微调完成后的模型正式上线
- 多版本 A/B 测试（如 `prod-qa-bot-v1` vs `v2`）
- 需要自动扩缩容与健康检查的长期服务 | - 高流量客服机器人（需保障 99.9% 请求 < 2s）
- AI 编程助手（依赖快速[流式输出](../concepts/streaming-output.md)）
- 大促期间临时扩容（TPM 预留防抖） | - 金融合同条款解析（定制法律语义）
- 电商商品图风格统一（万相微调）
- 企业专属语音播报（CosyVoice 音色克隆）
- 数学推理 Agent（RL 强化策略） | + +## 各方案适用场景建议 + +### ✅ 推荐选择 **Fine-tuning** 当： +- 您拥有高质量、领域专属的标注数据（如客服对话、设计稿描述、医学报告）； +- 标准大模型在关键指标（准确率、风格一致性、专业术语召回）上未达业务要求； +- 您需要模型具备**不可迁移的知识或行为偏好**（如公司 SOP、品牌话术、IP 形象）； +- 您能接受 1~24 小时训练周期，并具备后续部署运维能力。 + +> ⚠️ 注意：Fine-tuning **不是**零样本/少样本推理替代方案。若仅需 [prompt](../guides/prompt.md) 工程优化，请优先使用 `inference` 模块。 + +### ✅ 推荐选择 **Model Production** 当： +- 您已完成 Fine-tuning 或已获得可部署模型（如第三方导出权重）； +- 您需要一个**生产就绪的服务实体**：具备唯一访问入口、自动扩缩容、健康探针、TLS 加密、权限隔离； +- 您需对多个模型版本进行生命周期管理（上线/下线/回滚）； +- 您希望规避手动维护 GPU 实例、负载均衡、监控告警等基础设施复杂度。 + +> ⚠️ 注意：Model Production **不提供训练能力**，也不支持 CPU 实例部署（当前仅 GPU 规格有效）。 + +### ✅ 推荐选择 **High-Speed Inference** 当： +- 您的模型服务**已通过 Model Production 上线并稳定运行**； +- 您面临明确的性能瓶颈：高并发下 TPS 不足、尾部延迟超标（P99 > 3s）、突发流量导致 429 错误频发； +- 您需要**SLA 保障**（TPM 预留）或**极致流式体验**（快速模式）； +- 您愿意为确定性容量或加速能力支付溢价（预付费或更高 token 单价）。 + +> ⚠️ 注意：High-Speed Inference 是**叠加在已部署服务之上的加速层**，无法独立存在；快速模式当前仅限 `glm-5.2-fast-preview`，不建议用于核心生产系统长期依赖。 + +## 技术选型参考（面向开发者） + +| 您的问题 | 推荐方案 | 关键依据 | +|----------|-----------|-----------| +| “我有一批销售话术数据，想让 Qwen3 更懂我们行业术语” | ✅ Fine-tuning | 需修改模型参数以注入领域知识，属训练范畴 | +| “微调好的模型怎么让前端调用？需要自己搭服务器吗？” | ✅ Model Production | 提供开箱即用的 HTTPS endpoint，免运维部署 | +| “上线后用户一多就卡顿，P95 延迟从 800ms 涨到 4s” | ✅ High-Speed Inference（TPM 预留） | 容量争抢导致排队，需专属资源保障吞吐 | +| “Agent 调用时输出太慢，用户等待感强” | ✅ High-Speed Inference（快速模式） | 针对[流式输出](../concepts/streaming-output.md)速度优化，TPS 提升 1.5~2 倍 | +| “能否用 CPU 部署低成本测试模型？” | ❌ Model Production（不支持）
✅ 可考虑 `inference` 模块（非本文对比项） | 当前 Model Production 仅接受 GPU 实例类型 | +| “想同时用 TPM 预留 + 快速模式，是否可行？” | ✅ 支持组合使用 | 在 TPM 预留的 `glm-5.2` 实例上，将 `model` 设为 `glm-5.2-fast-preview` 即可 | +| “微调时发现 `qwen3.5-9b` 不支持 efficient_sft？” | ✅ 以控制台/API 实际选项为准 | 文档存在表述差异，但实操中 `efficient_sft` 是主流推荐方式，兼容性优于全参训练 | + +> 💡 **最佳实践路径**： +> `Fine-tuning` → `Model Production` → `High-Speed Inference` +> 三者串联构成完整 MLOps 闭环。切勿跳过 Model Production 直接对微调模型做 TPM 预留（因预留对象必须是已部署的 `deployment_name` 或标准模型 ID）；也勿在未微调前过度投入 High-Speed Inference（基座模型能力不足时，加速无法解决根本效果问题）。 + +## 被对比主题页 + +- [model production](../api/model-production.md) +- [model high speed inference](../guides/model-high-speed-inference.md) +- [fine tuning](../guides/fine-tuning.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md deleted file mode 100644 index 048238f7..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md +++ /dev/null @@ -1,45 +0,0 @@ -# 模型部署与高速推理对比 - -在百炼平台上落地生产级推理服务时，开发者常面临两条技术路线：一是通过**模型部署**为预置或调优模型建立独立、资源专享的推理实例（PTU、模型单元、按 Token 三种计费）；二是通过**高速推理**能力（TPM 预留、快速模式 Fast mode）在标准调用之上叠加容量保障或输出提速。二者定位不同——前者关注「拥有一套专属推理服务」，后者关注「让现有模型调用更稳、更快」。本页梳理两者关键差异，帮助开发者按业务诉求做选型。 - -## 关键维度对比 - -| 对比维度 | 模型部署（PTU / MU / 按 Token LoRA） | 高速推理（TPM 预留 / 快速模式 Fast mode） | -| --- | --- | --- | -| 核心目标 | 建立独立、资源专享的推理服务，满足高并发、低延迟、私有模型落地 | 为已有模型调用锁定专属吞吐（TPM 预留）或提升输出速度（快速模式） | -| 支持模型 | 平台预置模型 + 调优/自训练模型（LoRA 导入，覆盖千问3/千问3-VL/千问2.5/千问2.5-VL） | TPM 预留：以控制台为准（千问3.x、GLM-5.x、DeepSeek-v4、Kimi-K2.6 等）；快速模式：有限模型如 `glm-5.2-fast-preview` | -| API 端点 / 接入 | 部署接口 `POST https://dashscope.aliyuncs.com/api/v1/deployments`，用 `plan` 区分计费；推理走 [DashScope SDK](../concepts/dashscope-sdk.md) 对专属服务调用 | TPM 预留：将 `model` 替换为专属模型 code，请求结构不变；快速模式：base_url 改为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`，`model` 指定 fast 版 | -| 计费方式 | PTU：时长×TPM 单价（后付费按小时/预付费按天）；MU：时长×单元数（后付费按分钟/预付费按月）；LoRA：按 Token 使用量 | TPM 预留：按 kTPM 预付费（部署成功即计费）；快速模式：按 token 计费，逻辑与标准 API 一致 | -| 计费切换限制 | 计费方式创建后不可更改，须下线重新部署 | TPM 预留可扩缩容/续费/退订（退费按已用 1.5 倍系数结算） | -| 超额处理 | PTU 超吞吐或超输入上限自动转按量计费（响应头 `x-dashscope-ptu-overflow:true`） | TPM 预留：自动降级公共池按量、不中断；快速模式：进入排队队列（不立即限流） | -| 输出速度 | PTU 相比按 Token TPS 提升约 1.5~2.0 倍；MU 支持 PD 分离降低首 Token 延迟 | 快速模式 TPS 提升至标准 API 的 1.5~2 倍（达 80~100 TPS） | -| 资源专享 | 是（资源独占，MU 性能可自定义） | TPM 预留：是（专属容量刚性兑付）；快速模式：否（按 token 共享，仅提速） | -| 代码改动 | 部署后用专属 `deployed_model` ID 调用 | TPM 预留/快速模式：替换 `model`（快速模式还需改 base_url） | -| 成熟度 | 正式能力（仅华北2·北京地域） | TPM 预留：正式；快速模式：preview 阶段，规格可能调整 | -| 典型场景 | 智能客服、实时内容审核、私有微调模型落地、长时任务 | TPM 预留：流量可预估、不能接受公共限流；快速模式：AI 编程助手、Agent 多步推理、实时对话 | - -## 适用场景建议 - -- **选模型部署（PTU）**：流量稳定、需要并发/延迟确定性的高负载生产环境；希望获得高于按量的 TPS，并利用长输入阶梯系数与前缀缓存折扣优化额度。建议部署前用控制台**容量计算器**估算所需 TPM。 -- **选模型部署（MU）**：需要部署私有微调（LoRA）模型、性能指标可自定义、或有长时任务，并希望用 PD 分离降低首 Token 延迟。 -- **选模型部署（按 Token / LoRA）**：仅用于验证 SFT 高效训练后的自定义模型效果，不使用不计费。 -- **选 TPM 预留**：不想新建独立部署实例，但流量可预估、无法接受被公共资源限流；需要「专属容量刚性兑付、超额自动降级不中断」的确定性保障。 -- **选快速模式（Fast mode）**：对输出速度（TPS）敏感、计费仍希望按 token 走标准逻辑的场景（编程助手、Agent、实时对话），可接受 preview 阶段的规格变动与独立接入域名。 - -## 技术选型参考 - -1. **先看诉求本质**：需要「一套专属服务 + 自定义/私有模型 + 计费确定性」→ 走**模型部署**；只想给现有模型「加容量保障或加速」→ 走**高速推理**。 -2. **计费柔性**：模型部署计费方式创建后不可改（须下线重建），选型前务必确认；TPM 预留支持不中断扩缩容，柔性更高。 -3. **超额行为要区分**：PTU 与 TPM 预留超额均转/降级为按量、服务不中断；而**快速模式超额是排队**，对延迟敏感业务需评估队列影响并做重试机制。 -4. **迁移成本**：模型部署与 TPM 预留仅需替换 `model`/`deployed_model` 参数；快速模式还需改 `base_url`（MaaS 域名），迁移时注意区分。 -5. **成熟度与地域**：模型部署当前限华北2·北京地域；快速模式处于 preview，能力可能随版本调整——正式生产建议优先选正式能力，并以百炼控制台展示的价格、容量起步值、支持模型为准。 -6. **可组合使用**：两条路线并非互斥。对于既要私有模型落地又要输出提速的场景，可分别评估 MU 部署与快速模式，按模型支持情况组合。 - -## 被对比主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md deleted file mode 100644 index 488f6c5d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md +++ /dev/null @@ -1,51 +0,0 @@ -# 模型部署与高速推理对比 - -百炼平台为推理调用提供两类专属容量方案：**模型部署**（含 PTU、模型单元、按 Token 用量三种计费形态）与**高速推理**（TPM 预留）。两者都通过预付费锁定专属吞吐、避免公共限流，但在隔离强度、接入方式、适用模型与计费粒度上存在差异。本文并排列出关键维度，帮助开发者根据业务负载特征做出选型。 - -## 关键维度对比 - -| 维度 | 模型部署（PTU） | 模型部署（模型单元） | 模型部署（按 Token 用量） | 高速推理（TPM 预留） | -| --- | --- | --- | --- | --- | -| 计费单位 | 按使用时长 + 输入/输出 TPM | 按使用时长 × 模型单元数量 | 按输入/输出 Token 数 | 按 kTPM（输入/输出分别计价） | -| 容量保障 | 专属部署实例，强隔离 | 专属部署实例，支持 PD 分离 | 不保障容量，仅验证效果 | 专属容量刚性兑付，与公共池隔离 | -| 超额处理 | 自动转按量计费，响应头 `x-dashscope-ptu-overflow:true` | — | 不适用 | 自动降级公共池按量计费，服务不中断 | -| 扩缩容 | 自助增减吞吐量 | 自助增减模型单元数量 | 控制台提交申请，人工审核 | 自助增减 kTPM | -| 接入方式 | `model` 替换为 `deployed_model`（专属服务 ID） | 同 PTU | 同 PTU | `model` 替换为专属模型 code | -| 创建入口 | 控制台或 `POST /api/v1/deployments`（华北2-北京） | 同 PTU | 同 PTU | 控制台「创建 TPM 预留」 | -| 支持模型 | 预置模型 + 导入的 LoRA 模型（千问3/2.5 系列） | 私有模型，性能指标自定义 | 部分 LoRA 调优模型 | 千问3.7-Max/Plus、千问3.6-Flash、GLM-5.2/5.1、DeepSeek-v4-Flash/Pro、Kimi-K2.6（分区域） | -| 长输入/缓存 | 支持阶梯系数与前缀缓存折扣 | 取决于模型 | 取决于模型 | 支持阶梯系数与前缀缓存折扣（容量计算器自动应用） | -| 退订规则 | 按天计费，无法提前退费 | 预付费首月内提前退订，日单价按 1.2 倍计费 | 一个月不使用自动释放 | 缩容/退订按 1.5 倍系数结算已用部分 | -| 到期保留 | 欠费保留 24 小时后停止计费、底层资源删除 | 同 PTU | — | 到期 2 小时内可续费，2~14 小时停止不可调用，14 小时后删除 | -| 地域 | API 部署仅华北2（北京） | 同 PTU | 同 PTU | 华北2（北京）、新加坡 | - -## 适用场景建议 - -- **高负载生产环境、需稳定吞吐与低延迟**：优先选择模型部署的 **PTU** 模式。它提供专属部署实例，支持长输入阶梯系数和前缀缓存折扣，适合对隔离性、性能、可观测性要求高的核心业务。 -- **私有模型部署、PD 分离、自定义性能指标**：选择模型部署的**模型单元**模式。通过 `deploy_spec`（如 `MU1`）与副本数精细控制算力，适合需要 Prefill/Decode 分离以降低首 Token 延迟的场景。 -- **调优后模型效果验证、不使用不计费**：选择模型部署的**按 Token 用量**模式。仅支持部分 LoRA 模型，扩缩容需控制台人工审核，适合实验性验证而非长期承载生产流量。 -- **流量可预估、不能接受限流、希望最小接入改动**：选择**TPM 预留**。预付费按 kTPM 锁定专属容量，超额自动降级公共池按量计费不中断，仅需将 `model` 替换为专属模型 code，适合中高流量但对极致隔离要求不高的在线服务。 -- **费用优化、无需专属容量**：考虑按量付费或资源包/节省计划，不在本对比页范围。 - -## 技术选型参考 - -1. **隔离强度**：PTU/模型单元为专属部署实例，隔离性最强；TPM 预留为专属容量兑付，与公共池共享底层但仍保障吞吐。对隔离性敏感（如合规、性能稳定性）选前者，仅对限流敏感选后者。 -2. **接入成本**：TPM 预留只需替换 `model` 参数为专属模型 code，无 API 端点变更；模型部署需通过 `POST /api/v1/deployments` 创建并等待 `status=RUNNING`，调用时使用 `deployed_model` 作为 `model`，并确保 API Key 与部署在同一业务空间。 -3. **模型范围**：模型部署支持导入自训练 LoRA 模型（千问3/2.5 系列，rank 须为 8/16/32/64，且 vocab、chat_template、VIT 须与基础模型一致）；TPM 预留仅支持平台预置模型，无法承载私有调优模型。 -4. **地域**：API 部署目前仅华北2（北京）；TPM 预留在华北2（北京）与新加坡均有开放，海外业务选 TPM 预留。 -5. **计费灵活性**：PTU 按天计费无法提前退费，模型单元首月内退订有 1.2 倍惩罚，TPM 预留缩容/退订按 1.5 倍系数结算。短期试算建议先用容量计算器评估 RPM、平均输入/输出长度、缓存命中率，避免额度过低频繁降级。 -6. **长输入与缓存**：两者均支持阶梯系数与前缀缓存折扣（如 glm-5.1 输入系数 1.33、缓存折扣 0.2；deepseek-v4-pro 缓存折扣 0.08）。PTU 通过响应头 `service_tier`、`provisioned_tokens`、`cached_tokens` 标识计费方式；TPM 预留在详情页「超额降级统计」中查看降级次数。 - -## 来源文档 - -- [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) -- [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) -- [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) -- [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md) -- [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - -## 被对比主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md deleted file mode 100644 index a61b79ec..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md +++ /dev/null @@ -1,62 +0,0 @@ -# 模型评估与应用评估对比 - -百炼平台提供了两套定位不同的评测能力：**模型评估**面向文本生成类模型本身的能力打分与横向对比，**应用评估**面向智能体/工作流应用的端到端输出质量。二者虽然都用到"评测集 + 评估器 + 报告"的组合，但评估对象、可用接口、归因粒度差异明显。本文从技术选型角度梳理关键维度，帮助开发者在正确的层面选用正确的工具。 - -## 核心差异一览 - -| 对比维度 | 模型评估 | 应用评估 | -|---------|---------|---------| -| 评估对象 | 单个文本生成模型（基座/调优模型）的推理结果 | 已发布的智能体应用、工作流应用的端到端输出 | -| 评测方式 | 自定义评测 + 基线评测 | 自动评测 + 手动评测（并存新旧两套系统） | -| 打分手段 | 大模型评估、规则评估（ROUGE/BLEU/Cosine 等）、人工评估 | 评估器（LLM 评估器 / Code 评估器）+ 人工标签 | -| 数据基础 | 评测数据集（Prompt+Completion）或推理结果集 | 评测集（旧版：对话分析/知识问答；新版：智能体/工作流/自定义） | -| 参考答案 | 规则/相似度评估需要；大模型/人工评估不需要 | 自动评测可基于知识库自动生成评测集，无需人工准备 | -| 基线能力评测 | 支持公开标准集（C-Eval、MMLU、GSM8K、BBH，仅北京地域） | 不涉及，聚焦应用业务表现 | -| 归因分析 | 侧重维度得分与通过率，无 RAG 环节归因 | 提供 BadCase 到 RAG 环节的归因（理解/重排/检索/切片/知识缺失）+ 调优建议 | -| 横向对比 | 通过排行榜对多模型排名对比 | 多应用横向评测，最多对比 8 个应用或版本 | -| 操作接口 | 仅控制台，无公开 API/SDK（编程化可参考 PAI Judge Model API） | 控制台操作，依赖应用观测能力 | -| 前置条件 | 准备评测集/结果集、创建评测维度 | 应用须已发布并配置知识库、开通应用观测、获取相应权限 | -| 计费方式 | 被评测模型推理费用 + 裁判模型评分费用（规则/人工评估无裁判费） | 评测产生的 Token 费用正常计费；评估器模型当前限时免费 | -| 典型场景 | 选型基座模型、验证微调/调优效果、模型能力回归 | 验证智能体回答质量、RAG 效果优化、应用版本迭代对比 | - -## 评分/评估器体系对比 - -| 能力 | 模型评估 | 应用评估 | -|------|---------|---------| -| 语义打分 | 大模型评估（数值型/分类型），推荐千问-Max 作裁判 | LLM 评估器，用于相关性、有害性、幻觉检测 | -| 确定性打分 | 规则评估：文本相似度、字符串匹配 | Code 评估器：Python 规则，格式校验、数值计算、精确匹配 | -| 人工打分 | 人工评估分类型（Pass/Fail） | 标签体系（分类/布尔/数字/文本）+ 快速标注 | -| 复用与固化 | 评测维度可创建为模板被多任务复用 | 评估器支持预置模板、自定义，或从历史评测任务标注结果抽象生成 | -| 组合建议 | 按标准答案有无与语义需求选单一评分方式 | 建议组合 3-5 个评估器（如 LLM 相关性 + Code 格式校验），单任务最多 10 个 | - -## 适用场景建议 - -- **选用模型评估：** - - 需要在多个基座模型之间做选型决策，或验证微调/调优后模型能力是否提升。 - - 有确定性标准答案（翻译、摘要、Function Calling），可用规则评估低成本快速打分。 - - 希望用公开标准集（C-Eval/MMLU/GSM8K/BBH）快速摸底模型基础能力（注意仅北京地域）。 - - 关注纯模型层面的语义质量，愿意用裁判模型打分。 - -- **选用应用评估：** - - 评估的是完整的智能体/工作流应用（含 RAG 检索、Prompt、知识库），而非单一模型。 - - 需要将 BadCase 归因定位到检索、重排、切片、知识缺失等具体环节并获得调优建议。 - - 需要在同一基准下横向对比多个应用或同一应用的不同版本（最多 8 个）。 - - 需要建立持续评测闭环：知识库更新、Prompt 调整、模型升级、检索策略变更后触发回归。 - -## 技术选型参考 - -1. **先看评估对象的层级**：只关心"模型答得好不好"用模型评估；关心"这个 Agent/工作流整体表现"用应用评估。 -2. **接口约束**：模型评估当前仅控制台操作，无公开 API/SDK，需编程化流水线时应提前评估（可考虑 PAI Judge Model API）；应用评估同样以控制台 + 应用观测为主。 -3. **成本控制**：两者都可用规则/Code 评估器规避裁判模型费用；模型评估可先用 50-100 条小规模验证并复用推理结果集，应用评估的评估器模型当前限时免费。 -4. **前置成本**：应用评估要求应用已发布、配置知识库并开通应用观测，接入成本更高；模型评估只需准备数据集与评测维度。 -5. **组合使用**：完整落地一个 RAG 智能体时，可先用模型评估锁定基座模型，再用应用评估在业务层做端到端质量把关与归因优化，形成"选模型 → 调应用"的两级评测链路。 -6. **注意评测噪声**：模型评估中 1-3% 的分差通常为噪声，LLM 评分器存在位置/自我偏好偏差，建议定期人工抽查校准；应用评估同样建议结合人工标签交叉验证。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [application evaluation](../guides/application-evaluation.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md deleted file mode 100644 index 4e13d227..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md +++ /dev/null @@ -1,60 +0,0 @@ -# 模型评估与模型监控对比 - -在百炼平台上，**模型评估**（模型评测）与**模型监控**分属模型生命周期的两个不同阶段，常被混淆但目标截然不同： - -- **模型评估**面向**上线前的选型与调优验证**，通过评测维度对模型的推理结果打分、对比，回答"哪个模型/哪个调优版本更好"的问题。 -- **模型监控**面向**上线后的运维与成本治理**，通过指标、告警、日志采集回答"线上跑得好不好、花了多少钱、有没有出错"的问题。 - -二者互补而非替代：评估帮你选出并验证模型，监控帮你把选出的模型稳定、经济地运行在生产环境。本文从技术选型角度对二者做关键维度对比，供开发者按阶段和目标选用。 - -## 关键维度对比 - -| 维度 | 模型评估（模型评测） | 模型监控 | -| --- | --- | --- | -| 定位阶段 | 上线前：选型、调优效果验证 | 上线后：性能/成本/错误可观测与治理 | -| 核心目标 | 对推理结果打分、对比、排名 | 采集指标、告警、审计对话日志 | -| 输入 | 评测数据集（Prompt+Completion）或推理结果集 | 线上真实调用流量（自动采集） | -| 输出 | 综合得分、通过率、逐条评分、排行榜 | 调用量、Token、延时、RPM/TPM、失败率、日志 | -| 支持模型 | 仅文本生成类模型 | 模型列表所有模型（含调优后自定义模型） | -| 评估/监控方式 | 大模型评估、规则评估、人工评估 | 用量视图 + 普通/高级监控 + 告警 | -| 地域限制 | 基线评测仅北京可用 | 高级监控限北京/新加坡/弗吉尼亚；告警限北京/新加坡；单次 Token/历史对话仅北京部分模型 | -| 数据延迟 | 任务提交后批量计算 | 用量约 1 小时；高级监控分钟级；日志分钟级 | -| API/SDK | 无公开 API/SDK，仅控制台操作 | 高级监控支持 Prometheus HTTP API，可接 Grafana | -| 计费方式 | 被评测模型推理费 + 裁判模型评分费（规则/人工评估无裁判费） | 监控本身无独立计费，聚焦成本查看与免费额度管理 | -| 典型场景 | 模型选型、调优前后对比、发布前质量把关 | 生产用量趋势、成本控制、故障告警、对话审计 | - -## 模型评估的适用场景 - -- **模型选型**：在多个候选模型间横向对比，选出最优模型（可绑定排行榜参与排名）。 -- **调优效果验证**：对比调优前后模型在自有数据集上的表现，确认微调是否有效。 -- **发布前质量把关**：上线前用 200-500 条数据做正式评测，验证问答质量、内容安全、翻译/摘要准确度等。 -- **评分方式按数据特征选择**： - - 有标准答案且格式固定 → 规则评估（字符串匹配） - - 有标准答案但表述多样 → 规则评估（文本相似度，ROUGE/BLEU/Cosine） - - 无标准答案需语义理解 → 大模型评估（裁判模型，推荐千问-Max） - - 需主观/专业判断 → 人工评估 -- **快速摸底基础能力**：用基线评测（C-Eval、MMLU、GSM8K、BBH）免数据集快速评估（注意仅北京地域）。 - -## 模型监控的适用场景 - -- **成本与免费额度治理**：查看调用量、Token 消耗、账单趋势，设置费用告警，开启"免费额度用完即停"避免超额扣费。 -- **线上性能可观测**：跟踪调用时长、首 Token 延时、RPM/TPM 等性能指标，评估服务质量。 -- **错误与安全监控**：监控失败率、限流（429）次数、内容安全拦截次数，及时发现异常。 -- **主动告警**：在北京/新加坡地域开启高级监控后配置告警规则，通过短信/邮件/电话/钉钉/企业微信/Webhook 通知。 -- **对话审计与排障**：开通审计日志与推理日志，查看单次调用的输入/输出与用量（当前仅北京部分模型）。 -- **自建可视化**：通过 Prometheus HTTP API 将高级监控指标接入 Grafana 或自建应用做深度分析。 - -## 开发者技术选型建议 - -1. **按生命周期阶段选择**：还没决定用哪个模型、或刚做完微调 → 用**模型评估**；模型已上线、关注花费和稳定性 → 用**模型监控**。二者通常先后使用，构成完整闭环。 -2. **是否需要编程化**：模型评估当前仅支持控制台操作，无公开 API/SDK，需自动化评测可参考 PAI Judge Model API；模型监控的高级监控则提供标准 Prometheus HTTP API，天然适合接入自动化可视化与告警体系。 -3. **注意地域约束**：基线评测、告警、单次 Token/历史对话等能力都有地域限制（多集中在北京），跨地域部署时需提前确认可用性。 -4. **控制成本**：评估阶段先用 50-100 条小规模验证并复用推理结果集、确定性场景优先规则评估；运维阶段通过监控控制 `max_tokens`、按任务选轻量模型、非实时任务用批量推理。 -5. **正确解读结果**：评估中 1-3% 的分差通常为噪声，LLM 评分器存在位置/自我偏好偏差需人工抽查校准；监控数据以控制台分钟级显示为准，30 天以前数据需到"费用与成本"页面查询。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md deleted file mode 100644 index 2a4e108f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型体验、生产与部署对比 - -在百炼平台上，从"选模型"到"训模型"再到"上线专属推理服务"是一条完整的链路，分别对应三个主题：**模型体验**（选型与全模态能力概览）、**模型生产**（微调训练 + 部署 API）、**模型部署**（专属资源推理服务的计费与落地）。三者面向不同阶段的开发者需求，容易混淆但各有侧重。本页对关键维度做横向对比，帮助开发者判断当前所处阶段应该查阅哪一部分文档、采用哪种能力。 - -## 定位对比 - -| 维度 | 模型体验 | 模型生产 | 模型部署 | -|------|----------|----------|----------| -| 核心目标 | 按场景选型、了解模型能力 | 微调训练定制模型 + 发布为服务 | 提供独立、资源专享的推理服务 | -| 所处阶段 | 需求调研 / 选型 | 模型定制与上线（训练→部署） | 生产落地与容量规划 | -| 面向对象 | 应用开发者、方案设计者 | 需要专属模型能力的开发者 | 需要高并发/低延迟私有推理的团队 | -| 文档类别 | guides（指南） | api（接口参考） | guides（指南） | -| 是否需要训练数据 | 否 | 是（微调数据集） | 视情况（可导入 LoRA 或用平台模型） | - -## 能力与技术维度对比 - -| 维度 | 模型体验 | 模型生产 | 模型部署 | -|------|----------|----------|----------| -| 主要内容 | 文本/视觉/视频/图片/3D/语音/音乐/向量全模态选型 | 模型调优（Fine-tuning）+ 模型部署 API | 三种计费方式、PTU 缓存、LoRA 导入、部署流程 | -| 核心接口 | 各品类模型调用 API（DashScope 等） | 调优 API + 部署 API（`POST /deployments`） | `POST/GET/DELETE /deployments`（同生产的部署接口） | -| 输入 | 场景需求（文本/图像/音频等多模态） | 基础模型 + 训练数据集 + 超参数 | 模型 ID / LoRA 产物 + 计费与资源配置 | -| 输出 | 推荐模型清单与参数指引 | 调优模型产物、在线推理服务 | 专属推理服务（`deployed_model` 唯一 ID） | -| 计费方式 | 按模型广场标注的计量（Token / 张 / 时长等） | 训练按量 + 部署按所选 plan | PTU / 模型单元（MU）/ 按 Token（LoRA），创建后不可改 | -| 地域限制 | 因模型而异（如音乐仅华北2·北京） | 依接口而定 | 仅华北2·北京地域 | -| 典型场景 | 聊天/内容生成/OCR/视频生成/TTS/ASR 等 | 定制专属模型、微调后上线 | 智能客服、实时审核等稳定高负载生产 | - -## 部署计费方式细分（属"模型部署"主题） - -| 计费方式 | plan 值 | 资源特性 | 计费粒度 | 适用场景 | -|----------|---------|----------|----------|----------| -| 预置吞吐 PTU | `ptu` | 预留资源保障 TPM，超额自动转按量 | 时长 × TPM 单价（后付费按小时 / 预付费按天） | 流量稳定、需延迟确定性的高负载 | -| 模型单元 MU | `mu` | 资源独占，支持 PD 分离 | 时长 × 模型单元数（后付费按分钟 / 预付费按月） | 私有微调模型、长时任务 | -| 按 Token（LoRA） | `lora` | 仅对 SFT/LoRA 自定义模型开放，不用不计费 | 按 Token 使用量 | 调优效果验证 | - -## 适用场景建议 - -- **还在挑模型 / 评估能力** → 查阅**模型体验**。先按模态（文本、视觉、视频、图片、3D、语音、音乐、向量）定位推荐模型，再到模型广场核对参数与计费，无需自己训练即可直接调用共享 API。 -- **需要专属能力、要微调训练** → 走**模型生产**流程：准备数据集 → 调优 API 提交微调 → 部署 API 发布为在线服务 → 调用端点推理。这是"训练 + 上线"的编排视角，关注接口定义与工作流顺序。 -- **要把模型跑成生产级私有服务** → 深入**模型部署**：根据流量特征选 PTU / MU / 按 Token，利用容量计算器规划 TPM，关注前缀缓存折扣、长输入阶梯系数、LoRA 导入要求（rank 8/16/32/64、必需文件、VL 需冻结 VIT）等落地细节。 - -## 技术选型参考 - -1. **共享推理 vs 专属推理**：仅需通用能力时直接用模型体验推荐的公共模型（按量共享）；对并发、延迟、数据隔离有确定性要求时，才走部署形成专属服务。 -2. **生产与部署的关系**：模型生产是"训练→部署"的**全流程视角**（含调优 API），模型部署是其中"部署"环节的**深度展开**（计费、缓存、导入、排障），两者共用 `POST /deployments` 接口。需要微调则从生产入手，只需部署平台模型可直接看部署文档。 -3. **计费不可逆**：部署计费方式一经创建无法更改，切换须先下线再重新部署；预付费提前退订首月按 1.2 倍计费——上线前务必确认。 -4. **自训练模型限制**：仅支持 LoRA 导入（不支持全参微调），且不能改动 vocab / chat_template，VL 模型须冻结 VIT，OSS Bucket 需打 `bailian-datahub-access` 标签。 -5. **计费自动兜底**：PTU 超出额度或输入超模型上限会自动转按量计费（响应头 `x-dashscope-ptu-overflow:true`），无需改代码，但需通过 `service_tier`、`provisioned_tokens`、`cached_tokens` 字段监控成本。 - -## 被对比主题页 - -- [model experience](../guides/model-experience.md) -- [model production](../api/model-production.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md deleted file mode 100644 index d76323b6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md +++ /dev/null @@ -1,69 +0,0 @@ -# 模型评测 vs 模型监控 vs 模型生产对比 - -百炼平台围绕模型全生命周期提供三大核心能力：**模型评测**用于上线前的质量验证，**模型监控**用于上线后的运行保障，**模型生产**用于模型的定制与部署。三者分别对应"评估选型 — 生产部署 — 运维保障"三个阶段，开发者需根据当前所处阶段选择合适的工具。本文从功能定位、输入输出、适用模型、费用模型等维度进行系统对比，帮助开发者快速定位所需能力。 - -## 关键维度对比 - -| 维度 | 模型评测 | 模型监控 | 模型生产 | -| --- | --- | --- | --- | -| **功能定位** | 模型质量量化验证 | 模型运行状态观测与告警 | 模型调优、压缩与部署 | -| **生命周期阶段** | 上线前 / 迭代验证 | 上线后持续运维 | 模型定制与交付 | -| **输入** | 评测数据集（Excel）或预置榜单 | 自动采集调用数据（无需手动输入） | 训练数据集 + 基础模型 | -| **输出** | 评测报告（综合得分、通过率、分数分布） | 监控看板、调用日志、告警通知 | 定制模型 + 在线推理服务端点 | -| **交互方式** | 控制台创建评测任务 | 控制台看板 + Prometheus API + Grafana | REST API（异步任务模式） | -| **支持模型范围** | 千问系列、开源模型、部分三方模型；基线评测仅支持调优后模型 | 所有模型（普通监控）；高级监控限北京/新加坡/弗吉尼亚 | 平台支持的基础模型（微调）+ 自定义导入模型（部署） | -| **[计费](../concepts/billing.md)方式** | 被评测模型推理费 + 裁判模型评分费 | 免费（高级监控依赖云监控 Prometheus 实例） | 训练算力费 + 部署推理资源费 | -| **数据延迟** | 任务式，分钟至小时级完成 | 普通监控约 1 小时；高级监控分钟级 | 异步任务，训练耗时视数据量而定 | -| **地域限制** | 无特殊限制 | 日志仅华北2（北京）；告警仅北京/新加坡 | 以控制台和 API 返回为准 | -| **典型操作** | 创建评测维度 → 上传数据 → 运行任务 → 查看报告 | 查看看板 → 配置告警 → 排查日志 | 创建调优任务 → 压缩 → 部署 | - -## 适用场景建议 - -### 模型评测适用场景 - -- **模型选型**：在多个候选模型之间进行横向对比，选择最适合业务的模型。 -- **调优验证**：微调后需要量化验证是否提升了目标能力。 -- **版本回归**：模型升级后确认核心能力未退化。 -- **能力基线**：使用预置榜单（C-Eval、MMLU、GSM8K 等）建立通用能力基准线。 - -### 模型监控适用场景 - -- **线上稳定性保障**：实时观测调用时长、首包延迟、失败率等性能指标。 -- **成本管控**：按[业务空间](../concepts/workspace.md)追踪 [Token](../concepts/token.md) 消耗，发现异常用量。 -- **故障排查**：通过历史对话日志定位推理异常的具体请求。 -- **主动告警**：[Token](../concepts/token.md) 突增、超时率升高时自动通知运维人员。 - -### 模型生产适用场景 - -- **领域定制**：通用模型无法满足垂直场景时，通过微调注入专业知识。 -- **成本优化**：对定制模型进行量化压缩，降低推理资源占用。 -- **服务化交付**：将训练好的模型部署为在线推理端点，供业务应用调用。 - -## 技术选型参考 - -| 开发者需求 | 推荐能力 | 说明 | -| --- | --- | --- | -| 需要知道哪个模型效果好 | 模型评测 | 创建自定义评测任务或基线评测进行量化对比 | -| 需要定制一个专属模型 | 模型生产 | 调优 → 可选压缩 → 部署为推理服务 | -| 模型已上线，需要保障稳定运行 | 模型监控 | 开启高级监控 + 告警规则 | -| 调优后想验证效果 | 模型评测 + 模型生产 | 先用模型生产完成调优，再用模型评测量化对比 | -| 线上模型出现异常，需要定位原因 | 模型监控 | 查看调用日志和失败详情排查 | -| 想在 Grafana 中统一看板 | 模型监控 | 接入高级监控的 Prometheus HTTP API | - -## 三者协作关系 - -在实际项目中，三大能力通常按以下流程串联使用： - -1. **模型生产**：基于基础模型微调出领域定制模型，压缩后部署上线。 -2. **模型评测**：部署前用评测任务验证定制模型的质量是否达标。 -3. **模型监控**：部署后持续观测运行状态，设置告警确保服务可靠性。 - -当监控发现质量劣化时，可回到模型评测进行定量验证，确认后再通过模型生产重新调优迭代，形成闭环。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model monitoring](../guides/model-monitoring.md) -- [model production](../api/model-production.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md deleted file mode 100644 index 1b246309..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md +++ /dev/null @@ -1,54 +0,0 @@ -# 模型监控 vs 应用监控 - -百炼平台提供两种互补的观测能力：**模型监控**关注单个模型 API 调用层面的性能与成本，**应用观测**则从应用维度端到端追踪内部调用链路。理解二者的定位差异，有助于开发者在不同阶段选择合适的监控手段，实现从模型选型到应用上线的全链路可观测。 - -## 关键维度对比 - -| 维度 | 模型监控 | 应用监控（应用观测） | -|------|----------|----------------------| -| 监控对象 | 单个模型的 API 调用 | 整个应用（智能体/工作流/高代码应用）的完整调用链路 | -| 核心关注点 | 模型性能、[Token](../concepts/token.md) 消耗、费用趋势 | 应用内部节点间的执行流程、端到端延时 | -| 支持范围 | 所有模型（含自定义调优模型） | 智能体应用、工作流应用、高代码应用 | -| 地域限制 | 高级监控仅北京/新加坡/弗吉尼亚；告警仅北京/新加坡 | 无特殊地域限制（需开通 OpenTelemetry 服务） | -| 典型指标 | 调用时长、首 [Token](../concepts/token.md) 延时、RPM/TPM、失败率、[Token](../concepts/token.md) 消耗 | 调用次数、失败率、Token 总量、平均首 Token 耗时、平均调用时长 | -| 数据粒度 | 普通监控小时级；高级监控分钟级 | 分钟级更新 | -| 数据保留 | 用量统计 30 天 | 调用记录最长 30 天 | -| 告警能力 | 支持（短信/邮件/电话/钉钉/企微/Webhook） | 不支持（需结合模型监控告警） | -| 外部集成 | Prometheus HTTP API + Grafana（Basic Auth） | OpenTelemetry 服务存储 | -| 日志能力 | 推理日志（输入/输出/Token 逐条记录） | Trace/Span 链路日志（含节点层级关系） | -| 筛选维度 | API-KEY、推理类型、时间范围、时间精度 | Request ID/Trace ID/Span ID、状态、Span Name、延时、Token 量、标签 | -| 数据标注 | 不支持 | 支持对 Span 添加标签（布尔/分类/数字/文本） | -| 评测集联动 | 不支持 | 支持将 Span 数据导入评测集 | -| 费用 | 免费（高级监控需开启） | 功能免费，存储费用由 OpenTelemetry 服务收取 | -| API 支持 | Prometheus HTTP API 可编程查询 | 仅控制台操作，无独立 API | - -## 适用场景 - -### 模型监控适合 - -- **模型选型与调优阶段**：对比不同模型的调用时长、首 Token 延时、成本效益比，为技术选型提供数据依据。 -- **成本管控**：按[业务空间](../concepts/workspace.md)维度统计 Token 消耗，配置免费额度用尽即停策略，避免超支。 -- **稳定性保障**：通过告警规则监控失败率、限流错误，在异常发生时第一时间通知相关人员。 -- **运维自动化**：利用 Prometheus API 接入自建监控系统或 Grafana 大盘，实现统一运维。 - -### 应用监控适合 - -- **应用调试与优化**：通过 Trace 链路追踪定位应用内部的性能瓶颈节点（如检索耗时、模型推理耗时）。 -- **智能体/工作流行为分析**：查看 AGENT、RETRIEVER、LLM、TOOL 等节点的执行顺序和中间结果，理解应用的实际行为。 -- **数据驱动迭代**：将线上真实调用数据标注后导入评测集，形成"观测-标注-评测-优化"闭环。 -- **问题排查**：按 Request ID 精确定位单次调用的完整执行链路，快速定位报错节点。 - -## 技术选型建议 - -1. **两者并非互斥**：生产环境建议同时开启。模型监控提供宏观的性能告警与成本管控，应用观测提供微观的链路诊断能力。 -2. **开发调试阶段**优先使用应用观测，快速定位逻辑问题和性能瓶颈。 -3. **上线运营阶段**优先配置模型监控告警，保障服务稳定性和成本可控。 -4. **高代码应用**目前应用观测仅支持入口节点，内部链路需依赖代码中集成 Tracing 模块并启用 `--telemetry enable` 参数。 -5. 如需将监控数据接入第三方平台，模型监控通过 Prometheus API 对接，应用观测数据则存储在 OpenTelemetry 服务中。 - -## 被对比主题页 - -- [model monitoring](../guides/model-monitoring.md) -- [application monitoring](../guides/application-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md deleted file mode 100644 index 31515cc0..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md +++ /dev/null @@ -1,42 +0,0 @@ -# 模型微调、压缩与高速推理对比 - -在百炼平台上，把一个模型从"能用"打磨到"好用、省钱、够快"，通常会经历三类相互独立又可串联的能力：**模型微调（Fine-tuning）** 把领域知识、任务能力与人类偏好写入参数；**模型压缩** 通过量化在保持能力的前提下降低部署规格与成本；**高速推理** 则在调用侧解决容量兑付与输出速度问题。三者定位不同——微调改变"模型本身"，压缩改变"部署精度/成本"，高速推理改变"调用时的容量与吞吐"，因此常见的完整链路为：模型调优 → 模型压缩（可选）→ 模型部署 → 高速推理（可选）。本文面向开发者，从关键维度对比三者，帮助做技术选型。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（量化） | 高速推理（TPM 预留 / 快速模式） | -| --- | --- | --- | --- | -| 解决的问题 | 深度定制：注入领域知识、指令遵循、偏好对齐、特定音色/风格 | 降低部署所需 MU 规格，减少推理成本 | 容量刚性兑付（预留）/ 提升单请求输出速度（快速模式） | -| 处理对象 | 基础模型 + 训练数据集 | 平台微调产出的全精度自定义模型 | 线上可调用的模型（含专属 code / fast 模型） | -| 输入格式 | 数据集：SFT 用 ChatML `{"messages":[...]}`；DPO 追加 `chosen`/`rejected`；CPT 用 `{"text":"..."}`；视觉 ZIP+`data.jsonl`；CosyVoice `{"wav_fn","text"}` | 源模型 + 量化模板（+ 可选校准数据集，最多 5 个） | 标准对话请求（`messages`）；预留替换 `model` 为专属 code；快速模式指定 fast 模型 ID | -| 输出产物 | 微调后模型（`finetuned_output` / `deployed_model`） | 低精度量化模型（新后缀，更小部署规格） | 无新模型产物，仅提升调用容量/速度 | -| 支持模型 | 千问文本/VL、万相图像/视频、CosyVoice 等多模态 | 仅平台微调产出的自定义模型（如 qwen3.5-flash 系列），不支持基础/第三方模型 | 预留：已开放预留的模型；快速模式：`glm-5.2-fast-preview` | -| 训练/处理方式 | CPT、SFT（全参/高效 LoRA）、DPO（全参/LoRA） | 量化（不含剪枝、蒸馏），MU 编号越大规格越小 | 预留：锁定 kTPM 吞吐；快速模式：提速至 1.5~2 倍 TPS | -| API 端点 / 接入 | `POST /api/v1/files`、`/fine-tunes`、`/deployments`（华北2 API Key） | 主要通过控制台创建压缩任务（模型训练 → 模型压缩） | 预留：标准 `dashscope.aliyuncs.com` + 专属 code；快速模式：`{workspace_id}.cn-beijing.maas.aliyuncs.com` 专属域名 | -| 计费方式 | 按 Token（API 创建仅支持按 Token）或训练单元（仅控制台）；成本与耗时较高 | 压缩任务限时免费；成本体现在部署阶段按 MU 规格计费（示例节省约 56%） | 预留：按 kTPM 预付费，超额降级按量；快速模式：按 token（同标准 API），超额排队 | -| 地域限制 | 仅华北2（北京） | 仅华北2（北京） | 预留：以控制台为准；快速模式：华北2（北京）、新加坡 | -| 可逆性 / 约束 | 可继续叠加训练（CPT→SFT→DPO） | 不可逆；压缩后不支持继续微调或二次压缩 | 预留退订/快速模式为 preview，规格可能变动 | -| 代码改动量 | 大：准备数据集、走训练→部署全流程 | 小：控制台配置任务，调用无需改代码 | 小：预留替换 `model`；快速模式换域名 + fast 模型 ID | -| 典型场景 | 领域问答、安全合规对齐、专属音色/图像风格 | 微调模型上线前的成本优化 | 高峰期专属容量保障；对输出速度敏感的编程助手/Agent/实时对话 | - -## 各方案适用场景建议 - -- **模型微调**：当 Prompt 工程、插件调用、RAG 等手段仍无法达到效果，需要把领域知识或人类偏好"写进参数"时选用。文本生成推荐按 `CPT（可选）→ SFT → DPO（可选）` 递进组合；对成本/时间敏感或数据集较小时用 LoRA 高效训练，追求效果且模型支持时优先全参。多模态（VL、万相、CosyVoice）按各自支持的方式与超参集处理。注意训练成本与耗时较高，且能力仅限华北2（北京）。 -- **模型压缩**：已有微调产出的全精度自定义模型、且部署成本偏高时选用。因压缩任务当前限时免费，建议在免费期内对同一模型尝试多个量化模板，分别部署后用业务测试集验证精度与成本的平衡，再选最优方案上线。切记压缩不可逆，需保留上游全精度模型以便重新压缩。 -- **高速推理 - TPM 预留**：面向高峰期需要"专属容量、不被公共限流"的生产业务，按 kTPM 预付费锁定吞吐，超额自动降级按量、不中断，适合对稳定性和容量兑付要求高的场景。 -- **高速推理 - 快速模式**：面向 AI 编程助手、Agent 多步推理、实时对话等对**输出速度**敏感的场景，TPS 可达标准 API 的 1.5~2 倍。当前仍为 preview，规格可能调整，生产接入前需评估稳定性；接入需使用专属域名，不能与预留的标准域名接入方式混用。 - -## 技术选型参考 - -1. **先定位需求属于哪一层**：要"改变模型能力/风格"→微调；要"降部署成本"→压缩；要"保容量或提速度"→高速推理。三者可组合：微调 → 压缩 → 部署 → 叠加 TPM 预留或快速模式。 -2. **成本与投入权衡**：微调投入最大（数据、训练时间、Token/训练单元费用）；压缩几乎零调用改造且当前限时免费，收益在部署阶段；高速推理中预留是预付费换刚性容量，快速模式按量计费换速度。 -3. **可逆性与前置条件**：压缩不可逆且要求上游为平台微调模型；微调与压缩均仅限华北2（北京）；快速模式为 preview 且模型/地域受限。 -4. **接入方式差异**：预留沿用标准域名 + 专属 model code；快速模式改用 workspace 专属域名 + fast 模型 ID，两者不可混用，选型时需评估现有调用代码的改造范围。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md deleted file mode 100644 index 47044f8e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md +++ /dev/null @@ -1,85 +0,0 @@ -# 模型微调、压缩与部署对比 - -在百炼平台上，将一个基础模型转化为可在生产环境中使用的定制化推理服务，通常涉及三个阶段：**模型微调（Fine-tuning）**、**模型压缩（Quantization）** 和 **模型部署（Deployment）**。三者在模型生产链路中依次衔接（调优 → 压缩（可选）→ 部署），各自解决不同层面的问题。本文从功能定位、适用范围、操作方式和成本等维度进行系统对比，帮助开发者在技术选型时做出合理决策。 - -## 功能定位与链路关系 - -模型微调、压缩与部署构成一条完整的模型生产流水线： - -- **模型微调**处于链路上游，目标是基于自有数据定制模型能力，使模型在特定领域或任务上表现更优。 -- **模型压缩**处于链路中间，是可选环节，通过量化技术降低模型参数精度，从而减小部署所需的算力规格和推理成本。 -- **模型部署**处于链路下游，将训练好（或压缩后）的模型发布为在线推理服务，供应用调用。 - -三者缺一不可地覆盖了"训练 → 优化 → 上线"的完整生命周期，但各自的输入输出、关注点和约束条件差异显著。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（Quantization） | 模型部署（Deployment） | -|------|------------------------|------------------------|----------------------| -| **核心目标** | 基于自有数据定制模型能力，提升特定场景表现 | 降低模型参数精度，减小部署规格与推理成本 | 将模型发布为在线推理服务，供应用调用 | -| **在链路中的位置** | 上游（第一步） | 中间（可选） | 下游（最后一步） | -| **输入** | 基础模型 + 训练数据集（JSONL / ZIP 等） | 微调产出的自定义模型 | 预置模型 / 微调模型 / 压缩后模型 | -| **输出** | 微调后的自定义模型 | 低精度量化模型 | 可调用的在线推理服务（API 端点） | -| **支持的模型范围** | 广泛：Qwen3.6/3.5/3/2.5 系列文本模型、Qwen-VL 视觉模型、Wan 图像/视频模型、CosyVoice 语音模型 | 较窄：仅支持百炼平台微调产出的特定自定义模型（如 qwen3.5-flash） | 最广泛：预置模型（Qwen、DeepSeek、GLM 等）+ 微调模型 + 压缩模型 + OSS 导入的 LoRA 模型 | -| **支持的模态** | 文本生成、视觉理解、图像生成、视频生成、语音合成 | 仅文本生成 | 文本生成、[多模态](../concepts/multimodal.md)、语音合成等 | -| **操作方式** | 控制台 + API | 仅控制台 | 控制台 + API | -| **是否可逆** | 可重复训练和迭代 | 不可逆：压缩后不支持继续微调或二次压缩 | 可随时上线/下线，支持重新部署 | -| **典型耗时** | 数小时至数天（取决于数据量和训练轮次） | 平台自动完成，排队后运行 | 分钟级（状态变为"运行中"即可调用） | -| **计费方式** | 按 Token 用量计费：训练数据 Token 数 x 训练轮次 x 训练单价 | 压缩任务本身限时免费 | 三种模式：预置吞吐（PTU）、模型单元（MU，按时长）、Token 用量（按调用量） | -| **地域限制** | 无特殊限制 | 仅华北2（北京） | 无特殊限制 | - -## 训练方法与压缩模板选择 - -### 微调方法 - -百炼平台提供三种递进式微调方法，推荐按 CPT → SFT → DPO 的顺序使用： - -| 方法 | 目标 | 数据要求 | 典型场景 | -|------|------|----------|----------| -| CPT（继续预训练） | 注入领域知识 | 1000万+ Token 无标签文本 | 金融/医疗/法律等垂直领域适配 | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条高质量问答对 | 客服、代码助手、Agent 工具调用 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组正负样本对 | 安全合规强化、降低幻觉 | - -每种方法支持全参训练和高效训练（LoRA）两种模式。全参训练效果更好但耗时更长、成本更高；LoRA 训练收敛快、成本低，适合快速验证。 - -### 压缩量化模板 - -量化模板决定压缩后的部署规格，MU 编号越大表示部署规格越小、成本越低，但精度损失可能越大。可选配校准数据以提升量化精度，建议选择与推理场景语义相近的数据集。 - -## 部署计费方式对比 - -| 计费方式 | 计费公式 | 适用场景 | 特点 | -|---------|---------|---------|------| -| 预置吞吐（PTU） | 使用时长 x (输入 TPM 单价 x 输入 TPM + 输出 TPM 单价 x 输出 TPM) | 流量稳定的高负载生产环境 | 保障吞吐额度内不限速，超额自动切换按量计费；TPS 约为按量的 1.5~2.0 倍 | -| 模型单元（MU） | 使用时长（小时）x 模型单元数量 x 模型单元单价 | 需要资源独占和自定义性能指标 | 支持 PD 分离计算模式，可降低首 Token 延迟 | -| Token 用量 | 输入 Token 数 x 输入单价 + 输出 Token 数 x 输出单价 | 调用量不稳定、用量较少的场景 | 仅支持部分 LoRA 调优模型，不使用不计费 | - -## 适用场景建议 - -**场景一：快速验证模型定制效果** -推荐路径：SFT 高效训练（LoRA）→ 直接部署（Token 用量计费）。跳过压缩环节，以最低成本快速上线验证。 - -**场景二：生产环境成本敏感** -推荐路径：SFT 全参训练 → 模型压缩 → 部署（MU 计费）。通过压缩降低部署规格（如从 MU1x2 降至 MU8x1，成本节省约 56%），适合长期运行的在线服务。 - -**场景三：高并发低延迟的核心业务** -推荐路径：微调（按需）→ 部署（PTU 计费）。PTU 模式提供预留吞吐保障，TPS 提升约 1.5~2.0 倍，适合流量稳定且对延迟敏感的场景。 - -**场景四：[多模态](../concepts/multimodal.md)模型定制** -推荐路径：视觉/图像/视频/语音 SFT → 直接部署（MU 计费）。当前压缩功能仅支持文本模型，[多模态](../concepts/multimodal.md)微调模型需直接部署。 - -## 技术选型要点 - -1. **是否需要微调**：如果预置模型已满足需求，可直接部署，无需微调。当模型在特定领域表现不佳、需要定制输出格式或降低幻觉时，再考虑微调。 -2. **是否需要压缩**：压缩可显著降低部署成本，但会带来一定精度损失且不可逆。建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选择最优方案。 -3. **如何选择部署计费方式**：流量稳定选 PTU，需要资源独占和灵活配置选 MU，用量少且不稳定选 Token 用量。注意 Token 用量模式仅支持部分 LoRA 模型。 -4. **地域约束**：模型压缩当前仅在华北2（北京）地域可用，规划链路时需考虑地域一致性。 -5. **不可逆操作提醒**：压缩后的模型不支持继续微调或二次压缩，务必保留上游全精度微调模型以备后续迭代。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md deleted file mode 100644 index bb5c6450..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md +++ /dev/null @@ -1,69 +0,0 @@ -# 模型优化方式对比（微调/压缩/高速推理） - -百炼平台围绕"让模型更贴合业务、跑得更省、调得更稳"提供了三类模型优化能力：**模型微调（Fine Tuning）**、**模型压缩（Model Compression）**、**高速推理（TPM 预留）**。三者作用阶段不同、目标不同，但在生产落地中常被串联使用：先用微调注入业务能力，再用压缩降低部署规格，最后用高速推理保障线上吞吐。本页从输入输出、支持模型、API/控制台入口、[计费](../concepts/billing.md)方式、典型场景等维度横向对比，供开发者在技术选型时参考。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine Tuning） | 模型压缩（Model Compression） | 高速推理（TPM 预留） | -| --- | --- | --- | --- | -| 一句话定位 | 把业务/场景知识写入模型参数 | 把全精度微调模型量化为低精度版本 | 为指定模型锁定专属吞吐量 | -| 优化阶段 | 训练阶段（改变模型权重） | 部署前阶段（不改权重，只做量化） | 服务运行阶段（不动模型，只保容量） | -| 是否改变模型 | 是，产出新的微调模型 | 是，产出量化后的低精度模型 | 否，仅锁定原模型的推理容量 | -| 前置条件 | 已开通百炼模型服务，准备好训练数据 | 当前工作空间内已有支持压缩的自定义微调模型 | 已开通百炼模型服务，已创建业务空间 | -| 输入格式 | 文本：JSONL 问-答对；图像/视频：`.zip`（含 `data.jsonl` + 训练样本）；语音：录音文件 | 全精度微调模型 + 校准数据集（部分模板要求） | 业务流量预估（RPM、平均输入/输出长度、缓存命中率） | -| 输出格式 | 新的自定义微调模型（`finetuned_output`） | 量化后的低精度模型（部署到更小规格单元） | 专属 model code（替换 API 的 `model` 参数即可） | -| 支持模型 | 千问系列（文本/视觉）、万相 wan2.7/wan2.2/wan2.5（图像/视频）、CosyVoice 语音 | 当前支持 `qwen3.5-flash-2026-02-23` 等基础模型对应的自定义微调模型（以控制台为准） | 按区域开放，如 qwen3.7-max、glm-5.2、kimi-k2.6、deepseek-v4-pro 等（以控制台为准） | -| 控制台入口 | 「模型调优」页面 | 「模型 > 模型训练 > 模型压缩」页面 | 控制台「创建 TPM 预留」 | -| API 端点 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`、`GET /api/v1/fine-tunes/{job_id}`、`POST /api/v1/deployments` | OpenAPI（`custom_calibration_file_ids` 等字段） | 替换 `model` 参数为专属 model code，调用 OpenAI 兼容模式 `/compatible-mode/v1/chat/completions` | -| [计费](../concepts/billing.md)方式 | 按 Token [计费](../concepts/billing.md)（API 创建的任务）；控制台创建可使用模型训练单元（预付费/后付费） | 当前限时免费；产出模型部署上线后按部署单元规格计费 | 按 kTPM 预付费，输入/输出 TPM 分别计价，按天计费；预留内调用不额外收费，超额自动降级按量 | -| 任务状态 | `PENDING` → `SUCCEEDED`；部署 `RUNNING` | 待开始/排队中/运行中/停止中/压缩成功/压缩失败/已取消 | 运行中/待生效/变配中/已停止/已过期/已取消 | -| 是否可逆 | 可迭代训练（CPT→SFT→DPO 递进） | 不可逆：不支持继续训练，不支持二次压缩 | 可扩缩容、续费、退订；退订不可恢复 | -| 典型场景 | 注入领域知识、对齐偏好、复刻风格、压低延迟、抑制幻觉 | 降低显存占用、部署到更小规格、降低部署成本、提升吞吐 | 流量可预估且不能接受限流、业务高峰期保障容量 | - -## 各方案适用场景建议 - -### 模型微调（Fine Tuning） - -适合"通用模型答不好、Prompt 工程已到瓶颈"的场景。当你需要把领域术语/事实、特定对话格式、人类偏好、特定风格写入模型本身时，应选微调： - -- **补知识**：CPT 注入领域文本（1000 万+ Token）。 -- **学做事**：SFT 教会指令遵循（1000+ 问-答对），文本/视觉/图像/视频/语音[多模态](../concepts/multimodal.md)均支持。 -- **做更好**：DPO 对齐人类偏好（100+ 组偏好对）。 -- **快速验证或数据量小**：优先用 LoRA 高效训练，速度快、成本低。 -- CosyVoice 调优仅支持 API 发起，且仅支持 SFT 高效微调。 - -### 模型压缩（Model Compression） - -适合"已经微调出好模型，但部署太贵/规格太大"的场景。它不改变模型能力定位，只做量化降本： - -- 前提是上游已产出全精度微调模型（压缩是微调的下游）。 -- 通过量化模板决定压缩后可部署规格（如 MU5、MU8），把 `MU1 * 2（¥108/小时）` 降为 `MU8 * 1（¥47/小时）`。 -- 适合追求更低部署成本、更高推理吞吐、更小部署单元的线上服务。 -- 注意：压缩产物不可继续训练、不可二次压缩，迭代须回到全精度模型重训。 - -### 高速推理（TPM 预留） - -适合"流量可预估、不能接受公共限流、要保障高峰期容量"的场景。它不动模型本身，只锁定推理吞吐： - -- 业务高峰期不能因公共池限流而抖动。 -- 需要 SLA 刚性兑付、专属容量不与他人共享。 -- 超额自动降级按量计费且不中断服务，适合可预估但偶发突刺的流量。 -- 接入极简：仅替换 `model` 参数为专属 model code，无需改代码逻辑。 - -## 技术选型建议 - -三者并非互斥，建议按"能力注入 → 规格压缩 → 容量保障"的顺序串联使用： - -1. **先微调**：用业务数据训练出符合场景的自定义模型（解决"答得好不好"）。 -2. **再压缩**：对微调产物做量化，部署到更小规格单元（解决"跑得省不省"）。 -3. **最后保容量**：用 TPM 预留锁定线上吞吐，保障高峰期稳定（解决"调得稳不稳"）。 - -如果只是临时验证或流量很小，可跳过压缩与预留，直接按量调用微调后的模型；如果是对外承诺 SLA 的生产服务，建议三步全走。若仅需保障容量而模型本身已满足需求，可单独使用 TPM 预留；若仅需降本而流量平稳，可单独使用模型压缩。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md deleted file mode 100644 index 7cd40cf3..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md +++ /dev/null @@ -1,51 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的监控能力：**应用监控（应用观测）**聚焦于应用层端到端调用链路，**模型监控**聚焦于模型层用量与运行状态。两者观测对象、指标口径、数据时效、地域支持各不相同，开发者需根据排查目标（应用行为 vs 模型行为）选择合适的能力，或组合使用以实现从业务入口到模型底层的全链路可观测。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 模型监控 | -| --- | --- | --- | -| 观测对象 | 业务空间内的应用（智能体应用、工作流应用、高代码应用） | 主账号下所有业务空间内的模型调用（按"模型 + 业务空间"维度） | -| 观测粒度 | 应用内部调用链路（CHAIN/AGENT/RETRIEVER/LLM/TOOL 等节点，支持嵌套） | 模型维度的聚合指标（调用次数、失败率、延时、Token、RPM/TPM 等） | -| 关键指标 | 延时、Token 量、调用记录（Prompt/输出）、CHAIN 节点状态 | 安全/成本/性能/错误四类：失败率、429 限流、首 Token 延时、Token 消耗等 | -| 数据时效 | 分钟级同步，调用记录最长可查 30 天 | 普通监控小时级、高级监控分钟级；用量数据延迟约 1 小时 | -| 模型范围 | 随应用调用自动覆盖应用内使用的模型 | 普通监控支持所有地域所有模型；高级监控仅北京/新加坡/弗吉尼亚；告警仅北京/新加坡 | -| 日志/历史对话 | 在 Trace 详情中查看 Prompt 内容、输出、Span 原始数据 | 仅华北2（北京）地域部分模型支持请求和响应日志（需开通推理日志） | -| 数据筛选 | Span 模式（Root/All/Model Span）+ 状态/Span Name/延时/Token/标签等多条件 | 按 API-KEY、推理类型、时间范围、时间精度筛选；失败详情可点击查看 | -| 数据导出 | 支持 JSONL / EXCEL 导出 | 高级监控指标存储于私有 Prometheus，支持标准 Prometheus HTTP API，可接入 Grafana | -| 主动告警 | 不支持告警，仅控制台查看 | 支持告警规则（短信/邮件/电话/钉钉/企微/Webhook），按 CRITICAL/ERROR/WARNING/INFO 分级 | -| 计费方式 | 应用观测功能本身不收费；数据存储费由 OpenTelemetry 服务收取 | 监控功能本身不收费；用量按 Token/张/秒计费，存储于云监控 Prometheus | -| 开通方式 | 控制台"应用观测配置"：授权 OTel 角色 → 开通 OTel 服务 → 初始化 LogStore | 控制台"模型监控配置"：开通审计/推理日志、高级监控、告警规则 | -| API 支持 | 无 API，仅控制台操作 | 高级监控提供 Prometheus HTTP API，可接入自建应用 | -| 标注与[评测](../concepts/evaluation.md) | 支持对 Span 加标签（布尔/分类/数字/文本），可加入[评测](../concepts/evaluation.md)集 | 不直接支持标注，但 Token 追踪与告警可辅助[评测](../concepts/evaluation.md)样本筛选 | - -## 适用场景建议 - -### 选用应用监控（应用观测） - -- 需要追踪智能体应用、工作流应用内部端到端调用链路（如检索 → 改写 → LLM → 工具调用）。 -- 排查应用层延时异常、Token 消耗突增的根因节点（定位到具体 Span）。 -- 希望将真实线上调用作为评测样本导入评测集。 -- 应用部署在任意地域，且不依赖主动告警（仅需控制台查询）。 -- 限制：不支持 Assistant API 创建的智能体应用；高代码应用仅能观测入口 CHAIN 节点。 - -### 选用模型监控 - -- 需要按业务空间/模型维度汇总调用量、失败率、延时、Token 消耗，做成本核算与稳定性保障。 -- 需要 429 限流、内容安全错误等模型层错误的明细与趋势分析。 -- 需要主动告警（超时、Token 突增等静默失败），或接入 Grafana/自建看板。 -- 需要 Token 消耗的汇总、追踪、阈值告警三层成本管理。 -- 需要查看请求和响应原文（仅北京地域部分模型，需开通推理日志）。 -- 限制：用量统计不支持按阿里云账号维度汇总；日志/告警有地域限制。 - -### 组合使用 - -当问题既涉及应用行为又涉及模型底层时，建议先用应用监控定位异常 Span（如某个 LLM 节点延时高或失败），再通过 Request ID / Trace ID 关联到模型监控中的对应调用记录，结合模型层失败率、限流、Token 消耗做联合诊断。应用监控提供"是什么调用出了问题"，模型监控提供"模型侧为什么会出问题"，二者互补构成完整的可观测体系。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md deleted file mode 100644 index d9d02b20..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md +++ /dev/null @@ -1,60 +0,0 @@ -# 应用监控与模型监控对比 - -阿里云百炼平台提供了两套互补的可观测能力：**应用监控（应用观测）**面向由智能体、工作流、高代码搭建的完整应用，追踪端到端的调用链路；**模型监控**面向底层模型调用本身，聚焦性能指标、Token 消耗与费用趋势。二者关注的抽象层次不同——前者回答"这个应用内部发生了什么"，后者回答"这个模型跑得怎么样、花了多少钱"。本文对比两者的关键差异，帮助开发者在做可观测性方案选型时快速定位合适的工具。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 模型监控 | -| --- | --- | --- | -| 观测对象 | 智能体应用、工作流应用、高代码应用的端到端调用链路 | 模型调用本身（含调优后的自定义模型） | -| 核心价值 | 追踪应用内部调用链路、节点耗时与思考过程 | 追踪模型性能、Token 消耗与费用趋势 | -| 支持范围 | 三类应用；不支持 Assistant API 创建的智能体；高代码仅观测入口 CHAIN 节点 | 用量统计支持全部模型；高级监控仅限北京/新加坡/弗吉尼亚；告警仅限北京/新加坡 | -| 数据延迟 | 分钟级 | 普通监控小时级、高级监控分钟级；用量统计约 1 小时 | -| 数据留存 | 调用记录最长 30 天 | 用量统计不支持查看 30 天以前数据 | -| 指标维度 | 调用次数/失败率、Token 量、平均首 Token 耗时、平均调用时长 | 安全、成本、性能（RPM/TPM 等）、错误四大类 | -| 筛选维度 | Request ID / Trace ID / Span ID、状态、Span、输入输出、延时、Token、标签 | API-KEY、推理类型（实时/批量）、时间范围、时间精度 | -| 链路/节点粒度 | 提供 CHAIN/AGENT/RETRIEVER/LLM/TOOL 等丰富节点类型与嵌套关系 | 无调用链路，按模型/API-KEY 维度聚合指标 | -| 告警能力 | 无内置告警 | 支持告警规则（短信/邮件/电话/钉钉/企业微信/Webhook），分 CRITICAL/ERROR/WARNING/INFO 等级 | -| 数据标注/评测 | 支持 Span 打标签、直接加入评测集 | 不涉及 | -| 日志能力 | 可查看 Prompt、输出、延时、Token 等调用记录 | 推理日志/审计日志，仅华北2（北京）部分模型，分钟级延迟 | -| 外部系统接入 | 数据导出为 JSONL / EXCEL | 高级监控数据存于私有 Prometheus，支持标准 HTTP API，可接 Grafana（Basic Auth） | -| 使用方式 | 仅控制台操作，无 API | 控制台面板 + Prometheus HTTP API | -| 计费方式 | 功能本身免费，观测数据存储由 OpenTelemetry 服务收费 | 监控功能查看无额外说明；聚焦模型调用本身的 Token/费用管理 | -| 前置开通 | 需授权 OpenTelemetry 角色、开通服务、初始化 LogStore | 高级监控/告警需在模型监控配置中开启；日志需开通审计与推理日志 | -| 典型场景 | 调试应用逻辑、定位慢节点、优化检索/Prompt、构建评测样本 | 成本核算、性能容量规划、异常主动告警、故障排查 | - -## 适用场景建议 - -### 优先选择应用监控（应用观测） - -- 应用由智能体、工作流或高代码搭建，需要**看清内部调用链路**（哪个节点慢、检索命中如何、模型思考过程）。 -- 需要按 Trace ID / Span ID **下钻单次请求**，排查复杂多节点应用的逻辑问题。 -- 希望把**真实线上调用直接沉淀为评测样本**，或对 Span 数据做标注管理。 -- 只在控制台侧使用、可接受分钟级更新、调用记录 30 天内可回溯。 - -### 优先选择模型监控 - -- 关注的是**模型层面的性能与成本**：RPM/TPM、失败率、限流、Token 消耗与费用趋势。 -- 需要**主动告警**（限流、失败率飙升、成本超标），并通过短信/电话/钉钉等渠道通知。 -- 需要**按业务空间精细化管理模型成本**，结合免费额度"用完即停"控制预算。 -- 希望把监控数据**接入 Grafana 或自建系统**做统一可视化（依赖高级监控的 Prometheus 接口）。 -- 需要通过推理日志做**内容审计或故障排查**（注意仅限北京地域部分模型）。 - -## 技术选型参考 - -两套能力并非互斥，成熟的生产系统通常**同时启用**： - -1. **分层定位问题**：先用模型监控发现"某模型失败率/延时异常"，再用应用监控下钻到具体应用与节点，确认是检索、插件还是模型环节导致。 -2. **注意地域限制**：模型高级监控与告警对地域敏感（高级监控限北京/新加坡/弗吉尼亚，告警限北京/新加坡，推理日志限北京）。若业务不在这些地域，应用监控的分钟级链路观测可作为主要抓手。 -3. **权衡数据延迟**：应用监控与模型高级监控均为分钟级；模型普通监控与用量统计为小时级，不适合实时排障，更适合趋势分析与成本复盘。 -4. **成本可观测归口**：需要账单/额度维度治理时以模型监控（用量统计 + 免费额度）为准；需要单次请求成本归因时用应用监控的 Token 量指标。 -5. **数据出口差异**：需要把数据导入外部 BI/表格用应用监控的 JSONL/EXCEL 导出；需要接入监控告警平台（Grafana/Prometheus）走模型高级监控的 HTTP API。 - -简言之：**排查"应用怎么跑的"用应用监控，管理"模型跑得好不好、花多少钱"用模型监控**，两者组合可覆盖从链路调试到成本告警的完整可观测闭环。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md deleted file mode 100644 index 3e3d1202..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md +++ /dev/null @@ -1,57 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的监控体系：**应用观测**和**模型监控**。应用观测侧重于端到端追踪应用内部的调用链路与节点级性能，帮助开发者定位应用层面的延时与逻辑问题；模型监控则聚焦于模型维度的调用性能、Token 消耗、费用趋势与告警，帮助开发者管控模型使用成本与稳定性。理解两者的定位差异，有助于开发者建立完整的可观测性方案。 - -## 核心维度对比 - -| 维度 | 应用观测 | 模型监控 | -|------|----------|----------| -| **监控对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（含自定义模型和调优模型） | -| **观测粒度** | 节点级（CHAIN、LLM、RETRIEVER、TOOL 等多种节点类型） | 模型级（按模型名称、API-KEY、推理类型筛选） | -| **核心指标** | 调用延时、Token 用量（输入/输出）、首 Token 耗时、调用次数与失败率 | 调用时长、首 Token 延时、RPM、TPM、Token 消耗、失败率、限流错误、内容安全错误 | -| **数据更新频率** | 分钟级 | 普通监控为小时级；高级监控为分钟级 | -| **数据保留时长** | 最长 30 天 | 用量统计不支持查看 30 天以前的数据 | -| **告警能力** | 无 | 支持告警规则配置，通知方式含短信、邮件、电话、钉钉机器人、企业微信机器人、Webhook | -| **用量统计** | 无独立用量统计，通过监控统计图表查看 Token 总量 | 按[业务空间](../concepts/workspace.md)维度统计，支持免费额度管理与用完即停 | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 高级监控数据可通过 Prometheus HTTP API 接入 Grafana 等外部系统 | -| **数据标注** | 支持对 Span 添加标签（布尔值/分类/数字/文本），可关联评测集 | 不支持 | -| **日志查看** | 通过 Trace 详情查看 Prompt 输入输出与原始数据 | 开通推理日志后可查看每次调用的输入、输出及 Token 消耗（仅华北2北京地域部分模型） | -| **地域限制** | 无特殊地域限制 | 高级监控仅支持北京、新加坡、弗吉尼亚；告警仅支持北京、新加坡 | -| **计费** | 功能免费，观测数据存储由 OpenTelemetry 服务收费 | 功能免费，高级监控数据存储在私有 Prometheus 实例中 | -| **操作方式** | 仅控制台操作，无 API | 控制台操作 + Prometheus API 接入 | - -## 开通与前置条件对比 - -| 维度 | 应用观测 | 模型监控 | -|------|----------|----------| -| **开通步骤** | 授权 OpenTelemetry 服务角色权限 → 开通 OpenTelemetry 服务 → 初始化 LogStore | 普通监控默认可用；高级监控需在模型监控配置中手动开启 | -| **子账号权限** | 需要 AliyunBailianFullAccess + 应用观测页面权限 + ram:CreateServiceLinkedRole 策略 | 标准百炼控制台权限即可 | -| **推荐操作账号** | 主账号 | 无特殊要求 | - -## 适用场景建议 - -### 应用观测适合以下场景 - -- **调用链路排查**：应用响应慢或出错时，需要逐节点定位瓶颈，例如区分是检索环节还是模型推理环节导致延时过高。 -- **Prompt 调试**：查看每次调用的完整输入输出，对比不同 Prompt 的效果。 -- **数据质量管理**：通过 Span 筛选与标注功能，对线上真实调用数据进行质量打分，并将优质样本导入评测集。 -- **工作流应用调试**：工作流包含多种节点类型（意图分类、脚本转换、条件判断等），需要观察每个节点的执行情况。 - -### 模型监控适合以下场景 - -- **成本管控**：按[业务空间](../concepts/workspace.md)统计模型用量与费用，配合免费额度管理控制预算。 -- **稳定性保障**：配置告警规则，在失败率上升或限流异常时及时收到通知。 -- **性能基线建立**：通过 RPM、TPM、首 Token 延时等指标建立性能基线，持续跟踪模型表现。 -- **多模型对比**：对比不同模型在相同业务场景下的调用时长、Token 消耗等指标，辅助模型选型。 -- **外部可视化集成**：将监控数据接入 Grafana 等系统，构建统一的运维大盘。 - -### 建议组合使用 - -在生产环境中，推荐同时启用两套监控：用模型监控建立全局的成本与稳定性视图并配置告警，用应用观测在出现异常时深入排查具体调用链路。两者从不同维度覆盖可观测性需求，互为补充而非替代。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md deleted file mode 100644 index a7e7444d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md +++ /dev/null @@ -1,76 +0,0 @@ -# [多模态](../concepts/multimodal.md)生成 API 对比（图像/视频/3D） - -百炼平台提供图像、视频、3D 三类[多模态](../concepts/multimodal.md)生成 API，分别面向不同的内容产出形态。三者都通过 DashScope HTTP 接口调用，统一使用 API Key 鉴权，并遵循"创建任务 → 轮询结果"的异步任务模式（部分图像模型支持同步调用）。本页从输入格式、输出格式、支持模型、API 端点、调用模式、[计费](../concepts/billing.md)与典型场景等维度做横向对比，帮助开发者根据产出目标与技术约束做选型。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 产出形态 | 静态图片（PNG） | 视频文件 | GLB 模型 + 预览渲染图 | -| 输入格式 | 文本、图像（图生图/编辑）、参考图 | 文本、图像（首帧/首尾帧）、参考图、视频、音频 | 文本、单图、多图（前/左/后/右 4 视角，固定数组长度 4） | -| 输出格式 | PNG，1–6 张或多图组图 | 视频 URL | PBR 材质 GLB（`pbr_model_url`）或无贴图基础模型（`base_model_url`），含 1 张预览渲染图 | -| 调用模式 | 同步（千问/万相2.6+/Z-Image 等新版）或异步（V1 及部分编辑/创意类） | 仅异步 | 仅异步 | -| API 端点 | 同步：`POST /api/v1/services/aigc/multimodal-generation/generation`；异步轮询：`GET /api/v1/tasks/{task_id}` | `POST /api/v1/services/aigc/video-generation/video-synthesis`（部分走 `image2video/video-synthesis`）；轮询：`GET /api/v1/tasks/{task_id}` | `POST /api/v1/services/aigc/video-generation/3d-generation`；轮询：`GET /api/v1/tasks/{task_id}` | -| 必需请求头 | `Authorization`；异步需 `X-DashScope-Async: enable` | `Content-Type`、`Authorization`、`X-DashScope-Async: enable` | `X-DashScope-Async: enable`（缺少报 `current user api does not support synchronous calls`） | -| 典型耗时 | 同步秒级返回；异步 1–2 分钟 | 1–5 分钟，万相2.1 视频编辑 5–10 分钟 | 较长，轮询建议间隔约 15 秒 | -| task_id 有效期 | 24 小时 | 24 小时 | 24 小时，超时返回 `UNKNOWN` | -| 产物下载链接有效期 | 随接口返回 | 随接口返回 | 2 小时，需及时下载 | -| 支持模型系列 | 千问图像、万相（Wan/wanx）、Z-Image、可灵 | 万相（HappyHorse/Wan/wanx）、爱诗 PixVerse、Vidu、可灵 | Tripo（`Tripo/Tripo-H3.1` 高精度、`Tripo/Tripo-P1.0` 专业快速） | -| 地域可用性 | 北京/新加坡/弗吉尼亚等多地域，地域独立鉴权不可混用；千问-图像翻译仅北京 | 同地域约束，模型/Endpoint/API Key 必须同地域；PixVerse、Vidu 仅北京 | 仅华北2（北京） | -| 业务空间专属域名 | 支持（`{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` 等） | 支持（北京 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com` 等） | 走默认 dashscope 域名 | -| SDK 支持 | 部分模型支持 DashScope SDK（Python/Java） | HTTP 为主 | HTTP | -| [计费](../concepts/billing.md)方式 | 按张数/模型[计费](../concepts/billing.md) | 按任务/时长计费 | 按任务计费（`usage` 记录任务类型与生成数量） | -| 典型场景 | 文生图、图生图、图像编辑、虚拟模特、试衣、海报、背景生成、擦除补全、画面扩展、人物写真 | 文生视频、图生视频、参考生视频、视频编辑、视频换人、数字人、肖像动态视频 | 文生 3D、单图生 3D、多图生 3D，游戏/电商/工业设计资产 | - -## 调用模式差异 - -三类 API 在调用流程上高度一致，均采用"创建任务 → 轮询查询"模式，但图像 API 额外提供**同步调用**能力： - -- **图像生成**：千问图像系列（qwen-image-2.0-pro/max/plus）、万相 2.6/2.7 文生图与编辑、Z-Image 等新版模型支持一次请求即返回结果的同步调用，走 `multimodal-generation/generation` 端点；V1 版及部分编辑/创意类模型仍需异步。同步模式流程更简单，适合交互式场景。 -- **视频生成 / 3D 生成**：因耗时较长（视频 1–10 分钟，3D 资产更久），统一仅支持异步。请求必须携带 `X-DashScope-Async: enable`，缺少该头会报错 `current user api does not support synchronous calls`。 - -三者都强调"请勿重复创建任务"，`task_id` 有效期 24 小时，直接轮询即可。 - -## 输入能力对比 - -| 输入方式 | 图像 | 视频 | 3D | -| --- | --- | --- | --- | -| 纯文本 | 支持，复杂文字渲染能力强（千问系列） | 支持（文生视频） | 支持，中英文等多语言，最大 1024 字符 | -| 单图输入 | 支持（图生图、图像编辑） | 支持（首帧生视频） | 支持，JPEG/PNG，宽高 [20,6000]，≤20MB | -| 多图输入 | 部分编辑模型支持多图输入/输出 | 支持（参考生、首尾帧） | 支持，固定 4 视角（前/左/后/右），有效 2–4 张 | -| 视频输入 | 不适用 | 支持（视频编辑、参考生视频） | 不适用 | -| 音频输入 | 不适用 | 万相2.7 支持[多模态](../concepts/multimodal.md)输入含音频 | 不适用 | - -3D 生成的多图输入有严格的视角顺序约束（前/左/后/右），不需要的视角传空对象 `{}`，这与图像/视频的"多图作为参考"语义不同。 - -## 产物与质量参数 - -| 项 | 图像 | 视频 | 3D | -| --- | --- | --- | --- | -| 输出规格 | 总像素 512×512~2048×2048，宽高比 1:4~4:1，1–6 张；万相2.7 支持 4K | 视频文件 URL | 面数：H3.1 最高 200 万面，P1.0 最高 2 万面 | -| 质量参数 | 分辨率、张数、宽高比 | 分辨率、时长、镜头叙事（`shot_type: multi`） | `texture_quality`（标清/高清）、`geometry_quality`（standard/ultra）、`pbr`、`texture` | -| 一致性能力 | 千问编辑支持角色一致性 | 万相2.7 参考生支持角色形象与音色一致性 | 多图视角约束保证几何一致性 | -| 预览能力 | 直接返回图片 | 直接返回视频 | 额外返回 `rendered_image_url` 预览渲染图 | - -## 适用场景建议 - -- **选图像生成 API**：需要静态视觉产出，强调文字渲染、风格化、精确编辑（增删移动物体、改动作）、虚拟模特/试衣/海报等电商与营销场景。优先用同步调用模型（千问图像、万相2.6+/2.7、Z-Image）以简化流程；批量或创意类任务再用异步。 -- **选视频生成 API**：需要动态叙事、数字人、肖像动态视频、视频编辑/换人。文生视频、图生视频（首帧/首尾帧）、参考生视频均可，万相2.7 是推荐的新版协议，支持多模态输入与角色/音色一致性。注意 PixVerse、Vidu 仅北京地域可用且需单独开通。 -- **选 3D 生成 API**：需要可直接导入引擎/3D 软件的 GLB 资产，适用于游戏、电商商品 3D 展示、工业设计。仅北京地域可用，需开通 Tripo。高精度选 `Tripo/Tripo-H3.1`（最高 200 万面），追求速度选 `Tripo/Tripo-P1.0`。 - -## 技术选型参考 - -1. **产出形态决定大类**：图片→图像 API；视频→视频 API；3D 模型→3D API。三者端点不同，不可混用。 -2. **延迟敏感优先同步**：仅图像 API 提供同步调用，适合交互式产品；视频与 3D 必须异步，需在业务侧实现轮询或配置异步任务回调（3D 查询接口默认 RPS 20）。 -3. **地域与鉴权**：三类均要求模型、Endpoint、API Key 同地域。3D 仅北京可用；千问-图像翻译、PixVerse、Vidu 也仅北京。建议迁移到业务空间专属域名以获得更好性能与稳定性。 -4. **任务复用**：`task_id` 24 小时有效，三类都要求轮询而非重复创建任务；3D 产物下载链接仅 2 小时，需及时落盘。 -5. **输入约束**：3D 多图必须按前/左/后/右 4 视角顺序；图像图文混排需开启 `enable_interleave=true` 并配合 SSE 流式；视频首尾帧、参考生有专属模型变体。 -6. **模型开通**：可灵、PixVerse、Vidu、Tripo 均需先在控制台搜索并开通授权，再调用 API。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md deleted file mode 100644 index 9ac91dd6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md +++ /dev/null @@ -1,71 +0,0 @@ -# 图像生成、视频生成与3D生成对比 - -百炼平台提供图像生成、视频生成和3D生成三大[多模态](../concepts/multimodal.md)内容创作能力。三者在输入输出格式、模型生态、调用方式和适用场景上各有侧重。本文从开发者技术选型角度，对这三类生成能力进行系统对比，帮助快速定位最适合业务需求的方案。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|---------|---------|--------| -| **输入格式** | 文本提示词、参考图像（单张/多张）、涂鸦草图 | 文本提示词、首帧/首尾帧图像、参考图像/视频/音频 | 文本提示词、单张图像、多图（4视角：前/左/后/右） | -| **输出格式** | PNG 图像（512x512 至 4K） | MP4 视频 | GLB 模型（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| **调用方式** | 同步调用为主，部分模型支持异步 | 全部异步（创建任务 → 轮询结果） | 全部异步（创建任务 → 轮询结果） | -| **主要模型系列** | 千问-图像、万相(Wan/Wanx)、Z-Image、可灵(Kling) | 万相(Wan)、HappyHorse、爱诗(PixVerse)、Vidu、可灵(Kling) | Tripo（H3.1 / P1.0） | -| **模型数量** | 20+ 款模型覆盖各类场景 | 6 大模型家族，任务类型丰富 | 2 款模型（高精度 / 专业快速） | -| **可用地域** | 部分全地域，部分仅华北2（北京） | 多数仅华北2（北京），HappyHorse 支持海外地域 | 仅华北2（北京） | -| **批量输出** | 单次可生成 1-9 张图像 | 单次生成 1 条视频 | 单次生成 1 个3D模型 | -| **产物有效期** | 即时返回，URL 有时效 | task_id 有效期 24 小时 | task_id 有效期 24 小时，下载链接有效期 2 小时 | -| **典型生成耗时** | 秒级至十秒级 | 分钟级 | 分钟级（耗时较长） | - -## 能力覆盖对比 - -| 能力 | 图像生成 | 视频生成 | 3D生成 | -|------|:-------:|:-------:|:-----:| -| 文本生成 | 支持 | 支持 | 支持 | -| 图像/图片参考生成 | 支持 | 支持（首帧/首尾帧） | 支持（单图/多图） | -| 内容编辑 | 支持（局部重绘、风格迁移、扩图等） | 支持（指令编辑、视频迁移） | 不支持 | -| [多模态](../concepts/multimodal.md)混合输入 | 支持（文+图） | 支持（文+图+视频+音频） | 不支持 | -| 中文文字渲染 | 支持（千问、Z-Image 等） | 不适用 | 不适用 | -| 人像/人物专项 | 支持（人像风格重绘、AI试衣） | 支持（数字人、舞动人像、悦动人像等） | 不适用 | -| PBR 材质输出 | 不适用 | 不适用 | 支持 | - -## 计费方式差异 - -- **图像生成**：按张计费，不同模型单价不同（如扩图 0.18 元/张）。部分创意工具仅提供免费体验额度，用完不可付费续用。 -- **视频生成**：按任务计费，费用与视频时长、分辨率、模型版本相关。 -- **3D生成**：按任务计费，费用与贴图质量（standard/detailed）和几何精度（standard/ultra）相关。 - -## 适用场景建议 - -**选择图像生成的场景：** -- 电商商品图、营销海报、社交媒体配图等静态视觉内容 -- 需要精细文字渲染或图文混排的场景（如带中文的宣传图） -- 图像编辑与风格迁移（如局部重绘、背景替换、AI试衣） -- 对生成速度要求高、需要批量出图的场景 - -**选择视频生成的场景：** -- 短视频创作、广告片制作、动态内容营销 -- 人像动画（数字人播报、舞蹈视频、唱演视频） -- 需要多镜头叙事或多角色互动的复杂视频 -- 视频风格转换和口型替换等后期编辑 - -**选择3D生成的场景：** -- 游戏资产、AR/VR 场景中的3D模型快速原型 -- 电商3D商品展示 -- 需要 PBR 材质的高精度3D资产生产（最高 200 万面） -- 从多视角图片重建3D物体 - -## 技术选型要点 - -1. **生成速度**：图像生成最快（秒级），视频和3D生成均需分钟级等待，且必须使用[异步调用](../concepts/async-invocation.md)模式。 -2. **模型生态丰富度**：图像生成和视频生成均拥有多个模型家族可选，3D生成目前仅有 Tripo 系列。 -3. **地域限制**：3D生成仅限北京地域；视频和图像生成的部分模型也有地域限制，选型前需确认目标地域的模型可用性。 -4. **输出后处理**：视频和3D的产物下载链接有时效限制（3D仅 2 小时），需在业务流程中及时下载存储。 -5. **开通流程**：3D生成和部分视频/图像模型需在百炼控制台额外搜索并开通服务，不是默认可用。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md deleted file mode 100644 index 45397164..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md +++ /dev/null @@ -1,74 +0,0 @@ -# Qwen API、应用调用与托管智能体 API 对比 - -百炼平台提供了多种 API 接入方式，开发者在集成大模型能力时常面临选型困惑：是直接调用 Qwen 模型 API，还是通过应用调用 API 使用已编排好的智能体/工作流，亦或是采用 Managed Agents API 获得平台全托管的智能体运行时？本文从接口定位、协议兼容性、会话管理、工具能力、计费模式等维度进行系统对比，帮助开发者根据实际场景做出技术选型。 - -## 定位差异 - -- **Qwen API**：直接调用 Qwen 系列大语言模型，获取文本生成能力。开发者自行管理 [prompt](../guides/prompt.md)、上下文和工具调用逻辑，灵活度最高。 -- **应用调用 API**：调用在百炼控制台中已创建并发布的智能体或工作流应用。应用内部已封装模型选择、知识库检索、插件调用等编排逻辑，开发者只需传入用户输入即可获取最终结果。 -- **Managed Agents API**：平台全托管的智能体运行时，提供 Agent、Session、Environment、Skill、File 等资源抽象。由平台负责会话状态机、沙箱执行、工具调用与事件流推送，适合需要长期运行、多步工具调用的复杂场景。 - -## 关键维度对比 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -|------|----------|-------------|-------------------| -| **定位** | 模型级调用，直接访问 Qwen 系列模型 | 应用级调用，调用已编排好的智能体/工作流 | 平台托管智能体运行时，全生命周期管理 | -| **兼容协议** | OpenAI Chat Completions、OpenAI Responses、Anthropic Messages、DashScope 原生 | OpenAI Responses（兼容模式）、DashScope | 百炼原生 REST API | -| **API 端点** | `POST /compatible-mode/v1/chat/completions` 等 | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` 或 `POST /api/v1/apps/{APP_ID}/completion` | `POST /api/v1/agentstudio/sessions/{session_id}/events` 等 | -| **认证方式** | [API Key](../concepts/api-key.md)（`Authorization: Bearer`） | [API Key](../concepts/api-key.md) + APP ID（+ 可选 Workspace ID） | [API Key](../concepts/api-key.md) + Workspace ID | -| **模型选择** | 请求体 `model` 字段指定任意 Qwen 模型 | 控制台配置，调用时无需指定模型 | Agent 创建时配置模型 | -| **会话管理** | 调用方自行维护（Responses 接口除外） | DashScope API 通过 `session_id` 自动维护；Responses API 需传完整历史 | 平台全托管，Session 状态机自动驱动 | -| **工具/插件** | Responses 接口内置联网搜索、代码解释器、网页提取；其他接口需自定义 | 控制台可视化编排插件、知识库、工具 | Skill（zip 包上传）+ Environment 沙箱执行 | -| **[多模态](../concepts/multimodal.md)支持** | 需选用 VL 系列模型 | Responses API 支持图像和文件输入 | 通过 File 资源挂载到 Session | -| **[流式输出](../concepts/streaming.md)** | 支持（`stream=true`） | 支持（`stream=true`） | SSE 事件流（`GET .../events/stream`） | -| **[异步调用](../concepts/async-invocation.md)** | 不支持 | Responses API 支持（`background=true`），DashScope 暂不支持 | 原生异步，Session 状态机驱动 | -| **支持地域** | 多地域 | 仅华北2（北京） | 仅 cn-beijing | -| **SDK 支持** | OpenAI SDK、Anthropic SDK、[DashScope SDK](../concepts/dashscope-sdk.md) | OpenAI SDK、[DashScope SDK](../concepts/dashscope-sdk.md) | 百炼原生 SDK / HTTP | -| **配置方式** | 纯代码，请求参数控制 | 控制台可视化编排 + API 调用 | API 全程管理（Agent/Environment/Session/Skill） | - -## 适用场景建议 - -### 选择 Qwen API - -- 需要直接、细粒度地控制模型推理参数（temperature、top_p 等）。 -- 已有基于 OpenAI 或 Anthropic SDK 的应用，希望低成本迁移到百炼平台。 -- 构建自定义的 RAG、Agent 框架，模型调用只是其中一环。 -- 对 [prompt](../guides/prompt.md) 工程有深度需求，需要完整掌控输入输出。 - -### 选择应用调用 API - -- 已在百炼控制台完成智能体或工作流的可视化编排，希望通过 API 将其集成到业务系统。 -- 需要使用控制台配置的知识库检索、插件、工作流节点等平台能力，不想在代码中重新实现。 -- 团队中非开发人员负责应用逻辑编排，开发人员只负责 API 集成。 -- 需要快速上线，应用逻辑变更通过控制台完成而非修改代码。 - -### 选择 Managed Agents API - -- 需要平台全托管的智能体运行时，不想自行管理会话状态和工具执行环境。 -- 智能体任务涉及多步工具调用、代码执行、文件读写，需要沙箱环境保障安全。 -- 希望通过 API 动态创建和管理多个智能体，实现多 Agent 协作。 -- 需要细粒度的事件流（SSE）来追踪智能体执行过程中的每一步操作。 -- 有自定义工具（Skill）需要安全审核后挂载，要求版本锁定和隔离。 - -## 选型决策参考 - -1. **"我只需要一个模型回答问题"** — 选 Qwen API。最简单直接，兼容主流 SDK。 -2. **"我已在控制台搭好应用，想 API 接入"** — 选应用调用 API。零编排代码，改逻辑只需改控制台配置。 -3. **"我需要平台帮我管理 Agent 的执行环境和工具调用"** — 选 Managed Agents API。平台托管状态机、沙箱和事件流，适合复杂任务。 -4. **迁移成本优先** — Qwen API 的 OpenAI/Anthropic 兼容接口迁移成本最低；应用调用 API 也提供 OpenAI 兼容模式。 -5. **功能完整度优先** — Qwen API 的 DashScope 原生接口参数最丰富；Managed Agents API 的资源模型最完整。 - -## 注意事项 - -- Qwen API 的兼容接口可能不暴露 DashScope 原生的全部参数，如需最全功能建议使用 DashScope 接口。 -- 应用调用 API 要求先在控制台创建并发布应用，APP ID 只能通过控制台手动获取。 -- Managed Agents API 当前仅支持 cn-beijing 地域，Skill 上传后需通过安全扫描才能挂载。 -- 三种 API 的计费方式均基于 token 消耗，但应用调用和 Managed Agents 可能涉及额外的平台资源费用（如沙箱、存储），请参考官方定价文档。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md deleted file mode 100644 index a3da8448..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md +++ /dev/null @@ -1,57 +0,0 @@ -# Qwen API vs 全双工实时API vs 托管智能体API - -百炼平台提供多种 API 接口满足不同开发场景。Qwen API 面向文本生成任务，提供多协议兼容的 HTTP 接口；全双工实时 API（Omni Realtime）基于 WebSocket 实现低延迟的音视频实时对话；托管智能体 API（Managed Agents）则提供完整的智能体托管运行时，由平台负责会话编排与沙箱执行。本文从协议、能力、适用场景等维度帮助开发者做出技术选型。 - -## 关键维度对比 - -| 维度 | Qwen API | 全双工实时 API | 托管智能体 API | -| --- | --- | --- | --- | -| 通信协议 | HTTP（REST） | WebSocket（长连接） | HTTP（REST）+ SSE 事件流 | -| 输入格式 | 文本（messages JSON） | 音频流（PCM 16kHz）、图像（Base64） | 文本消息、文件附件 | -| 输出格式 | 文本（支持流式 SSE） | 音频流（PCM 24kHz）+ 文本转录 | 事件流（SSE），含文本、工具调用回执等 | -| 支持模型 | Qwen 系列文本生成模型 | Qwen3.5-Omni-Realtime、Qwen3-Omni-Flash-Realtime、Qwen-Omni-Turbo-Realtime | 可配置任意百炼平台模型 | -| API 端点 | `https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions` 等 | `wss://{WorkspaceId}.{region}.maas.aliyuncs.com/api-ws/v1/realtime` | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio` | -| 兼容协议 | OpenAI Chat Completions、OpenAI Responses、Anthropic Messages、DashScope 原生 | 无（百炼专有 WebSocket 协议） | 无（百炼专有 REST 协议） | -| 工具调用 | Responses 接口内置联网搜索/代码解释器/网页提取；其他接口需自定义 | 支持 Function Calling 和联网搜索（仅 Qwen3.5 系列） | 平台托管技能（Skill zip 包），沙箱内执行 | -| 会话管理 | 仅 Responses 接口自动管理历史；其他需客户端维护 | 平台维护 WebSocket 会话上下文 | 平台全托管（Session 状态机：idle → running → idle/terminated） | -| 延迟特性 | 标准 HTTP 请求-响应，流式可逐 token 返回 | 超低延迟（毫秒级音频帧推送） | 异步任务式，SSE 实时推送中间事件 | -| 多模态支持 | 纯文本（部分模型支持图像输入） | 音频 + 视频 + 图像 + 文本 | 文本 + 文件（通过 File 资源挂载） | -| 计费方式 | 按 token 计费（输入/输出分计） | 按音频时长 + token 计费 | 按底层模型 token + 沙箱资源用量计费 | - -## 适用场景建议 - -### Qwen API - -- **文本问答与对话**：聊天机器人、客服对话、内容生成等标准 NLP 任务 -- **快速迁移**：已有 OpenAI 或 Anthropic 代码的项目，可通过兼容接口低成本切换到 Qwen 模型 -- **轻量工具调用**：使用 Responses 接口可直接获得联网搜索、代码解释器能力，无需额外开发 -- **批量处理**：适合离线或准实时的文本处理流水线 - -### 全双工实时 API - -- **语音助手**：需要实时语音输入并即时语音回复的场景 -- **智能客服**：基于 VAD 自动检测用户说话意图，实现自然的对话轮转 -- **多模态交互**：需要同时处理音视频输入的实时应用（如视频通话中的 AI 助手） -- **声音定制**：利用声音复刻能力打造品牌专属语音形象 - -### 托管智能体 API - -- **复杂任务编排**：智能体需要多轮推理、工具调用、代码执行的场景 -- **平台托管运行时**：不想自行管理会话状态、沙箱环境和工具执行的团队 -- **企业级智能体**：需要版本管理、权限控制、文件交互等完整生命周期管理 -- **多技能组合**：通过 Skill 机制灵活组装工具链，且由平台保证安全审核 - -## 技术选型指引 - -1. **只需文本生成** → 选择 Qwen API。如已有 OpenAI/Anthropic SDK 代码，使用对应兼容接口可零改动迁移。 -2. **需要实时语音交互** → 选择全双工实时 API。它是唯一支持音频流式双向通信的接口，延迟最低。 -3. **需要平台托管的智能体运行时** → 选择托管智能体 API。适合需要沙箱执行、多工具编排、会话生命周期管理的复杂 Agent 应用。 -4. **组合使用**：三者并非互斥。例如可用托管智能体 API 编排复杂工作流，其底层模型调用仍走 Qwen API；或在语音助手前端使用全双工实时 API，后端通过托管智能体执行复杂任务。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [omni realtime api](../api/omni-realtime-api.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md new file mode 100644 index 00000000..2e9fadb6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md @@ -0,0 +1,59 @@ +# 实时 API 方案对比：Omni Realtime API 与 Realtime API + +本对比旨在帮助开发者清晰理解百炼平台两大实时交互接口——**Omni Realtime API** 与 **Realtime API** 的定位差异、能力边界与适用条件，避免因选型偏差导致集成成本上升、功能缺失或性能不达预期。二者虽均面向低延迟多模态实时场景，但在协议设计、模型覆盖、部署灵活性及工程约束上存在系统性差异。本文基于最新 v3.5 系列模型能力与生产环境实践整理，适用于语音助手、智能客服、实时翻译、音视频互动等核心场景的技术选型决策。 + +## 关键维度对比 + +| 维度 | Omni Realtime API | Realtime API | +|------|-------------------|--------------| +| **协议基础** | 仅支持 WebSocket（`wss://.../api-ws/v1/realtime`） | 支持 **WebSocket / WebRTC / AOQ** 三协议，可按终端类型动态选择 | +| **输入格式** | - 音频：PCM（16 kHz，Base64）
- 图像：JPG/JPEG（≤1080p，≤256 KB Base64）
- 视频帧：通过 `append_video` 事件流式传入
- **图像必须在首次音频输入后发送，且与音频缓冲区协同提交** | - WebSocket：同 Omni（文本/音频/图像分通道）
- WebRTC/AOQ：原生支持音视频+文本**混合传输**（如带时间戳的 AV sync stream）
- 所有协议均支持 `input_audio_buffer.append` 等标准事件 | +| **输出格式** | - 文本 + PCM 音频（24 kHz）
- 可配置 `modalities: ["text"]` 或 `["text","audio"]`，**不支持纯音频输出**
- 增量事件：`response.text.delta`、`response.audio.delta`、`conversation.item.input_audio_transcription.delta`（内置 ASR） | - 同 Omni 的文本/音频增量输出
- WebRTC/AOQ 额外支持**端到端音视频流直出**（如 H.264 视频帧 + Opus 音频包）
- WebSocket 模式下不支持原生视频流输出 | +| **支持模型与能力** | - 专属模型系列：
 ✓ `qwen3.5-omni-realtime`（语义 VAD、联网搜索、工具调用）
 ✓ `qwen3.5-omni-plus/flash-realtime`（`idle_timeout_ms`、`smooth_output`）
 ✓ `qwen3-omni-flash-realtime`（默认音色 `Cherry`）
 ✗ 不支持 `livetranslate`、`multimodal-dialog`、`Fun-ASR`、`CosyVoice` 等非 Omni 系列模型 | - **统一模型网关**，覆盖更广：
 ✓ 全模态：`qwen3.5-omni-plus/flash-realtime`
 ✓ 实时翻译：`qwen3.5-livetranslate-flash-realtime`
 ✓ 多模态套件：`multimodal-dialog`（WebRTC/WebSocket）
 ✓ 专用 ASR/TTS：`Fun-ASR`、`CosyVoice`（WebSocket）
 ✓ 语音对话：`qwen-audio-3.0-realtime-plus/flash`（WebSocket） | +| **API 端点** | 固定 WebSocket 地址：
`wss://{WorkspaceId}.{region}.maas.aliyuncs.com/api-ws/v1/realtime`
（推荐业务空间专属域名，已弃用 `dashscope.aliyuncs.com`） | 协议差异化端点：
- WebSocket：`wss://dashscope.aliyuncs.com/...`（或工作空间域名）
- WebRTC：`{workspace_id}.{region}.maas.aliyuncs.com`（SDP 交换需后端代理）
- AOQ：通过 `aoqTokenForClient` 动态协商，无固定 URL | +| **计费方式** | 按 **实际调用时长（秒） + 输出 token 数量** 计费
- 音频输入/输出按采样点折算为等效时长
- 工具调用、联网搜索不额外计费，但计入会话总耗时 | 按 **协议 + 模型 + 资源消耗** 分层计费：
- WebSocket：同 Omni（时长 + token）
- WebRTC/AOQ：**增加媒体处理资源费**（如 AEC、弱网对抗、编解码开销）
- `multimodal-dialog`、`livetranslate` 等应用类模型有独立定价单元 | +| **典型场景** | - 低延迟语音助手（端侧 VAD 触发、快速响应）
- 智能客服坐席辅助（实时转录+语音回复）
- 需深度定制会话状态机的 B2B 交互系统（事件驱动精细控制） | - 浏览器端网页客服（WebRTC，免插件、强弱网适应）
- 原生 App 音视频通话（AOQ，毫秒级延迟、离线降级）
- 多语言实时会议翻译（`livetranslate` + WebRTC）
- 快速验证原型（WebSocket，零客户端依赖） | +| **VAD 与交互控制** | - 强事件驱动：`server_vad`（声学）或 `semantic_vad`（语义，仅 `qwen3.5-omni-realtime`）
- Manual 模式需显式 `commit` + `response.create`
- 支持 `idle_timeout_ms`（主动引导） | - VAD 类型一致（`server_vad`/`semantic_vad`），但 WebRTC/AOQ 内置硬件级 AEC+降噪，VAD 更鲁棒
- WebRTC/AOQ 支持**端侧触发响应**（如按键说话、手势唤醒），不依赖服务端 VAD 判定 | +| **安全与部署约束** | - 鉴权：`Authorization: Bearer `（建连时携带）
- **禁止在前端暴露 API Key**，必须由业务服务端代理
- 无 [Token](../concepts/token.md) 机制，依赖长期密钥管理 | - WebSocket：同 Omni
- WebRTC：SDP 交换必须经业务后端代理（防 CORS + 密钥泄露）
- AOQ：**强制 [Token](../concepts/token.md) 鉴权**（`aoqTokenForClient`），时效短、可撤销、支持细粒度权限 | + +## 各方案适用场景建议 + +### ✅ 推荐选用 **Omni Realtime API** 当： +- 业务聚焦于 **纯语音/语音+图像** 的轻量级实时交互（如车载助手、IoT 设备语音控制）； +- 需要 **语义级语音活动检测（`semantic_vad`）** 或 **联网搜索能力**（仅 `qwen3.5-omni-realtime` 提供）； +- 已有成熟 WebSocket 客户端栈，追求 **最小接入成本** 与 **确定性低延迟**（端到端 P99 < 400ms）； +- 对音色复刻、工具链深度集成（如自定义工具回调流程）有强需求； +- 运行环境受限（如嵌入式设备、无 WebRTC 支持的旧浏览器）。 + +### ✅ 推荐选用 **Realtime API** 当： +- 面向 **多终端统一体验**：需同时支持 Web（WebRTC）、iOS/Android（AOQ）、服务端（WebSocket）； +- 场景涉及 **真实音视频通话**（如远程医疗、在线教育），要求端到端 AEC、抗丢包、弱网自适应； +- 需要 **非 Omni 系列模型能力**：如专业语音翻译（`livetranslate`）、多模态对话套件（`multimodal-dialog`）、高保真 TTS（`CosyVoice`）； +- 安全合规要求严格：需 **短期 [Token](../concepts/token.md) 鉴权（AOQ）** 或 **后端代理 SDP（WebRTC）**，杜绝密钥暴露风险； +- 开发团队具备跨协议调试能力，愿为极致体验承担稍高集成复杂度。 + +> ⚠️ 注意：若项目需同时使用 `qwen3.5-omni-realtime`（语义 VAD）和 `multimodal-dialog`（多轮视觉引导），**必须选用 Realtime API + WebRTC 协议**——Omni Realtime API 不支持该套件。 + +## 技术选型参考（致开发者） + +| 你的需求 | 推荐方案 | 关键理由 | +|----------|-----------|-----------| +| “我要做一个微信小程序里的语音客服，用户说一句话，立刻听到回答” | **Omni Realtime API**（WebSocket） | 小程序 WebView 支持 WebSocket；Omni 的 `semantic_vad` 可精准截断口语停顿，避免“等说完才响应”；无需处理 WebRTC 信令复杂度。 | +| “我们开发 iOS/Android App，要做一个带美颜和实时字幕的视频面试系统” | **Realtime API + AOQ** | AOQ 提供原生 SDK、毫秒级音画同步、内置美颜/降噪/字幕渲染管线；`multimodal-dialog` 模型可理解面试者微表情与肢体动作。 | +| “客户要求支持 Chrome/Firefox/Safari 全浏览器实时翻译，且能应对 4G 弱网” | **Realtime API + WebRTC** | WebRTC 是浏览器唯一原生低延迟音视频协议；内置 ICE-lite 和拥塞控制，弱网下自动降码率保流畅；`livetranslate` 模型专为跨语言对话优化。 | +| “我们是 SaaS 厂商，需为不同客户提供隔离的语音助手，且要审计每次调用” | **Realtime API（WebSocket） + 业务服务端代理** | 通过服务端统一注入 `workspace_id` 和鉴权头，天然实现租户隔离；所有请求经代理可记录完整 trace 日志，满足 SOC2 审计要求。 | +| “想快速验证 Qwen-Omni 的语音生成效果，本地 Node.js 脚本跑通就行” | **Omni Realtime API**（WebSocket） | 无需安装 SDK、无需处理信令，几行代码即可建立连接、发送音频、接收流式响应，原型验证效率最高。 | + +> 💡 **终极建议**： +> - **从协议出发选型**：先明确终端类型（Web/iOS/Android/Server）与网络环境（强网/弱网/局域网），再决定用 WebRTC/AOQ/WebSocket； +> - **再匹配模型能力**：若需 `semantic_vad` 或联网搜索，锁定 `qwen3.5-omni-realtime` → 选 Omni；若需 `livetranslate` 或 `multimodal-dialog` → 选 Realtime； +> - **最后评估工程成本**：团队是否熟悉 WebRTC SDP？是否有能力维护 AOQ Token 服务？这些将直接影响上线周期。 + +如需进一步比对具体模型参数、错误码处理或迁移路径，请查阅对应 SDK 文档与 [实时 API 最佳实践](https://help.aliyun.com/zh/model-studio/realtime-api-best-practices)。 + +## 被对比主题页 + +- [omni realtime api](../api/omni-realtime-api.md) +- [realtime api user guide](../api/realtime-api-user-guide.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md b/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md deleted file mode 100644 index 5b2f9be9..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md +++ /dev/null @@ -1,81 +0,0 @@ -# 智能体应用 - -智能体应用（Agent）是阿里云百炼平台的核心应用构建模式之一，通过自然语言零代码配置，让大模型基于角色设定自主决策、动态规划并调用知识库、MCP、Skill 等工具来完成任务。相较于流程固定的工作流应用，智能体强调 AI 的自主性，适合意图开放、需要动态编排的对话与轻量任务场景。 - -## 版本演进：Agent 1.0 与 Agent 2.0 - -百炼提供两代技术架构不同的智能体，**不支持直接升级或版本切换**，迁移需重新创建： - -- **新版智能体（Agent 2.0）**：2025 年 12 月 26 日上线。将知识库、MCP 等能力统一抽象为「工具」，由智能体自主规划调用时机与顺序，并完整展示「规划-执行-反思」链路。无旧版依赖时推荐使用。**仅支持 API 调用，不支持任何分享渠道**（魔笔/UI、钉钉、微信、组件、音视频互动）。 -- **旧版智能体（Agent 1.0）**：通过知识库（RAG）+ 插件扩展能力，先检索知识再决策是否调用工具，适合意图单一、流程固定的简单任务。自定义插件有 **5 秒超时限制**。分享渠道均为 1.0 功能。 - -## 核心能力配置（Agent 2.0） - -- **模型选择**：推荐具备强工具调用能力的模型（如千问-Max 系列）；可配置最长回复长度、`temperature`、`enable_thinking`（思考模式）等参数。 -- **提示词（System Prompt）**：定义角色、行为指令与能力边界，支持自定义变量嵌入。 -- **内置工具**：沙箱环境下的 `bash`、`write`、`read`、`edit`、`glob`、`grep`、`download_file`，默认关闭需按需开启。 -- **知识库**：作为工具由智能体自主调用，支持标签过滤限定查询范围。 -- **MCP**：外部工具以 MCP 协议接入，支持动态非固定顺序调用。 -- **Skill**：可扩展能力包，智能体在对话中自动识别匹配任务并调用，无需额外编码。 -- **记忆**：短期记忆支持 0-30 轮上下文；长期记忆暂未支持。 -- **ReAct 最大轮次**：取值 1-50，限制单次会话内工具调用最大次数。 - -## 文件问答 - -智能体应用支持上传文件进行问答，提供三种处理模式： - -| 模式 | 适用场景 | 特点 | -|------|---------|------| -| 全文引用 | 文档总结、全文翻译 | 简单直接，受上下文长度限制 | -| 切片检索（RAG） | 长文档问答、知识库检索 | 能处理超长文件，效果依赖检索策略 | -| 自定义处理 | 图片转换、视频分析等 | 功能灵活，依赖配置的工具 | - -限制：单会话最多 10 个文件，单文件不超过 10 MB（超出需用文件上传 API）。支持文档、图片、视频、音频等格式。 - -## 发布与调用 - -所有应用类型均需先发布才能通过 API 集成，核心步骤：在应用配置页点击「发布」→ 在「发布渠道」查看调用方式。RAM 账号发布前需拥有 `ram:CreateServiceLinkedRole` 权限。 - -API 调用与工作流应用完全一致，通过 `Application.call` / `POST /apps/{app_id}/completion` 触发： - -```python -import os -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) -print(response.output.text) -``` - -响应结构为 `{"output": {"finish_reason", "session_id", "text"}, "usage": {...}, "request_id": "..."}`，业务侧主要消费 `output.text`。 - -## 分享与组件化（仅 Agent 1.0） - -- **分享渠道**：UI 应用/魔笔、钉钉、微信公众号、音视频实时互动。UI 体验链接与音视频临时二维码有效期均为 **24 小时**。分享产生的费用由应用创建者 UID 账号承担。 -- **组件化**：智能体或工作流可发布为模块化组件供其他应用复用。接入智能体时组件作为工具，大模型据「组件描述」自动判断调用；预设系统参数 `query`、`imageList` 无法删除。注意避免嵌套调用（A↔B）和多级调用（A→B→C 易超时）。 - -## 与 Managed Agents 的区别 - -Managed Agents 是服务端托管运行时，与无状态的智能体应用不同：它在独立云端沙箱容器中维护会话状态，支持中断与续接、事件历史持久化，面向多步工具调用、代码执行、文件处理等长时运行任务。 - -## 计费说明 - -- **模型调用**：按模型类型和 Token 用量计费。 -- **知识库**：按量付费，召回的文本切片会增加输入 Token；自 2026 年 1 月 4 日起正式计费。 -- **MCP/插件**：部分官方 MCP 按调用计费，第三方 MCP 由第三方收取。 - -百炼提供限时免费额度，可在模型广场查看。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [managed agents](../guides/managed-agents.md) -- [application publishing and sharing](../guides/application-publishing-and-sharing.md) -- [start using](../guides/start-using.md) -- [skill](../guides/skill.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md b/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md deleted file mode 100644 index 72e01839..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md +++ /dev/null @@ -1,90 +0,0 @@ -# 智能体编排 - -智能体编排是指在阿里云百炼平台上，通过配置或代码组织智能体（Agent）、工具、知识库、子智能体等资源之间的调用关系与执行流程，让大模型在受控边界内自主规划并完成业务任务的能力。 - -## 在百炼中的使用方式 - -百炼支持三种互补充的编排形态，开发者可单独使用或组合集成： - -| 形态 | 编排方式 | 控制力 | 适用场景 | -| --- | --- | --- | --- | -| 新版智能体（Agent 2.0） | 自然语言配置，零代码 | 由大模型自主规划 | 智能客服、知识问答、任务助理 | -| 工作流应用 | 可视化节点编排，低代码 | 节点固定、流程确定可复现 | 报告生成、订单处理、审批流 | -| 高代码应用 | Python 编码 | 完全由代码定义逻辑 | 私有算法部署、深度定制 | - -### 1. 新版智能体编排 - -Agent 2.0 将知识库、MCP 统一抽象为「工具」，由智能体在每轮「规划—执行—反思」（ReAct）链路中自主决定调用顺序。关键配置： - -- **模型**：推荐 `千问-Max` 等工具调用能力强的模型；可配 `temperature`、`enable_thinking`、最长回复长度。 -- **系统提示词**：定义角色、行为指令、能力边界；支持 `/变量` 引用自定义变量。 -- **知识库**：作为工具被自主调用，可用标签限定查询范围。 -- **MCP / 插件**：以 MCP 协议接入外部工具，支持动态非固定顺序调用；插件可一键转 MCP。 -- **内置工具**：`bash`、`read`、`write`、`edit`、`glob`、`grep`、`download_file`，在隔离沙箱中默认关闭、按需开启。 -- **记忆**：短期 0–30 轮上下文。 -- **ReAct 最大轮次**：1–50，限制单次会话工具调用次数，超出自动退出并生成最终回复。 - -> 注意：旧版智能体与新版本架构不兼容，不支持升级、降级或版本切换。 - -### 2. 工作流编排 - -通过可视化节点将复杂任务拆解为有序步骤，逻辑确定。核心节点： - -- **开始 / 结束**：定义输入输出参数，开始节点预置 `query`、`historyList`、`imageList` 等。 -- **大模型节点**：执行 LLM 推理，配置模型、提示词、用户提示词、记忆。 -- **意图分类节点**：按输入分流到不同下游分支。 -- **变量处理节点**：文本输出或变量加工。 -- **智能体群组节点**：把任务分解给多个已发布的子智能体协同完成，是工作流编排多智能体协作的核心。 - -会话变量作为全局变量在工作流全生命周期内记录参数，可在画布右上角配置并在各节点引用。 - -### 3. 高代码编排 - -基于完整 Python 项目结构部署 AI 后端： - -- **部署形态**：Serverless Function（无状态、快速拉起、低成本）或 K8s（高性能、有状态、长程任务）。 -- **代码提交**：使用预置模板（基础对话 / 工具调用 / 深度研究 Agent）或上传本地 `.whl` 包。 -- **MCP 工具接入**：控制台直接关联知识库、工作流、插件等 MCP 服务。 -- **网关**：应用稳定后建议开启网关，通过自定义域名和路由在生产环境访问。 - -## API 层编排 - -### Managed Agents API（托管运行时） - -平台托管会话、沙箱、工具执行与事件流，开发者通过 REST 或 DashScope SDK 管理 Agent、Environment、Skill、File、Session 等资源： - -- **Agent**：可复用的智能体配置（模型、系统提示词、工具包、技能）。 -- **Environment**：工具调用的执行沙箱与预装依赖，可被多会话复用。 -- **Session**：智能体的一次运行实例，绑定 Agent 与 Environment 快照。 -- **Skill**：以 zip 包封装的工具组合与文档，挂载时锁定版本。 -- **File**：独立文件资源，可挂载到沙箱供工具读写或作为消息内容。 - -调用流程四步：创建 Agent（一次创建长期复用）→ 创建 Session（每轮对话新建）→ 发送 Event（写入用户消息触发 `running`）→ 订阅 SSE 流式接收回复直至回到 `idle`。 - -### 应用调用 API - -将控制台编排好的应用集成进业务系统，两种接口： - -- **OpenAI 兼容 Responses API**：`POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`，复用 OpenAI 生态，支持同步/异步、流式、多模态。 -- **DashScope 原生 API**：`POST /api/v1/apps/{APP_ID}/completion`，功能更全，新版智能体、工作流、旧版智能体均支持。 - -调用前需准备 APP ID 与 API Key（推荐写入 `DASHSCOPE_API_KEY` 环境变量）。多轮对话可通过 `session_id`（系统自动加载历史，有效期 1 小时、最多 50 轮）或自行维护 `messages` 数组实现。 - -## 选型建议 - -- 需要 AI 自主决策、动态规划 → 新版智能体。 -- 流程固定、要求稳定可复现 → 工作流，复杂协作用智能体群组节点。 -- 深度定制、私有算法 → 高代码应用。 -- 需要平台托管调度与执行基础设施 → Managed Agents API。 -- 仅把控制台应用接入业务系统 → Application.call / completion。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [managed agents api](../api/managed-agents-api.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [application call](../api/application-call.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md b/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md deleted file mode 100644 index a6884fa8..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md +++ /dev/null @@ -1,80 +0,0 @@ -# API Key 鉴权 - -API Key 是调用阿里云百炼平台模型与应用的核心鉴权凭证，以 `Authorization: Bearer ` 的形式随请求携带，用于标识调用者身份并界定其可访问的模型、应用与业务空间范围。 - -## 在百炼平台的使用场景 - -API Key 贯穿百炼的各类调用入口，不同场景下的获取与使用方式略有差异： - -- **模型 API 直连**：调用文本、图像、视频、语音、向量等模型前，先在控制台对应地域的 **API Key** 页面创建 Key，配合服务端点 `base_url` 发起请求。 -- **应用与知识库调用**：智能体应用、工作流应用、知识检索/问答接口（DashScope 应用网关）均通过 API Key Bearer 鉴权，Base URL 常需拼接业务空间 ID，如 `https://{workspaceId}.cn-beijing.maas.aliyuncs.com`。 -- **框架集成**：LlamaIndex（RAG）、Spring AI Alibaba（智能体/工作流/知识库检索）均以 API Key 鉴权复用百炼能力，通常通过环境变量注入。 -- **百炼 CLI（`bl`）**：支持控制台登录、`--api-key`、环境变量、配置文件、临时传入等多种认证方式，可组合使用。 -- **不可信环境（浏览器/移动端）**：应由后端生成临时 API Key，避免永久 Key 泄露。 -- **子业务空间**：需使用该空间自身的 API Key，并预先为其授予相应模型的调用权限。 - -## Key 的类型与前缀 - -百炼的 API Key 因计费方式不同而彼此隔离，前缀是重要区分标识： - -| 类型 | 前缀 | 说明 | -| --- | --- | --- | -| 按量付费（安全升级后） | `sk-ws` | 仅创建时展示一次明文，关闭弹窗后无法再查看，务必立即保存 | -| 按量付费（升级前旧 Key） | `sk-` | 仍可正常使用 | -| Token Plan / Coding Plan 专属 | `sk-sp-` | 套餐专属，与通用 Key 不可混用 | -| 临时 API Key | `st-` | 由后端接口生成、限时有效 | - -> API Key 与其配套的 Base URL 必须成对使用。按量付费、Token Plan、Coding Plan 三者的 Key 与 Base URL 混用会导致意外扣费或返回 401/403 鉴权失败。 - -## 创建时的关键选项 - -- **归属业务空间**：决定 Key 的调用权限。默认空间的 Key 可调用所有标准模型及默认空间应用；子空间的 Key 仅能调用已授权模型及本空间应用。同一空间内的 Key 权限相同，无需按模态分别创建。 -- **权限**：可选「全部」或「自定义」。自定义可配置 IP 白名单（最多 20 个 IPv4/IPv6 地址或网段）及可访问的模型/应用范围。 -- **创建主体**：需主账号，或具备「管理员」/「API-Key」页面权限的子账号。RAM 子账号在订阅套餐前需主账号授予 `AliyunBailianFullAccess` 等权限。 - -## 配置与使用 - -推荐将 API Key 写入环境变量，避免硬编码泄漏： - -- 通用约定使用 `DASHSCOPE_API_KEY`。 -- Spring AI Alibaba 应用集成用 `DASHSCOPE_API_KEY`，知识库检索用 `AI_DASHSCOPE_API_KEY`（约定不一致，关键是 `application.yml` 中占位符与实际变量名匹配）。 - -调用时除 API Key 外，还需指定与地域、协议匹配的 `base_url`（OpenAI 兼容协议与 Anthropic 兼容协议不同）。 - -## 临时 API Key - -面向浏览器、移动端等不可信环境，通过后端接口用永久 Key 换取限时凭证： - -``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800" \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" -``` - -关键参数与特性： - -- `expire_in_seconds`：有效期（TTL），单位秒，范围 `[1, 1800]`，默认 60 秒。 -- 返回体中 `token`（`st-` 前缀）为临时 Key，`expires_at` 为过期 UNIX 时间戳。 -- 临时 Key 继承永久 Key 的全部权限（含模型/知识库访问限制），到期自动失效，无法提前删除。 -- 各地域（北京 / 新加坡 / 弗吉尼亚）的 Key 与 Endpoint 不互通，需配套使用。 - -## 编程化管理 - -除控制台外，百炼提供 OpenAPI（`CreateApiKey` / `GetApiKey` / `ListApiKeys` / `UpdateApiKey` / `DeleteApiKey` / `EnableApiKey` / `DisableApiKey` / `ResetApiKey`）以编程方式管理 Key。这些接口使用阿里云账号 AccessKey 签名认证，并需具备相应 RAM 权限，与业务调用所用的 API Key Bearer 鉴权不同。 - -## 开发者要点 - -- 优先用环境变量注入，切勿硬编码或提交到代码仓库。 -- 升级后的 `sk-ws` Key 只显示一次，创建后立即保存。 -- 明确当前场景使用的是按量付费、Token Plan、Coding Plan 中的哪种 Key，并搭配对应 Base URL。 -- 不可信环境一律走临时 API Key，生产环境的文件存储不要依赖临时 URL。 - -## 关联主题页 - -- [preparations](../api/preparations.md) -- [more](../api/more.md) -- [frameworks](../api/frameworks.md) -- [more about models](../api/more-about-models.md) -- [token plan guide](../guides/token-plan-guide.md) -- [knowledge](../api/knowledge.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md b/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md deleted file mode 100644 index 8e78a250..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md +++ /dev/null @@ -1,53 +0,0 @@ -# 异步调用与任务轮询 - -异步调用是百炼平台为耗时较长的模型任务（图像生成、视频生成、3D 生成等）设计的调用模式：客户端先提交任务拿到 `task_id`，再通过轮询或事件通知获取最终结果，从而避免长连接等待与请求超时。 - -## 核心流程：创建任务 → 轮询获取 - -异步调用统一分为两步： - -1. **创建任务**：向对应能力的生成端点发起 `POST` 请求，请求头必须携带 `X-DashScope-Async: enable`，否则会报错 `current user api does not support synchronous calls`。请求成功后返回一个 `task_id`。 -2. **轮询查询结果**：用该 `task_id` 发起 `GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`，读取 `output.task_status` 直到任务进入终态（`SUCCEEDED` / `FAILED`），再从结果中取回产物 URL。 - -任务状态的典型流转为：`PENDING`（排队中）→ `RUNNING`（处理中）→ `SUCCEEDED`（成功）/ `FAILED`（失败）。不同能力还可能出现 `SUSPENDED`（挂起）、`CANCELED`（已取消）、`UNKNOWN`（任务不存在或已过期）等状态。 - -## 在不同场景中的使用 - -- **3D 生成（Tripo）**：仅支持异步，端点为 `.../aigc/video-generation/3d-generation`，轮询建议间隔约 15 秒，生成耗时较长，产物下载链接有效期仅 2 小时。 -- **视频生成**：所有厂商模型（万相、PixVerse、Vidu、可灵等）统一走异步，端点通常为 `.../aigc/video-generation/video-synthesis`（部分数字人/换人类模型使用 `.../aigc/image2video/video-synthesis`）。单次任务通常耗时 1-5 分钟，个别统一编辑模型约 5-10 分钟。 -- **图像生成**：多数传统模型仅支持异步（`text2image` / `image2image` 等端点），生成通常需 1-2 分钟；而新版模型（wan2.6 / wan2.7、z-image-turbo 等）走 `multimodal-generation/generation` 端点，支持 HTTP 同步一次拿结果。请勿把同步协议用在旧模型上。 -- **应用调用（智能体/工作流）**：Responses API（OpenAI 兼容）通过设置 `background=true` 开启异步，创建任务后返回任务 ID 再轮询；DashScope 应用调用接口目前暂不支持异步。 - -## 关键参数与配置 - -- **请求头**：创建任务必须带 `X-DashScope-Async: enable`；同时需 `Authorization: Bearer $DASHSCOPE_API_KEY`、`Content-Type: application/json`。 -- **`task_id` 有效期**：一般为 **24 小时**，过期查询返回 `UNKNOWN`。请勿重复创建任务，轮询即可。 -- **`background`（应用 Responses API）**：布尔值，默认 `false`，设为 `true` 开启异步执行。 -- **查询限流**：任务查询接口默认约 20 QPS / RPS（主账号维度）。 - -## 异步任务管理与完成通知 - -百炼提供三个通用的异步任务管理接口： - -- **查询单个任务**：`GET .../api/v1/tasks/{task_id}`。 -- **批量查询**：`GET .../api/v1/tasks/?start_time=xxx&end_time=xxx&status=xxx`，单次时间跨度不超过 24 小时。 -- **取消任务**：`POST .../api/v1/tasks/{task_id}/cancel`，仅能取消 `PENDING` 状态的任务。 - -为避免频繁轮询浪费资源并触发限流，百炼已接入阿里云事件总线 EventBridge，支持任务完成后主动推送通知（HTTP 回调 URL 或 RocketMQ 两种方案）。事件源为 `acs.dashscope`，事件类型为 `dashscope:System:AsyncTaskFinish`，收到通知后只需调用一次查询接口即可获取结果。 - -## 开发者建议 - -- 妥善保存 `task_id`，切勿因未及时轮询到结果而重复创建任务。 -- 轮询间隔不宜过密（如约 15 秒），高频或生产场景优先使用异步完成通知。 -- 及时下载产物：图像/视频 URL 有效期通常为 24 小时，3D 模型下载链接有效期仅 2 小时。 -- 处理失败时读取响应中的 `code` 与 `message`，对照百炼错误码文档排查。 - -## 关联主题页 - -- [3d generation](../api/3d-generation.md) -- [video generation api](../api/video-generation-api.md) -- [image generation](../api/image-generation.md) -- [more about models](../api/more-about-models.md) -- [application call](../api/application-call.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/asynchronous-task.md b/skills/bailian-docs-llm-wiki/wiki/concepts/asynchronous-task.md new file mode 100644 index 00000000..1345167d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/asynchronous-task.md @@ -0,0 +1,51 @@ +# 异步任务 + +异步任务是百炼平台对耗时较长的模型推理请求（如图像生成、视频生成、3D建模等）所采用的标准执行模式：客户端发起请求后立即返回任务标识符（`task_id`），实际计算在服务端后台异步执行，结果需通过轮询或事件回调方式获取。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +- **图像生成**：`kling`、`vidu`、`wanx-v1` 等模型强制使用异步模式。调用 `POST /api/v1/services/aigc/multimodal-generation/generation` 创建任务，返回 `task_id`；再通过 `GET /api/v1/tasks/{task_id}` 轮询状态，直至 `task_status == "SUCCEEDED"` 后提取 `output.results[].url`（图片 URL 有效期 24 小时）。 + +- **视频生成**：全部模型统一启用异步调用，必须设置请求头 `X-DashScope-Async: enable`。创建任务后获得 `task_id`（有效期 24 小时），后续轮询 `/api/v1/tasks/{task_id}` 获取视频下载链接（URL 有效期通常为 2 小时）。 + +- **3D生成**：Tripo 模型仅支持异步，且**强约束地域为华北2（北京）**。请求需携带 `X-DashScope-Async: enable`，成功响应含 `task_id`；轮询间隔建议 ≥15 秒，结果中 `pbr_model_url` 或 `base_model_url` 有效期仅 2 小时，需及时下载。 + +- **模型微调与部署**：微调作业（`fine_tuning_jobs`）和部署操作本身即为异步任务。提交后返回 `job_id` 或 `deployment_id`，需轮询对应资源 endpoint（如 `GET /v1/fine_tuning_jobs/{job_id}`）确认完成状态，不可同步等待。 + +- **通用原则**： + - 所有异步任务均以 `task_id` 为唯一追踪凭证，格式为标准 UUID； + - 任务状态流转为 `PENDING` → `RUNNING` → `SUCCEEDED` / `FAILED` / `CANCELLED`； + - 避免高频轮询（推荐 ≥15 秒间隔），生产环境应优先配置 [EventBridge 回调](../../raw/model-api-reference/more-about-models/async-task-api.md) 接收 `dashscope:System:AsyncTaskFinish` 事件，实现低延迟、零限流的结果通知。 + +## 关键参数和配置 + +- `X-DashScope-Async: enable`：**必需请求头**，显式声明启用异步模式（部分 API 如视频生成强制要求，缺失将报错）； +- `task_id`：字符串类型，由平台生成并返回，用于后续查询、取消或获取结果； +- `expire_in_seconds`（可选）：部分异步管理接口支持指定任务保留时长（默认 24 小时），超期后任务记录与输出 URL 自动清理； +- `X-DashScope-OssResourceResolve: enable`（文件类任务）：当输入含 OSS 临时 URL 时需添加，确保平台正确解析资源； +- 回调配置（高级）：通过 EventBridge 绑定 HTTP Endpoint 或 RocketMQ Topic，接收结构化事件（含 `data.task_id`, `data.status`, `data.output`），替代轮询。 + +## 面向开发者，简洁实用 + +✅ **最佳实践**： +- 始终检查响应状态码 `200` 和 `task_id` 字段，勿假设请求已立即完成； +- 使用 SDK 的 `get_task_result(task_id)` 封装轮询逻辑（Python/Java SDK 均内置），避免手写重试； +- 生产环境禁用裸轮询，务必配置 EventBridge 回调——单次事件触发 + 单次结果查询，性能与稳定性双优； +- 所有异步输出 URL 均有时效性（2–24 小时不等），请在收到结果后立即下载或持久化存储； +- `task_id` 是调试核心线索：日志中记录它，控制台中可通过「异步任务管理」页面直接检索全量状态与错误详情。 + +⚠️ **避坑提示**： +- 不同模型/地域的 `task_id` 不互通，跨地域调用将失败； +- 未配置 `X-DashScope-Async: enable` 却调用异步专属 endpoint（如 `/api/v1/tasks/xxx`），将返回 `400 Bad Request`； +- 轮询频率超过 20 QPS 可能触发限流，导致 `429 Too Many Requests`； +- 任务超时（如 24 小时未完成）后 `task_id` 失效，无法再查询，需重发请求。 + +## 关联主题页 + +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) +- [3d generation](../api/3d-generation.md) +- [more about models](../api/more-about-models.md) +- [model production](../api/model-production.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md b/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md deleted file mode 100644 index 6b73c92e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md +++ /dev/null @@ -1,85 +0,0 @@ -# 计费 - -计费是百炼平台围绕模型推理、模型训练、模型部署及增值服务所建立的费用体系，涵盖按量付费、订阅制（[Token](token.md) Plan / Coding Plan）、预留容量（TPM 预留 / PTU）等多种方式，开发者可根据用量规模和业务场景灵活组合，实现成本最优。 - -## 计费场景与方式 - -百炼平台的计费覆盖以下主要场景： - -| 场景 | 计费方式 | 说明 | -|------|---------|------| -| 模型推理调用 | 按量付费（[Token](token.md) 计价） | 按输入/输出 [Token](token.md) 分别计价，部分模型支持阶梯计费 | -| 模型训练 | 按训练 Token 计费 | （训练数据 Token + 混合训练数据 Token）× 循环次数 × 单价 | -| 模型部署 | PTU / 模型单元 / Token 用量 | 三种方式创建后不可更改，需下线重建才能切换 | -| 知识库 | 按规格计费 | 标准版 0.03 元/小时，旗舰版 0.2 元/RCU/小时 | -| TPM 预留 | 按 kTPM 预付费 | 按天计费，输入/输出分别定价 | -| Token Plan 团队版 | 坐席订阅制 | 按 Credits 抵扣，198 元/坐席/月起 | -| Coding Plan | 月度订阅制 | 按模型调用次数限额，200 元/月 | - -## 费用抵扣顺序 - -多种付费方式并存时，系统按以下优先级自动抵扣： - -**免费额度 → 资源包 → 其他模型节省计划 → AI 通用型节省计划 → 按量付费** - -Token Plan 和 Coding Plan 的套餐额度独立于按量计费体系，不参与上述抵扣链路。 - -## 免费额度 - -首次开通百炼时自动发放新人免费额度，有效期 30~90 天。关键限制： - -- 仅适用于华北2（北京）地域、中国内地服务部署范围的实时推理 -- 不支持抵扣 Batch 调用、模型调优、模型部署、自定义模型 -- 主账号与 RAM 子账号共享 -- 建议开启"免费额度用完即停"功能，防止额度耗尽后自动扣费 - -## 成本优化方案 - -### AI 通用型节省计划 - -承诺每月消费金额获取阶梯折扣，最高可享 5.3 折。覆盖阿里直供全部模型，承诺金额 1,000 元起，支持 3/6/12/24 个月周期。当月未用完的额度自动清零，不可累积。 - -### 资源包 - -预先购买具体 Token 数量，用于抵扣特定模型的实时推理用量。适合用量明确且集中在单一模型的场景。 - -### Batch 调用优惠 - -支持 Batch 调用的模型，输入和输出 Token 单价均按实时推理价格的 50% 计费。 - -### 上下文缓存折扣 - -部分模型支持上下文缓存，命中缓存的输入 Token 按折扣系数消耗额度（例如 deepseek-v4-pro 为 0.08，即按 8% 折算）。Batch 调用与上下文缓存不能同时生效。 - -## 多地域定价差异 - -同一模型在不同地域的价格可能不同。华北2（北京）使用人民币定价，新加坡等国际地域按国际价格结算，通常高于国内价格。[API Key](api-key.md) 必须与 Base URL 同一地域，否则会报 `401` 错误。 - -## 账单查询 - -- **费用概览**：控制台"用量 & 费用 > 费用概览"查看当月总消费，支持按模型或 [API Key](api-key.md) 筛选 -- **模型用量**：按[业务空间](workspace.md)维度统计，数据延迟约 1 小时，不支持查看 30 天以前的数据 -- **出账时间**：大模型推理分钟级出账（2~10 分钟），批量推理和训练小时级出账 -- **分账管理**：通过[业务空间](workspace.md)标签按部门或项目归集费用，T+1 天生效 - -## 欠费与停止计费 - -账户可用额度小于 0 时视为欠费。Token Plan 和 Coding Plan 的套餐额度独立于账户余额，欠费期间可继续使用。停止计费的方式： - -- **模型推理**：停止 API 调用，删除不再使用的 [API Key](api-key.md) -- **模型部署**：在控制台下线已部署模型；包月预付费需额外退订实例 -- **模型训练**：无进行中的训练任务即不产生费用 -- **订阅制**：关闭自动续费，到期自动停止 - -## 关联主题页 - -- [test 1](../guides/test-1.md) -- [token plan guide](../guides/token-plan-guide.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [model monitoring](../guides/model-monitoring.md) -- [support](../guides/support.md) -- [knowledge base](../guides/knowledge-base.md) -- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md b/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md deleted file mode 100644 index 2c3d22ce..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md +++ /dev/null @@ -1,32 +0,0 @@ -# 上下文窗口 - -上下文窗口（Context Window）是模型在单次推理中能够接收并处理的输入与输出 Token 总长度的上限，单位通常为 Token，是衡量模型承载长文本、多轮对话、多模态输入能力的核心指标。在百炼平台，千问系列文本生成模型（如 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-flash`）及视觉理解模型普遍支持 1M（约百万）Token 的上下文窗口，第三方模型如 `deepseek-v4-pro` / `deepseek-v4-flash` 同样提供 1M 上下文。 - -## 在百炼中的使用场景 - -- **长文档处理**：1M 上下文可一次性容纳数十万字的中文文档或数十张高分辨率图像，适合合同审阅、代码仓库分析、长报告摘要等场景，避免分段截断导致信息丢失。 -- **多轮对话**：聊天与 Agent 应用需在请求中累积历史消息，上下文窗口决定可保留的对话轮数。超出窗口后早期消息会被截断，需结合检索或摘要策略压缩历史。 -- **多模态输入**：视觉理解模型同时接收图像、视频与文本，图像 Token 按 `高 × 宽 / (32 × 32) + 2` 估算，单张图最高约 1600 万像素；视频最长支持 2 小时 / 2GB（Qwen3.7/3.6/3.5 系列），1 小时 / 2GB（Qwen3.5-Omni 系列含音频）。多模态输入会快速占用上下文窗口，需在长文本与多图之间权衡。 -- **工具调用与思考模式**：开启 `enable_thinking` 的思考模式、Function Calling 的工具定义与中间结果都会计入上下文消耗，复杂 Agent 流程需为工具输出预留充足窗口预算。 - -## 关键参数与配置 - -- **模型选型**：上下文长度随模型档位而定，文本/视觉模型以 1M 为主；向量、[重排序](rerank.md)、语音等模型按各自规格独立设定，调用前应以模型广场标注为准。 -- **对话历史管理**：仅 OpenAI 兼容 Responses 接口由平台自动管理对话历史，无需手动拼接 `messages`；OpenAI Chat Completions、Anthropic Messages、DashScope 原生接口均需调用方自行维护上下文长度与轮次，避免超出窗口导致请求失败或内容截断。 -- **Token 计费与限流**：上下文窗口内的输入与输出 Token 均参与计费和限流计量，长上下文请求会显著提升单次调用成本与首字延迟，建议按需裁剪历史与文档片段。 -- **超时与域名**：长上下文推理耗时较高，生产环境推荐使用业务空间专属域名 `{WorkspaceId}.{region}.maas.aliyuncs.com`（请求超时 3600 秒、SLA 99.9%），避免中心化或试用域名的较短超时与限流影响。 - -## 注意事项 - -- 上下文窗口是模型能力上限，不代表稳定可用长度；接近上限时模型对长尾内容的注意力可能下降，关键信息宜放在输入首尾。 -- 跨接口迁移时需确认目标兼容接口是否完整透传长上下文相关参数，部分兼容接口为协议一致可能不暴露全部原生参数，最全参数请使用 DashScope 原生接口。 -- 各地域、各模型对上下文窗口的支持存在差异，API Key、模型列表与接入域名不能跨地域混用，调用前请在目标地域的模型广场核对规格。 - -## 关联主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [get started with models](../guides/get-started-with-models.md) -- [model experience](../guides/model-experience.md) -- [more about models](../api/more-about-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md b/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md deleted file mode 100644 index 2b0d787d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md +++ /dev/null @@ -1,112 +0,0 @@ -# 跨会话记忆 - -跨会话记忆是指将对话中提取的关键信息和用户画像持久化存储，并在后续会话中通过语义检索召回并注入 Prompt 的能力，用于解决大模型上下文窗口无法跨会话延续的问题。百炼平台通过「记忆库（Memory Library）」与「长期记忆 API」提供该能力。 - -## 在百炼平台中的使用场景 - -跨会话记忆在以下场景中发挥作用： - -- **个性化智能体**：在多轮对话之间保留用户偏好、习惯和重要事件（如「每天上午 9 点提醒我喝水」），让智能体在新会话中仍然理解用户历史。 -- **长期用户画像**：通过自定义画像模板提取结构化属性（年龄、职业、偏好等），在后续对话中以固定字段持久化注入，适用于需要稳定属性支撑的业务。 -- **Agent 自动捕获/召回**：OpenClaw Agent 通过记忆插件在 `before_agent_start`（自动召回）和 `agent_end`（自动捕获）两个生命周期钩子中与长期记忆 API 交互，实现零侵入的跨会话记忆。 -- **应用观测中的记忆追踪**：在应用观测的调用链路中，记忆的写入与检索会作为 RETRIEVER、EMBEDDING 等节点出现，便于开发者定位记忆相关调用的延时与 Token 消耗。 - -## 接入方式 - -### 方式一：API 直连 - -通过 HTTPS 调用 `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` 系列接口，请求 Header 携带 `Authorization: Bearer $DASHSCOPE_API_KEY`。典型流程为：对话结束调用 `AddMemory` 写入记忆 → 下次对话调用 `SearchMemory` 语义检索 → 将结果注入 Prompt。 - -```bash -# 写入记忆（自动从对话提取） -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午9点提醒我喝水"}, - {"role": "assistant", "content": "好的，已记录"} - ], - "user_id": "user_001" - }' - -# 语义检索记忆 -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "我需要做什么？"}], - "top_k": 5 - }' -``` - -Python 用户可安装 `agentscope-runtime`，使用 `AddMemory`、`SearchMemory`、`ListMemory`、`CreateProfileSchema`、`GetUserProfile` 等封装类（均需在 `finally` 中调用 `close()`）。 - -### 方式二：OpenClaw 记忆插件 - -```bash -openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw -openclaw plugins info modelstudio-memory-for-openclaw -openclaw modelstudio-memory stats -openclaw gateway restart -``` - -插件配置写入 `~/.openclaw/openclaw.json`，关键项：`slots.memory` 注册为记忆槽位（会自动禁用内置 `memory-core` 和 `memory-lancedb`）；`apiKey` 填 DashScope API Key；`userId` 用于隔离不同用户记忆空间。所有读写均由百炼服务端完成提炼、向量化和语义检索。 - -## 记忆内容类型 - -记忆库提供两类持久化内容，可独立或组合使用： - -- **记忆片段**：从对话中自动提取的关键事件和信息，支持自动去重、动态更新，也可通过 `custom_content` 直接写入指定内容。适用于大多数长期记忆场景。 -- **用户画像**：基于画像模板（profile schema）从对话中提取的结构化属性，适用于需要固定属性持久化存储的场景。属性字段应清晰具体，避免「姓名/名称/名字」等同义字段并存，且不应期望一次对话就提取全部信息。 - -## 关键参数 - -### AddMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 记忆实体 ID，用于标识归属对象，最大 64 个字符 | -| `messages` | 与 `custom_content` 二选一 | 对话消息列表，每个消息含 `role`（user/assistant）和 `content`，最多 50 条 | -| `custom_content` | 与 `messages` 二选一 | 自定义内容，最大 512 个字符，传入后忽略 `messages` | -| `profile_schema` | 否 | 画像模板 ID，在记忆库详情页获取 | -| `memory_library_id` | 否 | 记忆库 ID，最大 32 个字符，不传则使用默认记忆库 | -| `project_id` | 否 | 记忆片段规则 ID，不传则使用指定记忆库的默认规则 | -| `meta_data` | 否 | 用户自定义信息 | - -### SearchMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 用户 ID，用于隔离记忆空间 | -| `messages` | 是 | 查询对话内容，系统据此做语义检索 | -| `top_k` | 否 | 返回的记忆片段数量 | - -## 使用限制 - -| 项目 | 限制 | -| --- | --- | -| 全部接口总计 | 3000 QPM（阿里云账号级别） | -| 记忆片段 add 接口 | 120 QPM | -| 记忆片段 search 接口 | 300 QPM | - -## 有效期说明 - -记忆有效期在不同入口存在差异：长期记忆 API 文档指出「生成的记忆片段与用户画像暂无失效日期」，而控制台默认记忆片段规则预置了「默认有效期 180 天」，并支持按规则配置 7/30/180 天或永不过期。以控制台记忆规则配置为准；通过 API 直写且不指定 `project_id` 时使用默认规则。 - -## 注意事项 - -- OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置；不支持阿里云百炼 Coding Plan 的 API Key。 -- 应用观测暂不支持通过 Assistant API 创建的智能体应用；对高代码应用，仅能观测到入口 CHAIN 节点，不支持追踪其内部调用链路。 -- 记忆的写入与检索在应用观测中会体现为 RETRIEVER、EMBEDDING 等节点，可用于定位记忆相关调用的延时与 Token 消耗。 - -## 关联主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) -- [application monitoring](../guides/application-monitoring.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md b/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md deleted file mode 100644 index de46d3bd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md +++ /dev/null @@ -1,99 +0,0 @@ -# DashScope SDK - -DashScope SDK 是阿里云百炼平台官方提供的软件开发工具包，封装了模型与应用调用的原生（DashScope）接口，让开发者用少量代码即可接入通义千问、万相、Qwen-MT、Qwen-OCR 等模型能力以及智能体/工作流应用。相比 HTTP 直连和 [OpenAI 兼容接口](openai-compatible-interface.md)，DashScope 原生接口暴露的参数最完整、功能集最丰富。 - -## 适用语言与安装 - -DashScope SDK 主要提供 Python 与 Java 两种官方实现，部分场景也可用 HTTP（如 Node.js 借助 `axios`）替代： - -- **Python**：`python3 -m pip install -U dashscope` -- **Java**：通过 Maven / Gradle 引入 `com.alibaba:dashscope-sdk-java`，建议版本 `>= 2.12.0` -- **Node.js / 其他语言**：目前无官方 SDK，直接走 HTTP API（发起 POST 请求） - -> 注意：不同能力对 SDK 语言与地域的支持存在差异。例如 `qwen-deep-research` **仅支持 Python DashScope SDK，且仅限华北2（北京）地域**，暂不支持 Java SDK 与 [OpenAI 兼容接口](openai-compatible-interface.md)。 - -## 在不同场景中的使用 - -### 1. 调用文本生成模型（Qwen 系列） - -Qwen 系列可通过 OpenAI 兼容、Anthropic 兼容或 DashScope 原生三类接口调用。其中 DashScope 是百炼原生接口，**功能集最完整、参数支持最丰富**；当需要使用最全的采样参数、插件或业务字段而兼容接口未暴露时，应改用 DashScope 原生接口。 - -### 2. 调用专用模型（[more](../api/more.md) models） - -法律、意图理解、翻译、OCR 等专用模型大多支持 OpenAI 兼容或 DashScope 两种方式调用： - -- `farui-plus`（法律大模型）：通过 DashScope SDK（Python / Java）调用 -- `tongyi-intent-detect-v3` / `qwen-mt-plus` / `qwen3.5-ocr`：OpenAI 兼容或 DashScope 均可 -- `qwen-deep-research`（深度研究）：仅 Python DashScope SDK - -### 3. 调用图像生成与编辑模型 - -千问-图像（Qwen-Image）、万相（Wan/Wanx）、Z-Image 等图像模型均可通过 HTTP 或 DashScope SDK 调用，覆盖文生图、图像编辑、图像翻译、风格迁移等能力。 - -### 4. 调用智能体应用与工作流应用 - -已创建并发布的智能体应用、工作流应用可通过 DashScope SDK 集成到业务系统，二者调用方式一致： - -- Python：`from dashscope import Application`，调用 `Application.call(...)` -- Java：构造 `ApplicationParam` 后调用 `application.call(param)` -- HTTP 等价接口：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` - -响应统一为 `{"output": {...}, "usage": {...}, "request_id": "..."}` 结构，业务侧主要消费 `output.text`。 - -## 关键参数与配置 - -### 鉴权（API Key） - -- SDK 通过 [API Key 鉴权](api-key.md)，**推荐将密钥写入环境变量 `DASHSCOPE_API_KEY`**，SDK 会自动读取，避免在代码中硬编码。 -- 调用特定地域（如华北2/北京）或子业务空间下的模型/应用时，需使用对应地域的 API Key，并按需提供 Workspace ID。 - -### 通用请求参数 - -- `model`（string，必选）：目标模型名称。 -- `messages`（array）：对话消息列表，按顺序排列，需由调用方维护上下文。 -- `app_id`（应用调用）：目标应用 ID，从控制台应用卡片复制。 -- `prompt` / `input`：用户输入内容。 -- `stream`（bool，可选）：是否[流式输出](streaming.md)；如 `qwen-deep-research` 的反问阶段需设为 `true`。 -- `session_id`（应用多轮对话）：由云端维护上下文，免去手动拼接历史。 - -### 模型专属参数示例 - -- **Qwen-MT（翻译）**：通过 `translation_options`（OpenAI SDK 中放入 `extra_body`）控制 `source_lang`、`target_lang`、`terms`（术语干预）、`tm_list`（翻译记忆）、`domain_prompt`（领域提示）。 -- **Qwen-OCR**：`messages.content` 为[多模态](multimodal.md)数组，可设 `min_pixels` / `max_pixels` 控制图像像素阈值。 - -## Python 快速示例（应用调用） - -```python -import os -from http import HTTPStatus -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) - -if response.status_code != HTTPStatus.OK: - print(f'code={response.status_code}, message={response.message}') -else: - print(response.output.text) -``` - -## 使用建议 - -- 优先使用环境变量管理 API Key，区分不同地域的密钥。 -- 需要最完整功能与参数时选 DashScope 原生接口；追求生态兼容、迁移成本最低时可选 [OpenAI 兼容接口](openai-compatible-interface.md)。 -- 接入前先对照具体模型的 API 参考，确认其支持的 SDK 语言、协议与地域。 - -## 关联主题页 - -- [image generation](../api/image-generation.md) -- [more models](../api/more-models.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md b/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md deleted file mode 100644 index 1334d698..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md +++ /dev/null @@ -1,54 +0,0 @@ -# 向量嵌入 - -向量嵌入（Embedding）是将文本、图像、视频等非结构化数据转换为固定维度的数值向量，使语义相近的内容在向量空间中距离也相近。百炼平台提供多类嵌入模型，支撑语义检索、RAG 召回、跨模态搜索、聚类推荐等下游任务。 - -## 在百炼平台的使用场景 - -- **RAG 知识库召回**：知识库创建时选择向量模型，对导入文档切片做嵌入入库；查询时对 Query 做同款嵌入，再与切片向量做相似度匹配，作为 RAG 流程的第一段召回。文档搜索、数据查询、音视频搜索类知识库支持 `text-embedding-v4` 或 `text-embedding-v3`（均为 512 维，维度不可更改）；图片问答类固定使用 `multimodal-embedding-v1`（1024 维）；视觉理解场景自动切换为 `qwen3-vl-embedding`。 -- **本地 RAG 应用**：本地知识库方案默认调用百炼 embedding API 生成向量，也可替换为本地部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`）。受 embedding API 限流影响，单文件不建议超过 100 MB。 -- **跨模态检索**：多模态向量模型（如 `qwen3-vl-embedding`、`multimodal-embedding-v1`）将文本、图像、视频映射到同一语义空间，支持以文搜图、以图搜视频等。支持「独立向量」（逐项生成）与「融合向量」（多输入合并为 1 个向量）两种模式。 -- **框架集成**：LlamaIndex 路线将知识库部署在百炼云端，使用官方向量模型与智能切分，不支持自定义嵌入模型；如需灵活选择嵌入模型，应改用本地知识库方案。 - -## 模型与关键参数 - -通用文本向量模型当前推荐 `text-embedding-v4`（Qwen3-Embedding 系列，支持 100+ 语种）。 - -| 模型 | 向量维度 | 最大行数 | 单行最大 Token | 语种 | -|------|---------|---------|---------------|------| -| text-embedding-v4 | 2048/1536/1024(默认)/768/512/256/128/64 | 10 | 8,192 | 100+ 语种 | -| text-embedding-v3 | 1024(默认)/768/512/256/128/64 | 10 | 8,192 | 50+ 语种 | -| text-embedding-v2 | 1,536 | 25 | 2,048 | 10 语种 | -| text-embedding-v1 | 1,536 | 25 | 2,048 | 6 语种 | - -请求参数： - -- `model`（必选）：模型名称。 -- `input`（必选）：字符串、字符串列表或文件。 -- `dimensions`（可选）：指定向量维度，仅 v3/v4 支持。 -- `encoding_format`（可选）：当前仅支持 `float`。 - -调用方式支持 OpenAI 兼容接口（base_url：`https://dashscope.aliyuncs.com/compatible-mode/v1`）和 DashScope SDK。 - -## 批处理接口 - -大规模文本向量化可使用异步批处理接口（`text-embedding-async-v1/v2`），单次最多 10 万行文本。需在 HTTP 请求头加 `X-DashScope-Async: enable` 启用异步模式，提交后通过 `task_id` 轮询结果。同时处理中任务不超过 50 个，并发运行上限 3 个，超出部分排队等待。 - -## 与 Rerank 的关系 - -向量嵌入负责「召回」，Rerank 模型负责「精排」。在知识库检索流程中，先由向量 + 关键词混合检索召回 TopK 切片，再由 `qwen3-rerank`、`qwen3-vl-rerank` 等排序模型对候选切片二次排序，最终按相似度阈值与最大召回数量返回。两者配合提升 RAG 命中准确率。 - -## 注意事项 - -- 知识库向量模型与维度在创建时选定，**维度不可更改**；Meta 抽取、多轮对话改写等索引配置在创建后也无法追加，需重建知识库。 -- 嵌入与 Rerank 是两类不同模型，不要混用接口：qwen3-rerank 走 `/compatible-api/v1/reranks`，qwen3-vl-rerank / gte-rerank-v2 走 `/api/v1/services/rerank/text-rerank/text-rerank`。 -- `gte-rerank` 系列将于 2026 年 5 月 30 日下线，建议迁移到 `qwen3-rerank`。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) -- [frameworks](../api/frameworks.md) -- [application use cases](../guides/application-use-cases.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md b/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md deleted file mode 100644 index a303b8d6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md +++ /dev/null @@ -1,77 +0,0 @@ -# 评测体系 - -评测体系是百炼平台用于系统化衡量输出质量的一整套能力，覆盖**应用评测**（智能体/工作流应用）与**模型评测**（文本生成模型）两大场景，通过评测集、评测维度/评估器、评分方式与评测报告构成完整的评估闭环。 - -## 两类评测场景 - -百炼的评测能力按被评测对象分为两条相对独立的链路： - -- **应用评测**：评估已发布智能体应用、工作流应用的回答质量，支持自动评测（大模型基于知识库生成评测集并打分）和手动评测（人工标注打分）。当前存在新旧两套系统，新版以「评测任务 + 评估器 + 标签」组织。 -- **模型评测**：评估文本生成类模型的能力，支持自定义评测（自有数据集 + 自定义维度）和基线评测（C-Eval、MMLU、GSM8K、BBH 等公开数据集）。 - -两者共享一套核心思路：准备评测数据 → 定义评分规则 → 创建评测任务 → 查看报告。 - -## 评分方式 - -无论应用还是模型评测，评分方式基本归为三类，选型取决于是否有标准答案与是否需要语义理解： - -| 评分方式 | 原理 | 适用场景 | 费用 | -|----------|------|----------|------| -| 大模型评估（LLM/AI 自动评测） | 裁判模型按 Prompt 语义打分或分类 | 相关性、幻觉、内容安全等语义场景 | 产生 Token/裁判模型费用 | -| 规则评估（Code/自动化指标） | ROUGE、BLEU、Cosine、字符串匹配等算法 | 翻译、摘要、格式校验、精确匹配等确定性场景 | 无额外费用 | -| 人工评估 | 人工逐条标注 Pass/Fail 或打分 | 创意写作、专业领域主观判断 | 无裁判费用 | - -**选型路径**：有标准答案且格式固定 → 字符串匹配；有标准答案但表述多样 → 文本相似度；无标准答案需语义理解 → 大模型评估；需主观判断 → 人工评估。 - -## 评测集（数据基础） - -评测集是评测任务的数据输入，需在数据管理模块或评测系统中准备： - -- **应用评测集**：旧版分对话分析（`.xls/.xlsx`）与知识问答（`.jsonl`）；新版分智能体、工作流、自定义三类，具备版本管理，创建后类型不可改。 -- **模型评测集**：单轮对话，Excel 格式，每行含 **Prompt**（用户输入）与 **Completion**（期望输出）。参评模型对每条 Prompt 推理，评分参考 Completion。 - -数据量建议：小规模验证 50-100 条，正式评测 200-500 条，全面评估 500 条以上。 - -## 评测维度与评估器（评分规则） - -评分规则是评测体系的核心组件，创建为模板后可被多个任务复用： - -- **模型评测的评测维度**提供五种评分器类型：大模型评估-数值型、大模型评估-分类型、规则评估-文本相似度、规则评估-字符串匹配、人工评估-分类型。维度类型创建后不可修改。 -- **应用评测的评估器**支持预置模板（通用质量、智能体、文本匹配、文本相似度、格式校验）、自定义（LLM 评估器 / Code 评估器）以及基于历史评测任务标注结果抽象生成。每个评测任务最多添加 10 个评估器，建议组合 3-5 个从不同维度评估。 - -## 关键参数与配置 - -- **评分范围**（数值型）：整数打分区间，默认 0-5，建议不超过 10，范围过大会降低 LLM 评分一致性。 -- **通过阈值**：判定 Pass/Fail 的分界线，数值型步长 0.1，相似度型步长 0.01。 -- **评分器 Prompt**（大模型评估）：至少含一个变量（`${prompt}`、`${output}`、`${completion}`），长度不超过 50000 字符；用于指导裁判模型如何打分。 -- **System Prompt**：配置于评测任务，为被评测模型设定角色/行为规范，通常可留空，注意勿与评分器 Prompt 混淆。 -- **数据来源**：评测数据集（含 Prompt+Completion，产生推理费用）或推理结果集（已含 Output，不产生推理费用）。 -- **推理参数**：Temperature、TopP 等，按所选模型动态加载。 -- **标签类型**（应用评测）：分类、布尔值、数字、文本，用于标注评测与观测数据。 - -## 计费说明 - -费用主要由**被评测模型推理费用**与**裁判模型评分费用**两部分构成： - -- 使用推理结果集可免去推理费用。 -- 规则评估、人工评估无裁判模型费用。 -- 已部署的调优模型评测不额外计费（推理费用包含在部署算力费用中）。 - -成本优化建议：先用 50-100 条小规模验证 → 保存推理结果集复用 → 确定性场景优先用规则评估。 - -## 限制与注意事项 - -- 模型评测当前仅支持文本生成类模型，且仅支持控制台操作，不提供公开 API/SDK。 -- 基线评测仅北京地域可用，不支持下载评测结果。 -- 数据管理与数据清洗/增强能力仅适用于华北2（北京）地域。 -- 任务提交后不可更换目标模型、维度类型创建后不可修改，选错需删除重建。 -- 应用自动评测仅面向已发布且已配置知识库的应用，须开通应用观测并具备相应权限。 -- LLM 评分器存在位置偏差与自我偏好偏差，1-3% 的分差通常为噪声，建议定期人工抽查校准。 - -## 关联主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning-and-deployment.md b/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning-and-deployment.md deleted file mode 100644 index 397e4bd9..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning-and-deployment.md +++ /dev/null @@ -1,92 +0,0 @@ -# 模型调优与部署 - -模型调优与部署是百炼平台“模型生产”链路的核心环节：先通过微调训练把领域知识、任务能力或人类偏好写入模型参数，再将调优产物发布为独立、资源专享的在线推理服务。完整链路为 **模型调优 → 模型压缩（可选）→ 模型部署**。 - -> **注意**：微调、压缩、部署与调用等能力**仅在华北2（北京）地域可用**，且必须使用该地域的 API Key；子账号（RAM 用户）需预先被授予调用、训练和部署权限。 - -## 在百炼平台的使用场景 - -- **深度定制模型**：当 Prompt 工程、插件调用仍无法满足效果时，通过微调把领域知识与任务能力固化到模型中，覆盖文本生成、视觉理解（Qwen-VL）、图像/视频生成（万相）、语音合成（CosyVoice）等多种模态。 -- **降低部署成本**：对全精度微调模型做量化压缩，在保持能力的前提下降低部署所需的 MU 规格。以 qwen3.5-flash 微调模型为例，压缩后部署成本可节省约 56%（MU1*2 → MU8*1）。 -- **上线专属推理服务**：将调优或导入（LoRA）的模型部署为独立服务，满足高并发、低延迟的生产需求。 -- **数据准备前置**：调优前通过数据管理功能创建、清洗、增强训练集与评测集（详见相关数据管理主题）。 - -## 模型调优 - -推荐按递进顺序组合使用文本生成的三种训练方式：`CPT（可选）→ SFT → DPO（可选）`。 - -| 方式 | 目标 | 最低数据量 | 数据形态 | -| --- | --- | --- | --- | -| CPT（持续预训练） | 补领域知识 | 1000 万+ Token | 无标签领域文本 `{"text":"..."}` | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条 | ChatML「问-答」对 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组 | `chosen`/`rejected` 回答对 | - -训练模式分**全参训练**与**高效训练（LoRA）**，两者费用相同。官方建议在模型支持全参训练时优先选择全参（效果更好、性价比更高）；LoRA 适合训练时间/成本敏感或数据集较小的场景。不同模型支持的训练方式差异明显：Qwen3-32B、Qwen2.5 系列等支持全部 5 种，而部分新模型（如 Qwen3.5-Plus/Flash）往往仅支持 `sft`。视觉理解仅支持 SFT，万相图像/视频与 CosyVoice 仅支持高效微调（LoRA / `efficient_sft`）。 - -### 关键超参数（文本生成） - -- `learning_rate`：高效训练建议 `1e-4` 量级，全参/CPT 建议 `1e-5` 量级（控制台默认值与 API 示例取值不一致，请以实际训练方式对应的量级为准）。 -- `n_epochs`：默认 `3`，范围 `[1, 200]`；数据量 <10000 建议 3~5 次，>10000 建议 1~2 次。 -- `batch_size`：一般 16/32。 -- `max_length`：建议设为模型支持的最大值；SFT 会**丢弃**超长数据，DPO 则**截断**后仍训练。 -- `lora_rank` / `lora_alpha` / `lora_dropout`：LoRA 专用，秩越大效果略好但更慢、更易过拟合。 -- 通过 API 创建任务时，`n_epochs`、`batch_size`、`max_length` 因影响计费而**必填**。 - -支持控制台（零代码入门）与 API/命令行（统一“上传数据集 → 创建训练任务 → 查询状态 → 管理产物”流程）两种方式。 - -## 模型压缩（可选） - -百炼的模型压缩特指**量化**，不涉及结构剪枝或知识蒸馏，位于调优与部署之间。仅支持平台微调产出的自定义模型。 - -- **不可逆**：压缩后的模型不支持继续微调，也不支持二次压缩。 -- **量化模板**：模板名中 MU 编号越大，部署规格越小、成本越低，但精度损失可能越大。 -- **校准数据**：可选，建议选择与目标推理场景语义相近的数据集（最多 5 个），用于提升量化精度。 -- 任务状态：PENDING → QUEUING → RUNNING → SUCCEEDED / FAILED / CANCELING → CANCELED；仅 PENDING/RUNNING 可停止，仅终态可删除。 -- 压缩任务本身限时免费，压缩后模型在部署阶段按 MU 规格计费。建议在免费期内尝试多个量化模板并用业务测试集验证后再上线。 - -## 模型部署 - -部署将模型发布为在线推理服务，可选三种**互斥**的计费方式（创建后不可更改，切换需先下线再重新部署）： - -- **预置吞吐（PTU）**：预留资源保障特定 TPM 吞吐，额度内不限速，TPS 通常提升约 1.5~2.0 倍，适合流量可预估的高负载生产。支持长输入（部分模型最高 200K token）与前缀缓存折扣；超额或超长时自动转按量计费，业务不中断。 -- **模型单元（MU）**：按使用时长 × 单元数量计费，资源独占，性能指标可自定义，支持 PD 分离降低首 Token 延迟。 -- **按 Token 使用量**：按调用输入/输出 Token 计量，不使用不计费，仅支持 SFT 高效训练后的自定义模型，主要用于效果验证。 - -### 模型导入（LoRA） - -可通过“我的模型”将本地训练的 LoRA 模型从 OSS 导入，**仅支持 LoRA，不支持全参微调模型**。关键约束： - -- OSS Bucket 需添加 `bailian-datahub-access` 标签（值 `read`），不支持归档存储与根目录文件。 -- 必需文件：`adapter_model.safetensors` 与 `adapter_config.json`。 -- `rank` 必须为 8、16、32、64 之一，且所有 LoRA 层使用相同 rank。 -- 不得修改原始 vocab 或 chat_template；VL 模型必须冻结 VIT。 - -### PTU 关键响应字段 - -- `service_tier`：`ptu-standard` 表示使用 PTU 额度，`default` 或不返回表示按量计费。 -- `provisioned_tokens`：折算后实际消耗的额度 token 数。 -- `cached_tokens`：前缀缓存命中数。 - -## 典型工作流 - -1. 准备并（可选）清洗/增强训练数据集。 -2. 提交微调训练任务，等待完成并管理调优产物。 -3. （可选）对全精度微调模型做量化压缩，降低部署规格。 -4. 将调优或导入的模型部署为在线服务，选择合适的计费方式。 -5. 通过标准 API 端点调用部署后的模型进行推理。 - -## 面向开发者的要点 - -- 全流程锁定华北2（北京）地域，注意 API Key 与子账号权限。 -- API 创建调优任务时，涉及计费的超参必填；部署计费方式一经选定不可改。 -- 压缩不可逆且免费期有限，先验证再上线；导入 LoRA 时严格遵守 rank 与文件约束。 - -## 关联主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model production](../api/model-production.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md b/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md deleted file mode 100644 index 3098445e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md +++ /dev/null @@ -1,89 +0,0 @@ -# 模型微调与生产链路 - -模型微调（Fine-tuning）是在 Prompt 工程、插件调用等手段仍无法满足效果时，把领域知识、任务能力、人类偏好或特定音色/风格直接写入模型参数的深度定制手段。它是百炼平台「数据准备 → 模型调优 → 模型压缩（可选）→ 模型部署」这条模型生产链路的核心环节。 - -> **注意**：微调、压缩、部署与调用能力**仅在华北2（北京）地域可用**，且必须使用该地域的 API Key；子账号（RAM 用户）需预先被授予调用、训练和部署权限。 - -## 完整生产链路 - -一条典型的自定义模型生产链路包含四个阶段，前后衔接： - -1. **数据准备**：创建、清洗、增强训练集与评测集（仅控制台，暂无数据处理 API）。 -2. **模型调优**：指定基础模型、训练集/验证集与超参数，提交微调任务，训练完成后得到自定义模型。 -3. **模型压缩（可选）**：对全精度微调模型做量化，降低部署所需 MU 规格、减少推理成本。 -4. **模型部署**：把微调或导入的模型发布为独立、资源专享的在线推理服务，再通过标准 API 调用。 - -## 场景一：模型调优 - -按模态划分，不同模型支持的训练方式差异明显： - -- **文本生成（千问系列）**：支持 CPT、SFT（全参 `sft` / 高效 `efficient_sft`）、DPO（全参 `dpo_full` / 高效 `dpo_lora`）。是否支持某种方式因模型而异（如 Qwen3-32B、Qwen2.5 系列支持全部 5 种，部分新模型仅支持 `sft`）。 -- **视觉理解（千问 VL）**：支持 SFT 全参与高效训练，不支持 CPT/DPO。 -- **图像/视频生成（万相）**：仅支持 SFT-LoRA 高效微调。 -- **语音合成（CosyVoice）**：仅支持 `efficient_sft`，且当前只能通过 API 发起，控制台暂不支持。 - -文本生成推荐按递进顺序组合：`CPT（可选）→ SFT → DPO（可选）`。 - -| 方式 | 目标 | 数据量 | 数据形态 | -| --- | --- | --- | --- | -| CPT（持续预训练） | 补领域知识 | 1000 万+ Token | 无标签领域文本 `{"text":"..."}` | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条 | ChatML「问-答」对 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组 | 同指令下 `chosen` / `rejected` 回答对 | - -训练模式分**全参训练**与**高效训练（LoRA）**，两者费用相同；官方建议模型支持全参时优先全参（效果更好、性价比更高），LoRA 适合训练时间/成本敏感或数据集较小的场景。 - -## 场景二：模型压缩（量化） - -百炼的模型压缩特指**量化**，不涉及剪枝或蒸馏。它把全精度微调模型转为低精度版本，在保持能力前提下降低部署 MU 规格。以 qwen3.5-flash-2026-02-23 为例，压缩前 MU1*2（108 元/小时）、压缩后 MU8*1（47 元/小时），成本节省约 56%。 - -> **注意**：压缩不可逆，压缩后模型不支持继续微调或二次压缩；仅支持百炼平台微调产出的自定义模型。 - -## 场景三：模型部署 - -部署提供三种互斥的计费方式，创建后无法更改： - -- **预置吞吐（PTU）**：预留资源保障特定 TPM，额度内不限速，TPS 通常提升约 1.5～2.0 倍，适合流量可预估的高负载生产。支持 PTU 长输入（部分模型最高 200K token）与前缀缓存折扣，超额自动转按量计费。 -- **模型单元（MU）**：按时长 × 单元数计费，资源独占，支持部分预置模型与所有调优后模型。 -- **按 Token 使用量**：不使用不计费，仅支持 SFT 高效训练后的自定义模型，主要用于效果验证。 - -此外可通过**我的模型**从 OSS 导入本地训练的 **LoRA** 模型（不支持全参微调模型），rank 须为 8/16/32/64 之一，必需 `adapter_model.safetensors` 与 `adapter_config.json`，且不得修改 vocab 或 chat_template。 - -## 关键参数与配置 - -文本生成调优的常用超参： - -- `learning_rate`：高效训练建议 `1e-4` 量级，全参/CPT 建议 `1e-5` 量级。 -- `n_epochs`：默认 `3`，范围 `[1, 200]`；数据量 <10000 建议 3~5，>10000 建议 1~2。 -- `batch_size`：一般 16/32。 -- `max_length`：建议设为模型最大值；SFT **丢弃**超长数据，DPO **截断**后仍训练。 -- `lora_rank` / `lora_alpha` / `lora_dropout`：LoRA 专用，秩越大效果略好但更慢、更易过拟合。 -- 通过 API 创建任务时，`n_epochs`、`batch_size`、`max_length` 因影响计费而**必填**。 - -> **注意**：不同文档默认学习率取值不一致（控制台面板显示 `3e-4`、API SFT 全参示例为 `1.6e-5`）。请以实际训练方式对应的量级为准，切勿照搬。 - -万相图像/视频、CosyVoice 各有独立超参集（如万相 `max_steps`/`generation_type`，CosyVoice 分 `lm_*` 韵律与 `fm_*` 音色两组网络的参数，8 个子字段全部必填）。 - -## 数据格式要点 - -- **SFT（文本）**：ChatML，每行一个 `{"messages":[...]}`，所有 assistant 输出都会被训练；不支持 OpenAI 的 `name`/`weight`。 -- **SFT 思考模型**：仅训练最后的 assistant 输出，思考内容用 `` 标签包裹并保留前后换行。 -- **DPO**：ChatML 加 `chosen` / `rejected` 字段。 -- **CPT**：纯文本 `{"text":"..."}`。 -- **视觉理解/图生视频**：需按字段规范提供图片、视频路径或帧列表。 - -## 面向开发者的使用方式 - -- **控制台（推荐入门）**：模型调优页面创建任务 → 选训练方式与模型 → 配置训练集/验证集与 Checkpoint → 训练 → 部署 → 评测。 -- **API / 命令行**：统一四步流程——上传数据集（`POST /api/v1/files`）→ 创建调优任务 → 轮询任务状态与训练指标 → 部署为在线服务并调用推理端点。 - -若调优后评测效果不佳，最简单的改进办法是收集更多高质量数据继续训练；压缩免费期内可对同一模型尝试多个量化模板，用业务测试集验证后再上线。 - -## 关联主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model production](../api/model-production.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md b/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md index 93ca5304..b45fa8b6 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md @@ -1,41 +1,47 @@ -# 函数调用（Function Calling） +# 函数调用 -函数调用（Function Calling）是让大模型在推理过程中根据用户输入，自主判断并"调用"外部工具（自定义函数、内置能力或插件）以获取实时信息、执行精确计算或操作外部系统的能力。它是弥补大模型原生局限、构建 Agent 与复杂应用的核心机制。 +函数调用（Function Calling）是百炼平台中模型主动识别用户意图、生成结构化工具请求并协同外部能力完成任务的核心机制。它不是简单的 API 封装，而是模型在理解对话上下文后，自主决策是否需要调用工具、选择哪个工具、构造合法参数，并将执行结果无缝融入后续推理链路的端到端能力。 -## 在百炼平台的使用场景 +## 在百炼平台的不同场景中，这个概念如何使用 -百炼在多个层面暴露了 Function Calling 能力： +函数调用在百炼中并非单一接口功能，而是贯穿多个技术路径的横切能力，具体体现为以下三类实践模式： -- **文本 / 视觉生成模型**：所有 Qwen3 及以上通用文本与视觉理解模型均支持自定义工具调用。以 `qwen3.7-plus` 为代表的旗舰模型工具调用完整、上下文长（1M），适合 AI 编程与 Agent 开发；效果确认后可切到 `qwen3.6-flash` 降本，功能与上下文保持一致。 -- **实时多模态（Omni-Realtime API）**：基于 WebSocket 的实时音视频对话同样支持工具调用。客户端通过 `session.update` 事件在会话中声明工具，模型触发调用后由 `response.function_call_arguments.done` 服务端事件返回调用参数，客户端执行后再用 `conversation.item.create` 事件回传工具结果。 -- **应用构建（智能体 / 工作流）**：新版智能体（Agent 2.0）将知识库、MCP、插件等能力统一抽象为"工具"，由智能体自主规划调用顺序，形成"规划-执行-反思"链路。工作流应用则把工具作为固定节点按编排顺序执行，不由模型主动规划。 -- **插件（Plug-in）与 Assistant API**：插件是工具集合，本质也是工具调用。智能体应用 / Assistant API 中，模型依据工具名称与描述判断是否调用；无需调用时直接生成结果。 +- **模型原生函数调用（推荐）**：通过 DashScope 原生接口（如 `/api/v1/services/aigc/text-generation/generation`）传入 `tools` 数组（含 `tool_id`、`description` 和可选 `parameters` schema），由 `qwen-max`、`qwen-plus`、`qwen-turbo` 等支持模型直接输出 `function_call` 结构。模型返回 `output.choices[0].message.tool_calls`，包含 `id`、`tool_name` 和 `tool_input`，开发者需解析后同步/异步调用对应工具，再将结果以 `tool_response` 形式回传继续对话。 -## 内置工具与自定义工具 +- **意图识别辅助调用**：使用专用意图模型（如 `tongyi-intent-detect-v3`）在 `INTENT_MODE` 下对用户输入做轻量级解析，返回标准化意图标签（如 `"search_news"`）和结构化参数（如 `{"keyword": "AI政策", "time_range": "7d"}`）。该方式延迟低、确定性强，适用于路由分发、规则引擎前置等场景，不依赖大模型生成，但需自行绑定工具逻辑。 -- **自定义工具（Function Calling）**：开发者自行定义工具名称、描述与参数结构，模型据此决定何时调用、如何填参，应用侧执行后将结果回填模型生成最终回复。 -- **内置工具**：联网搜索、代码解释器、网页抓取等由平台预置，无需复杂配置即可开启，是 Function Calling 的开箱即用形态。 +- **智能体/工作流编排调用**：在可视化应用中，插件（Plugin）作为已注册的工具单元被显式添加至智能体或拖入工作流节点。此时函数调用由平台运行时自动触发——模型输出 `tool_use` 指令后，平台根据 `tool_id` 查找已授权插件，注入参数并执行，结果自动注入上下文。此模式屏蔽底层协议细节，适合非代码型开发者快速集成。 -## 关键参数与配置 +> ⚠️ 注意：[OpenAI 兼容接口](openai-compatible-interface.md)（`/v1/chat/completions`）**不原生支持函数调用**；其 `functions` / `function_call` 参数被忽略。若需兼容 OpenAI 客户端，必须自行封装：将工具描述注入 `system` 提示词，解析模型 `content` 中的 JSON-like 调用指令，再手动调度。 -- **模型选型**：推荐具备强工具调用能力的模型（如千问-Max / `qwen3.7-plus` 系列）。 -- **思考模式**：可通过 `enable_thinking` 开启（Responses API 用 `reasoning.effort` 控制），配合工具调用提升规划质量。 -- **ReAct 最大轮次**：智能体中取值 1-50，限制单次会话内工具调用的最大次数，防止无限循环。 -- **实时 API 工具配置**：通过 `session.update` 的 `tools` 字段声明可用工具；工具调用结果需经 `conversation.item.create` 回传。 -- **插件调用**：通过 Assistant API 调用时需正确传递工具 ID（如 `calculator`、`code_interpreter`），且每个智能体应用最多添加 10 个工具。 +## 关键参数和配置 -## 开发建议 +| 参数 | 位置 | 类型 | 说明 | 必填 | +|------|------|------|------|------| +| `tools` | 请求体 `input` 或 `messages` 同级 | array | 工具定义列表，每个元素含 `tool_id`（字符串，如 `"quark_search"`）、`description`（自然语言描述）、`parameters`（JSON Schema，用于约束输入格式） | 是（启用函数调用时） | +| `tool_choice` | 请求体 `parameters` 内 | string 或 object | 控制调用策略：`"auto"`（默认，模型自主决定）、`"none"`（禁用）、`{"type": "function", "function": {"name": "xxx"}}`（强制指定） | 否 | +| `enable_search` / `enable_code_interpreter` | 请求体 `parameters` 内 | boolean | DashScope 原生接口快捷开关（仅限内置工具），等价于预置对应 `tools` | 否（推荐用 `tools` 显式声明） | +| `tool_response` | 后续请求 `messages` 中 | object | 上一轮工具执行结果，格式为 `{"role": "tool", "content": "...", "tool_id": "...", "tool_call_id": "..."}`，必须与前次 `tool_calls[0].id` 匹配 | 是（多轮调用时） | -- 优先用内置工具满足通用需求（搜索、计算、代码执行），减少自定义成本。 -- 自定义工具时，工具名称与描述要清晰准确，直接影响模型是否正确触发调用。 -- 需要精确、可控流程时用工作流把工具固化为节点；需要动态规划时用智能体让模型自主调用。 -- 旧版智能体的自定义插件有 5 秒超时限制，设计工具时注意执行时长。 +- **工具 ID 规范**：必须与插件市场注册的 `tool_id` 完全一致（区分大小写），如 `calculator`、`text_to_image`；自定义插件需确保 `tool_id` 在业务空间内唯一且已授权。 +- **参数校验**：模型生成的 `tool_input` 会依据 `parameters` schema 进行基础校验（如类型、必填字段），但**不执行业务逻辑验证**（如搜索关键词长度、图片尺寸合法性），需在工具侧二次校验。 +- **流式响应注意**：函数调用结果在流式响应中可能分块到达（如 `delta.tool_calls`），需按 `index` 和 `id` 组装完整 `tool_call` 对象，不可仅依赖首块。 + +## 面向开发者，简洁实用 + +- ✅ **首选 DashScope 原生接口**：功能最全、错误反馈明确（如 `invalid_tool_id`）、支持长上下文与私有模型。 +- ✅ **始终显式声明 `tools`**：避免依赖隐式开关（如 `enable_search`），确保行为可预测、可审计。 +- ✅ **验证 `tool_call_id` 回传**：多轮调用中，`tool_response` 的 `tool_call_id` 必须严格匹配模型返回的 `id`，否则平台拒绝处理。 +- ❌ **勿在 [OpenAI 兼容接口](openai-compatible-interface.md)中传 `functions`**：该字段被静默忽略，会导致调用逻辑失效。 +- ❌ **勿跳过 `tool_response` 格式校验**：`content` 字段必须为字符串（即使返回 JSON），且 `role` 必须为 `"tool"`。 +- 🚀 **生产建议**：对高并发场景，使用连接池（Java 设置 `connectionPoolSize`，Python 复用 `requests.Session`）；对耗时工具（如 `quark_search`），结合[异步任务](asynchronous-task.md) + EventBridge 回调，避免阻塞主线程。 ## 关联主题页 -- [omni realtime api](../api/omni-realtime-api.md) -- [model experience](../guides/model-experience.md) -- [llm application](../guides/llm-application.md) +- [qwen api reference](../api/qwen-api-reference.md) +- [more models](../api/more-models.md) - [plug in](../guides/plug-in.md) +- [toolkits and frameworks](../api/toolkits-and-frameworks.md) +- [more about models](../api/more-about-models.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md b/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md deleted file mode 100644 index 3062222e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md +++ /dev/null @@ -1,87 +0,0 @@ -# 知识库 - -知识库是阿里云百炼平台基于 RAG（[检索增强生成](rag.md)）技术构建的私有数据管理能力，用于为大模型补充私有数据与最新信息，使应用能够准确回答特定领域问题。知识库仅支持在中国站华北2（北京）地域开通和使用，提供标准版与旗舰版两种规格，并配套日志监控、API、效果优化与计费体系。 - -## 核心能力 - -知识库对私有数据或文件进行语义检索，可找出语义相同或相近的内容，即使关键词匹配度极低甚至为零。检索结果可作为上下文喂给大模型生成回答，也可仅作为检索能力单独使用。支持挂载到[智能体应用](agent-application.md)、工作流应用，或通过阿里云百炼 SDK、LlamaIndex、Spring AI Alibaba 等框架集成到外部应用。 - -知识库类型分为四类，单一知识库不支持同时选择多个类型： - -- 文档搜索：对非结构化文档做语义检索，细分为基础文档问答、图文并茂回复、视觉理解（富文本文档）、极速问答四种使用场景。 -- 数据查询：对结构化表格数据做检索。 -- 图片问答：基于 multimodal-embedding-v1（1024 维）对图片做检索。 -- 音视频搜索：对音频、视频内容按时间轴结构化检索。 - -## 在百炼平台中的使用场景 - -### [智能体应用](agent-application.md) - -在 Agent 2.0 架构中，知识库作为工具由智能体自主规划调用，与 MCP 等外部工具统一调度；在旧版（Agent 1.0）中，则先检索知识库再决策是否调用其他工具。可在应用配置页文档知识库右侧点击「+ 添加知识库」接入，并设置相似度阈值与权重，还可通过标签限定查询范围以提升准确性。开启「展示回答来源」后，回答会以角标形式展示知识来源与源文件/源网页地址。 - -### 工作流应用 - -将知识库节点拖入画布，配置输入变量（query）、选择固定知识库或动态引入、设置 TopK，用于在预定义流程中精确控制检索环节。 - -### 外部应用集成 - -- SDK：通过阿里云百炼 SDK 调用检索能力，子账号需获取 `AliyunBailianDataFullAccess` 策略并加入[业务空间](workspace.md)。 -- LlamaIndex（Python 3.9+）：将知识库部署在云端，使用默认智能文档切分与官方向量模型，不支持自定义切分与嵌入模型；如需灵活切分应改用本地知识库方案。 -- Spring AI Alibaba（Spring Boot 3.x，JDK 17+）：可调用百炼智能体/工作流应用并检索百炼知识库，应用集成推荐变量名 `DASHSCOPE_API_KEY`，知识库检索推荐 `AI_DASHSCOPE_API_KEY`。 -- 本地 RAG：检索环节在本地执行，生成环节调用通义千问 API，适合需要灵活切分与嵌入模型选择的场景。 - -### API 编排 - -通过百炼 OpenAPI（`bailian/2023-12-29`，ROA 风格）可程序化管理知识库全链路：申请上传租约 → 上传文件 → 添加文件到类目 → 轮询文件解析状态（INIT/PARSING/PARSE_SUCCESS）→ 初始化知识库 → 提交索引任务 → 轮询任务状态直至 COMPLETED。每[业务空间](workspace.md)最多创建 500 个类目，类目类型目前仅支持 `UNSTRUCTURED`。 - -## 关键参数与配置 - -### 规格与并发 - -| 规格 | 最高检索并发 | 平台存储空间 | 价格 | -| --- | --- | --- | --- | -| 标准版 | 1 QPS（固定，不可调） | ≤ 100 GB | 0.03 元/知识库/小时 | -| 旗舰版 | 50–10,000 QPS（可调，对应 1–200 RCU） | ≤ 9,999 GB | 0.2 元/RCU/小时 | - -RCU（Retrieval Compute Unit）是检索并发能力度量单位，1 RCU 约支撑最高 50 QPS，所需 RCU = 向上取整（检索峰值 QPS ÷ 50）。变配按发生时间分段计费，同一知识库 1 个自然日内最多变配 1 次。 - -### 向量与切片 - -- 向量模型：文档搜索、数据查询、音视频搜索类支持 text-embedding-v4、text-embedding-v3（均为 512 维）；图片问答类仅支持 multimodal-embedding-v1（1024 维）。向量维度不支持更改。 -- 文本切片长度上限：单个切片 6,000 [Token](token.md)；编辑切片（UpdateChunk）长度限制为 10–6,000 字符；删除切片（DeleteChunk）单次最多 10 个。 -- 召回文本切片数量：单次查询最多召回 20 个切片。 -- 切片方式：智能切分（保留语义完整性）或按长度切分。**知识库一旦创建，无法再配置 metadata 抽取，也无法更改文档切分 chunk**，需在创建时一次性规划好元数据与切片策略。 - -### 检索参数 - -- 相似度阈值：仅语义相似度高于此阈值的文本切片才会被召回。阈值过高（如 0.60）可能导致无召回结果或丢弃相关切片。 -- 召回片段数（K 值）：取值范围 1–20，调大可提升完整性但增加 [Token](token.md) 消耗；拼装后总长度超出大模型输入限制会被截断，并非越大越好。 -- 初步向量检索 TopK / 初步关键词检索 TopK：默认 50，取值范围 10–100，影响送入排序模型的切片数量与成本。 -- 权重：仅在**同类型知识库之间生效**，用于干预多知识库召回顺序。 - -### 解析方式 - -导入文件时可选择解析方式：电子文档解析（不支持插图与图表）、文档智能解析（提取插图文本与摘要）、大模型文档解析（支持对插图和图表提问，需配合选择模型）、Qwen VL 解析（仅图片，可传入 Prompt）、音视频解析（语音识别、视频帧提取、剧情解析按时间轴结构化对齐）。 - -## 效果优化 - -当出现召回不完整或内容不准确时，建议先建立评测基线（至少 100 组问题，覆盖事实型/比较型/教程型/分析型），再按 RAG 三阶段诊断改进： - -1. 建立索引：优化源文件排版（优先 Markdown、移除水印、避免复杂表格）、统一实体表述、启用多轮对话改写（创建时开启，后续无法补开）。 -2. 检索召回：为文件添加标签过滤、配置元数据做结构化搜索、采用智能切分保留语义完整性、调整相似度阈值与召回片段数。 -3. 生成答案：更换为能力更强的商业模型（如通义千问 Max/Plus/QwQ）、优化提示词模板（限定输出、少样本提示、内容分隔标记且 `${documents}` 只出现一次）。 - -## 日志监控 - -检索日志由日志服务（SLS）承载，首次使用需授权角色 `AliyunServiceRoleForSFMAccessSLS` 并创建 LogStore。每条日志 topic 为 `log_dispatch`，包含 request_id、pipeline_id、workspace_id、latency、response_status_code、response_code 等字段。建议搭建调用量趋势、TopN 知识库排名、业务错误率与 HTTP 5xx 错误率等监控。 - -## 关联主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [frameworks](../api/frameworks.md) -- [llm application](../guides/llm-application.md) -- [application use cases](../guides/application-use-cases.md) -- [data connection overview](../guides/data-connection-overview.md) -- [application component api reference](../api/application-component-api-reference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md b/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md deleted file mode 100644 index f1f996e6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md +++ /dev/null @@ -1,45 +0,0 @@ -# 长上下文 - -长上下文指模型在单次调用中能够接收并处理的输入 token 上限（部分模型还包含输出）。上下文窗口越大，单次请求可携带的对话历史、文档、图像、视频等数据越多，能减少多轮拼接与外部检索的复杂度。 - -## 在百炼平台中的使用场景 - -- **文本生成**：Qwen3.7-max / Qwen3.7-plus / Qwen3.6-flash 以及 deepseek-v4 系列均提供 1M token 上下文窗口，可在单轮内喂入长文档、长对话历史或多文件做总结、问答与代码调试。 -- **视觉理解**：Qwen3.7-plus 等模型支持 1M 上下文，可接收最长 2 小时视频（2GB）或大段图文混合输入，进行视频内容分析与 OCR。 -- **PTU 部署**：预置吞吐部署支持长输入（部分模型最高 200K token），通过阶梯容量系数折算 TPM，超出额度或输入超过模型上限时自动转为按量计费，业务不中断。 -- **对话历史管理**：OpenAI 兼容 Responses 接口由平台自动维护对话历史，无需在请求中手动拼接 messages；其他接口需由调用方维护上下文长度与轮次，避免超出模型上下文窗口。 - -## 关键参数与配置 - -- **上下文长度**：按模型不同，常见档位为 32K / 64K / 128K / 200K / 1M。选型时确认目标模型在对应兼容接口下的实际上下文上限。 -- **PTU 长输入阶梯系数**：超出 32K 的输入按更高系数折算 TPM。以 glm-5.1 为例，`[0, 32K)` 系数 1.0，`[32K, 200K]` 输入系数 1.33 / 输出 1.17。 -- **前缀缓存折扣**：命中缓存的输入 token 按模型对应折扣折算容量（如 glm-5.1 为 0.2，deepseek-v4-pro 为 0.08）。 -- **自动溢出阈值**：千问 128K、DeepSeek 64K 为模型上限阈值，超过即转按量计费，响应头返回 `x-dashscope-ptu-overflow: true`。 -- **图像 token 计算**：视觉模型每张图片 token 数按 `h x w / (32 x 32) + 2` 估算，叠加在上下文总额内。 - -## 容量与额度评估 - -长输入场景下建议先用控制台的 PTU 容量计算器评估额度：根据每分钟请求数（RPM）、平均输入/输出长度、预估缓存命中率推算输入 TPM 和输出 TPM，避免意外转为按量计费。PTU 部署的响应包含以下与额度相关的字段： - -- `service_tier`：`ptu-standard` 表示使用 PTU 额度；`default` 或不返回表示按量计费。 -- `provisioned_tokens`：折算后实际消耗的 PTU 额度（含阶梯系数和缓存折扣）。 -- `cached_tokens`：前缀缓存命中的 token 数。OpenAI Chat 兼容为 `usage.prompt_tokens_details.cached_tokens`；OpenAI Responses 为 `usage.input_tokens_details.cached_tokens`；Anthropic 兼容暂不返回。 - -## 限制与注意事项 - -- **功能完整度**：兼容接口为保证协议一致性，可能不暴露百炼原生全部参数；如需最全采样参数与插件能力，建议改用 DashScope 原生接口。 -- **上下文窗口 ≠ 输出上限**：1M 上下文主要指输入侧，单次输出长度仍受模型自身限制，长输入场景需评估输出截断风险。 -- **历史轮次管理**：迁移到不自动管理历史的接口时，需自行裁剪历史消息，避免累积超出窗口导致请求失败或尾部被截断。 -- **跨接口迁移**：从 OpenAI / Anthropic 迁移时应先确认目标 Qwen 模型在对应兼容接口下是否支持所需上下文长度与参数（如 `temperature`、`tools`、`stream` 等），再决定接口选型。 -- **地域差异**：各地域（北京、新加坡、弗吉尼亚）的 API Key 不互通，长上下文模型的可用性可能与地域相关，部署类操作仅适用于华北2（北京）地域。 - -## 关联主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [more about models](../api/more-about-models.md) -- [model experience](../guides/model-experience.md) -- [model deployment 1](../guides/model-deployment-1.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md b/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md new file mode 100644 index 00000000..0de6103f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md @@ -0,0 +1,43 @@ +# 长期记忆 + +长期记忆是百炼平台提供的结构化、语义化用户记忆持久化能力，用于突破大模型上下文窗口限制，实现跨会话、跨对话的智能信息沉淀与精准召回。它将对话中的关键事实（如用户偏好、习惯、承诺）自动提炼为可检索、可更新的记忆片段，并支持基于 Schema 的结构化用户画像建模。 + +## 在百炼平台的不同场景中如何使用 + +- **智能体（Agent）应用**：通过 `autoCapture`（自动捕获）与 `autoRecall`（自动召回）插件闭环，实现在对话结束时自动提炼记忆、在新会话开始前自动注入相关记忆。适用于个性化推荐、日程提醒、客服历史复现等需上下文延续的场景。 +- **工作流（Workflow）应用**：作为外部状态服务调用，可在关键节点（如“用户确认后”）显式调用 `AddMemory` 写入结果，或在决策前调用 `SearchMemory` 检索历史依据，增强流程确定性与业务合规性。 +- **高代码应用**：直接集成 `agentscope-runtime` 提供的 `AddMemory`、`SearchMemory` 等工具类，结合自定义逻辑控制记忆生命周期（如按业务事件触发写入、按时效策略批量清理），适用于对数据主权和控制粒度要求高的生产系统。 +- **RAG 增强场景**：与知识库检索正交协同——知识库承载静态业务文档，长期记忆承载动态用户专属信息（如“张三上周投诉过物流延迟”），二者可联合注入提示词，实现“公域知识 + 私域上下文”的混合推理。 + +## 关键参数和配置 + +| 参数名 | 类型 | 必填 | 说明 | 典型值 | +|--------|------|------|------|--------| +| `user_id` | string | 是 | 记忆隔离的唯一标识，同一用户必须复用相同 ID；不同 ID 数据完全隔离 | `"usr_abc123"` | +| `messages` / `custom_content` | array / string | 互斥 | `messages`：最多 50 条对话消息（一问一答计 2 条），由平台自动提炼；`custom_content`：≤512 字符纯文本，直写内容 | `[{role:"user",content:"每周三晚上8点健身"}]` | +| `project_id` / `project_ids` | string / list | 否 | 单条操作指定记忆规则（如“健康习惯”）；搜索时传数组实现多规则联合召回 | `["proj_health", "proj_reminder"]` | +| `memory_library_id` | string | 否 | 指定记忆库 ID；不填则使用默认库（每个应用默认绑定一个） | `"lib_default"` | +| `top_k` | integer | 否 | `SearchMemory` 返回最大条数，范围 1–100，默认 10 | `5` | +| `min_score` | double | 否 | 相似度阈值 [0.0, 1.0]，低于此值的结果被过滤，默认 0.3 | `0.45` | +| `meta_data` | object | 否 | 自定义键值对，支持增量更新（`UpdateMemory` 中），用于业务分类、来源标记等 | `{"category": "reminder", "source": "chat"}` | +| `expiration_time` | string | 否 | ISO 8601 时间格式（如 `"2025-12-31T23:59:59Z"`），或预设值 `"7d"`/`"30d"`/`"180d"`/`"never"`；不填则按规则默认有效期（通常 180 天） | `"never"` | + +> ⚠️ 注意：`SearchMemory` 支持语义增强开关（`enable_rewrite`, `enable_judge`, `enable_rerank`），调试阶段建议开启以提升召回质量；生产环境可根据延迟敏感度关闭重排序（`enable_rerank=False`）。 + +## 面向开发者的实用提示 + +- **ID 隔离是底线**：务必确保 `user_id` 在整个用户生命周期内稳定一致（如用业务系统 UID 而非会话 ID），否则记忆无法关联。 +- **写入优先用 `messages`**：相比 `custom_content`，`messages` 能触发更准确的语义提炼（如识别时间、实体、意图），尤其适合对话场景。 +- **检索要设 `min_score`**：默认 0.3 可能召回噪声，建议根据实际效果调至 0.4–0.6，避免低质记忆干扰生成。 +- **更新需全量覆盖**：`UpdateMemory` 当前不支持字段级 patch，需传入完整记忆节点内容 + 新 `meta_data`。 +- **限流需统筹规划**：阿里云账号级总配额 3000 QPM，其中 `AddMemory` ≤120 QPM、`SearchMemory` ≤300 QPM，高并发应用应做好本地缓存或批量合并。 +- **调试推荐控制台**：使用百炼控制台「记忆库」→「记忆检索」标签页，可实时测试 query 改写、多规则混合召回效果，无需写代码。 + +## 关联主题页 + +- [long term memory new](../api/long-term-memory-new.md) +- [memory library overview](../guides/memory-library-overview.md) +- [application support](../guides/application-support.md) +- [llm application](../guides/llm-application.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/mcp-and-tools.md b/skills/bailian-docs-llm-wiki/wiki/concepts/mcp-and-tools.md deleted file mode 100644 index da52b78e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/mcp-and-tools.md +++ /dev/null @@ -1,94 +0,0 @@ -# MCP 与工具扩展 - -MCP 与工具扩展是百炼平台为大模型补齐能力边界的统一机制：通过 MCP（Model Context Protocol）标准协议、插件（Plugin）和 Skill 等形式，让智能体、工作流应用能够调用外部工具、执行代码、处理文件，从而突破大模型在实时信息、精确计算、私有数据访问等方面的原生局限。 - -## 三类扩展能力对比 - -百炼提供多种工具扩展形态，适用场景各有侧重： - -| 扩展形态 | 本质 | 接入方式 | 典型场景 | -| --- | --- | --- | --- | -| MCP 服务 | 大模型与外部工具间的标准协议通道 | MCP 广场开通/部署后接入智能体或工作流 | 地图、搜索、图表、阿里云 OpenAPI 等第三方工具 | -| 插件（Plugin） | 工具集合，一个插件含多个工具（API） | 组件广场添加到智能体，或经 Assistant API 调用 | 代码解释器、计算器、图片生成、搜索等 | -| Skill | 可扩展能力包，封装端到端任务流程 | 上传 ZIP 技能包添加到智能体 | 文件处理、数据分析（如 xlsx 处理） | - -三者可组合使用：新版智能体（Agent 2.0）和 Managed Agents 将知识库、MCP、插件、Skill 统一为工具，由智能体自主规划调用顺序。 - -## MCP：标准协议接入 - -MCP 是 Anthropic 提出的开源标准，百炼基于它提供全周期服务，免去为每个外部工具编写专用接口的成本。 - -### 服务类型 - -- **官方 MCP 服务**：百炼云端部署，开通即用（如 Amap Maps、Sequential Thinking、QuickChart、联网搜索 WebSearch）。 -- **自定义 MCP 服务**：开发者自行部署，三种方式——脚本部署（托管到函数计算 FC，支持 `npx` / `uvx` / `http`）、从 AI 网关导入（RESTful API 升级为 MCP）、从阿里云 OpenAPI 导入（操作 OSS、ECS 等）。 - -### 使用方式 - -- **智能体应用**：模型根据对话内容自主判断是否调用，单个智能体最多同时添加 **5 个** MCP 服务，支持多工具组合与动态非固定顺序调用。 -- **工作流应用**：每个 MCP 节点只能用一个工具，需手动指定输入参数；通常先用大模型节点把自然语言解析为工具所需参数，再接入 MCP 节点。 -- **外部调用**：可集成到 Cherry Studio、Cursor 等第三方客户端，或通过 MCP SDK 编码接入个人项目。 - -### 关键参数与配置 - -- **传输协议**：`type` 字段须与端点路径一致，`"sse"` 对应 GET `/sse`，`"streamableHttp"` 对应 POST `/mcp`，不匹配会触发 405/404 错误。百炼已从旧版 SSE 升级为 **Streamable HTTP**，已开通用户需在 MCP 广场执行"取消开通 → 立即开通"完成升级，SDK 使用 `streamablehttp_client`。 -- **鉴权**：外部调用使用 `Authorization: Bearer `，端点如 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`。 -- **脚本部署配置示例**： - -```json -{ - "mcpServers": { - "memory": { - "command": "npx", - "args": ["-y", "@modelcontextprotocol/server-memory"] - } - } -} -``` - -- **敏感信息**：涉及敏感数据的服务在创建时用 KMS 凭据加密管理。 -- **部署后修改**：仅支持编辑名称和描述；改部署方式、地域、安装方式或配置须先停止再重新部署。 - -### MCP 限制 - -- 不能直连千问 API，必须集成在智能体或工作流中使用。 -- 自定义 MCP 托管在 FC，暂不支持访问本地资源；访问远程资源需配置 IP 白名单或打通 VPC。 -- 私有 npm 仓库不支持；npx/uvx 部署的服务源版本更新后不会自动更新，需手动重新部署。 -- 调用会将返回内容作为上下文传入模型，增加输入 Token 消耗。 - -## 插件：工具集合 - -插件是一个工具集合，分官方、三方、自定义三类。官方插件（如 `code_interpreter`、`calculator`、`text_to_image`、`quark_search`、`generate_qrcode`、`github_search`）无需配置输入输出参数即可直接调用。 - -- **支持模型**：qwen-turbo、qwen-plus、qwen-max、qwen-vl-max、qwen-vl-plus。 -- **调用机制**：智能体应用 / Assistant API 由模型判断是否调用；工作流应用中插件作为节点按编排执行。 -- **调用配额**：每个智能体应用最多添加 **10 个** 工具；通过 API 调用需正确传递工具 ID（如 `calculator`）。 -- **授权**：首次访问需授权服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI`；RAM 子账号需主账号先授予 `ram:CreateServiceLinkedRole` 权限。 - -## Skill：能力包 - -Skill 是智能体的可扩展能力包，让智能体在对话中自动识别并处理特定任务（如文件处理、数据分析），无需额外编码。分官方 Skill（自动更新）和自定义 Skill（上传 ZIP）。 - -- ZIP 包根目录必须含 `SKILL.md`，用 YAML 定义 `name`（唯一，建议小写连字符）和 `description`。 -- `description` 质量直接决定智能体调用准确率，建议包含适用输入类型、支持的操作、触发关键词、不适用场景。 -- ZIP 包大小不超过 **10 MB**；重新上传同名包生成新版本，已添加的智能体自动使用最新版。 - -## Managed Agents 中的工具 - -Managed Agents 智能体托管运行时内置 7 个工具：命令执行 `bash`，文件操作 `read`、`write`、`edit`、`glob`、`grep`、`download_file`，并支持接入 MCP 服务与 Skill。工具在独立云端沙箱执行，事件历史服务端持久化，支持中断与续接，适合多步工具调用、代码执行等长时任务。 - -## 计费要点 - -- **云部署 MCP**：限时免部署费；联网搜索免费额度 2000 次，用尽后 29 元/千次，限流 15 QPS。 -- **自定义部署 MCP**：基础模式无部署费、按调用时长 0.000156 元/秒计费（有冷启动）；极速模式部署费 0.000036 元/秒 + 调用费 0.000156 元/秒，适合高频在线场景。 -- **插件**：多为免费或限时免费（部分需申请开通）。 - -## 关联主题页 - -- [model context protocol](../guides/model-context-protocol.md) -- [plug in](../guides/plug-in.md) -- [managed agents](../guides/managed-agents.md) -- [skill](../guides/skill.md) -- [llm application](../guides/llm-application.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md b/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md deleted file mode 100644 index 0f2719f1..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md +++ /dev/null @@ -1,92 +0,0 @@ -# 模型部署与高速推理 - -模型部署是将百炼平台的预置模型、微调模型或导入模型发布为独立、资源专享的在线推理服务的过程；高速推理则是在部署之上（或独立）为调用提供容量刚性兑付与更高输出速度的一组能力。两者共同解决生产环境对高并发、低延迟和确定性吞吐的需求。 - -## 在百炼平台的使用场景 - -模型生产的完整链路为：**模型调优（Fine-tuning）→ 模型压缩（可选量化）→ 模型部署 → 推理调用**。开发者可通过调优 API 定制专属模型，用模型压缩降低部署规格与成本，再通过部署 API 发布为在线服务，最后调用端点进行推理。 - -围绕"部署 + 推理"，平台提供多种资源与容量形态，按业务诉求选型： - -- **需要私有推理服务、资源独占**：用模型部署（PTU / 模型单元 / 按 Token）创建专属服务。 -- **只需锁定容量、抵御公共限流**：用 **TPM 预留**，为指定模型预留专属吞吐量。 -- **只需更快出字、计费不变**：用**快速模式（Fast mode）**，把输出速度提升到标准 API 的 1.5~2 倍。 -- **需要降低部署成本**：先对微调模型做**模型压缩（量化）**，再部署。 - -## 部署的三种计费方式 - -计费方式在服务创建时选定，创建后不可更改（需下线后重新部署）： - -- **预置吞吐（PTU）**：预留资源保障特定 TPM，额度内不限速，TPS 通常较按 Token 提升约 1.5~2.0 倍。适合流量稳定、需并发/延迟确定性的高负载场景。支持按小时后付费与按天预付费。 -- **模型单元（MU）**：按时长 × 单元数计费，资源独占、性能可自定义，支持 PD 分离（拆分 Prefill/Decode 降低首 Token 延迟）。适合私有微调模型与长时任务。支持按分钟后付费与按月预付费。 -- **按 Token 使用量**：仅对 SFT 高效训练（LoRA）后的自定义模型开放，主要用于调优效果验证。 - -> 预付费无法提前退费，首月内提前退订按单价 1.2 倍计费；PTU 超出购买吞吐或输入超模型上限时，自动切换为按量付费（响应头 `x-dashscope-ptu-overflow:true`）。 - -## PTU 长输入与前缀缓存 - -- **长输入阶梯系数**：部分模型对超过 32K 的输入按更高系数折算 TPM（如 glm-5.1 在 [32K, 200K] 区间输入 1.33 / 输出 1.17），部分模型无阶梯（1.0）。 -- **前缀缓存折扣**：命中缓存的输入 token 按折扣系数消耗额度（glm-5.1 为 0.2，deepseek-v4-pro 低至 0.08），显著降低多轮对话与重复前缀场景成本。 -- **额度识别字段**：`service_tier`（`ptu-standard` 走 PTU 额度）、`provisioned_tokens`（折算后实际消耗）、`cached_tokens`（缓存命中数）。这些字段在 OpenAI Chat 兼容、OpenAI Responses、Anthropic 兼容、DashScope 四种协议下的 JSON 路径不同，需分别读取；Anthropic 兼容格式暂不返回 `cached_tokens`。 - -建议创建或扩容前用控制台**容量计算器**（依据 RPM、平均输入/输出长度、预估缓存命中率）估算所需 TPM，避免额度不足产生意外按量费用。 - -## 模型导入（LoRA） - -部署自训练模型前需从 OSS 导入 LoRA 微调版本，核心要求： - -- 仅支持 **LoRA**，不支持全参微调；必需文件 `adapter_model.safetensors` 与 `adapter_config.json`。 -- **rank** 必须为 8/16/32/64 之一，且同模型各 LoRA 层 rank 一致。 -- 不得新增 token、修改 vocab 或 chat_template，须与开源基础模型完全一致；VL 模型必须冻结 VIT（含 `visual` 权重则无法导入）。 -- **OSS 前提**：Bucket 需添加 `bailian-datahub-access` 标签（值 `read`）、文件须放子目录、不支持归档类存储；首次导入需完成服务关联角色授权。 - -支持基础模型涵盖千问3、千问3-VL、千问2.5、千问2.5-VL 系列。 - -## 使用 API / 命令行部署 - -部署接口统一为 `POST https://dashscope.aliyuncs.com/api/v1/deployments`（仅华北2·北京，需先配置 `DASHSCOPE_API_KEY`），通过 `plan` 字段区分计费方式： - -- **PTU**：`"plan": "ptu"`，配合 `ptu_capacity.input_tpm` / `output_tpm`。 -- **模型单元**：`"plan": "mu"`，配合 `deploy_spec`（如 `MU1`）、`capacity`（副本数）、`enable_thinking`、`max_context_length`、`rpm_limit`、`tpm_limit` 等。 -- **按 Token（LoRA）**：`"plan": "lora"`，`capacity` 必填但无效，扩缩容需在控制台申请。 - -部署自定义模型时 `model_name` 使用**模型 ID**（在"我的模型"页面获取）。查询状态用 `GET /deployments/{id}`。 - -## 模型压缩（量化）降本 - -模型压缩特指量化（不含剪枝/蒸馏），将全精度微调模型转为低精度版本以降低部署所需 MU 规格。例如 qwen3.5-flash 微调模型压缩前 MU1*2（108 元/小时），压缩后 MU8*1（47 元/小时），成本节省约 56%。压缩不可逆，压缩后不支持继续微调或二次压缩，仅华北2（北京）可用。量化模板中 MU 编号越大规格越小、成本越低但精度损失可能越大；校准数据应选择与推理场景语义相近的数据集。 - -## 高速推理能力 - -### TPM 预留：锁定专属容量 - -- **专属模型 code**：创建预留后系统生成专属 `model` code，需将请求中的 `model` 替换为该 code 才命中预留容量。 -- **超额不中断**：超出预留自动降级为公共池按量计费，无需改代码，可在详情页查看超额降级统计。 -- **计费**：按 kTPM（1 kTPM = 1000 Tokens/分钟）预付费，部署成功即计费。 -- **管理**：支持在线扩缩容；退订不可恢复，缩容/退订退费按已用部分 1.5 倍系数结算；到期后 2 小时内可调用，2~14 小时转已停止（可续费），14 小时后删除。 - -### 快速模式（Fast mode）：更高输出速度 - -当前处于 **preview 阶段**，面向 AI 编程助手、Agent 多步推理、实时对话等对输出速度敏感的场景： - -- **高速输出**：TPS 达标准 API 的 1.5~2 倍（约 80~100 TPS）。 -- **计费不变**：仍按输入/输出 token 计费。 -- **特殊限流**：超出 TPM 不立即限流，请求进入排队队列（区别于 TPM 预留的"超额降级按量"）。 -- **使用方式**：将 `model` 指定为支持快速模式的 model ID（如 `glm-5.2-fast-preview`），域名格式为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。`glm-5.2` 默认返回 `reasoning_content`，[流式输出](streaming.md)时思考与回答分别通过 `delta.reasoning_content` 与 `delta.content` 推送。 - -## 选型建议 - -- 追求成本优化且流量波动大：**按量付费**。 -- 流量可预估、不能接受公共限流：**TPM 预留**。 -- 高吞吐 + 高性能确定性：**PTU 专属部署**。 -- 只想更快出字、代码改动最小：**快速模式**。 -- 私有微调模型且要降本：先**模型压缩**再按 **MU** 部署。 - -## 关联主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model production](../api/model-production.md) -- [model compression](../guides/model-compression.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md b/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md deleted file mode 100644 index 22db84ab..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md +++ /dev/null @@ -1,83 +0,0 @@ -# 模型选型 - -模型选型是指在百炼平台提供的自研千问与第三方大模型矩阵中，按业务场景、能力档位、成本与延迟需求匹配最合适的模型，并选定对应地域、接入域名与调用方式的过程。选型结果直接决定效果、吞吐、合规性与单位成本。 - -## 选型总体思路 - -1. **先定场景**：明确任务是文本生成、视觉理解、视频/图像/3D 生成、语音、向量检索，还是 RAG、调优、深度研究等专用领域。 -2. **再定能力档**：在同类模型中按「高能力档 / 平衡档 / 轻量低成本档」对位选择，效果稳定后再用低档降本。 -3. **最后定接入**：根据合规与并发要求选择地域、服务部署范围与接入域名，并确认调用协议（OpenAI 兼容 / DashScope SDK / WebSocket / 异步任务）。 - -## 文本生成场景选型 - -对话、抽取、改写、Agent 等文本任务推荐从 `qwen3.7-plus` 入手——能力、速度与成本均衡，支持 1M 上下文、完整 Function Calling 与内置工具；需要最强推理选 `qwen3.7-max`；快速响应的简单任务用 `qwen3.6-flash` 降本。 - -从闭源或第三方模型迁移时按能力档对位： - -| 能力档 | 千问 | 第三方对位 | -| --- | --- | --- | -| 高能力档 | `qwen3.7-max` | — | -| 平衡档 | `qwen3.7-plus` | `deepseek-v4-pro` / `glm-5.2` | -| 轻量低成本档 | `qwen3.6-flash` | `deepseek-v4-flash` / `MiniMax-M2.5` | - -> 注意：第三方模型（`deepseek-v4-pro`、`kimi-k2.7-code`、`MiniMax-M2.5`、`mimo-v2.5-pro` 等）通常不支持内置工具与结构化输出，迁移时需评估能力差异。 - -## [多模态](multimodal.md)场景选型 - -| 模态 | 推荐入口 | 备选 / 降本 | -| --- | --- | --- | -| 视觉理解 / OCR | `qwen3.7-plus`（1M 上下文、2h 视频） | `qwen3.6-flash` 降本；`qwen-vl-ocr` 专做文档/表格/手写提取 | -| 图像生成 | `wan2.7-image-pro` / `qwen-image-2.0-pro` | `z-image-turbo` | -| 视频生成 | `happyhorse-1.1-t2v`（有声 1080P） | `wan2.7-t2v/i2v/r2v` 系列（自定义音频、首尾帧、角色一致性） | -| 3D 生成 | `Tripo/Tripo-H3.1` / `Tripo/Tripo-P1.0` | 仅北京地域，异步任务 | -| 语音合成（TTS） | `cosyvoice-v3.5-plus` | `qwen3-tts-instruct-flash` | -| 语音识别（ASR） | `fun-asr-realtime` | `qwen3.5-omni-plus-realtime` | -| 语音转语音 / 全模态 | `qwen3.5-omni-plus-realtime` | `qwen3.5-livetranslate-flash-realtime` | -| 音乐生成 | `fun-music-v1` | 邀测，仅北京 | -| 向量与重排序 | `text-embedding-v4` / `qwen3-rerank` | `qwen3-vl-embedding`（[多模态](multimodal.md)向量） | - -视频、图像、3D、音乐等生成类模型统一走异步任务模式；语音类多走 WebSocket / HTTP；向量与重排序走 HTTP。 - -## 专用领域模型选型 - -除通用对话模型外，百炼提供面向特定任务的专用模型，选型时需关注地域与协议差异： - -- **法律**：`farui-plus`，法律咨询/文书生成，DashScope SDK。 -- **意图理解**：`tongyi-intent-detect-v3`，同时输出意图与[函数调用](function-calling.md)信息。 -- **深度研究**：`qwen-deep-research`，两阶段（反问确认 + 深入研究），**仅 Python DashScope SDK、仅华北2（北京）**，不支持 OpenAI 兼容接口。 -- **翻译**：`qwen-mt-plus`，支持术语干预、翻译记忆、领域提示。 -- **OCR 与结构化抽取**：`qwen3.5-ocr`，可设 `min_pixels` / `max_pixels` 控制图像分辨率与 token 消耗。 -- **界面交互**：`gui-plus-2026-02-26`，通过 `computer_use` 工具操控桌面 GUI，仅北京。 - -## 选型时的关键参数与能力差异 - -- **思考模式**：通过 `enable_thinking` 开启（Responses API 用 `reasoning.effort` 控制深度），Qwen3 及以上支持，多为可按请求切换的混合模式。 -- **Function Calling 与内置工具**：自定义工具全系列通用模型支持；内置工具（联网搜索、代码解释器、网页抓取）仅部分模型支持，如 `qwen3.7-plus`、`qwen3.6-flash`、`glm-5.2`、`mimo-v2.5-pro`。 -- **结构化输出**：非思考模式下返回合法 JSON，用于字段抽取；第三方模型多不支持。 -- **上下文与[多模态](multimodal.md)限制**：视觉模型单张图片最高 1600 万像素，token 计算 `h × w / (32 × 32) + 2`；视频时长上限因模型而异（2 小时 / 2GB 或 1 小时 / 2GB）。 -- **[流式输出](streaming-output.md)**：`stream` 参数控制；`qwen-deep-research` 反问阶段必须设为 `true`。 - -## 地域、部署范围与接入域名 - -地域决定接入点与数据存储位置，服务部署范围决定推理执行位置，接入域名决定并发承载与隔离性。三者共同影响可选模型与稳定性。 - -- **地域**：华北2（北京）、新加坡、美国（弗吉尼亚）、德国（法兰克福）、日本（东京）。各地域模型列表、API Key、接入域名独立，**不能跨地域混用**。 -- **服务部署范围**：无合规需求选「全球」（资源池更大）；有合规需求选特定地理边界（中国内地/美国/欧盟/日本/国际）。北京、新加坡各仅支持一种范围，无需选择。 -- **接入域名**：生产环境推荐专属域名 `{WorkspaceId}.{region}.maas.aliyuncs.com`（99.9% SLA、超时 3600 秒、支持 HTTP/SSE/WebSocket/WebRTC）；共享域名 `dashscope.aliyuncs.com` 兼容存量；试用域名仅用于快速体验。 -- **模型后缀**：美国地域使用带 `-us` 后缀的模型名（如 `qwen-plus-us`）可限定美国境内推理，不带后缀默认全球推理。 - -## 选型实践建议 - -- 先在控制台「模型体验」中迭代 Prompt，效果稳定后再固化到代码；切换模型后应回归测试。 -- 穷尽 Prompt 工程与 RAG 方案后再投入模型调优（SFT/CPT/DPO），性价比最高。 -- 低延迟批量场景可使用批量推理降本；遇 429 限流时用指数退避 + 客户端令牌桶平滑流量。 -- 切换模型前关注上下文长度、[函数调用](function-calling.md)、[流式输出](streaming-output.md)、[计费](billing.md)方式与地域支持的差异。 - -## 关联主题页 - -- [model experience](../guides/model-experience.md) -- [get started with models](../guides/get-started-with-models.md) -- [use cases](../guides/use-cases.md) -- [more models](../api/more-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal-input.md b/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal-input.md new file mode 100644 index 00000000..8e13a43c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal-input.md @@ -0,0 +1,57 @@ +# 多模态输入 + +多模态输入是指在一次请求中同时提交两种或以上类型的数据（如文本、图像、音频、视频、3D 图像等），使模型能够联合理解并处理跨模态语义信息。百炼平台通过统一的 API 设计与模型能力支持，将多模态输入作为核心交互范式，覆盖智能体调用、图像/视频/3D 生成、实时音视频对话等全场景。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +- **智能体（Agent）与工作流（Workflow）调用**： + 通过 `input` 字段传入消息数组，支持混合 `text`、`image`、`file` 类型内容。图像输入需选用通义千问 VL 系列模型，并在智能体中配置为“自定义处理”，或在工作流模型节点入参中显式填写 `imageList`；文件输入仅限智能体应用，且需在应用内启用“全文引用”或“切片检索”。 + +- **图像生成（Image Generation）**： + 支持 `prompt`（文本）+ `input.messages[].content[].image`（图像）联合输入，用于图生图、局部编辑、风格迁移等任务。`qwen-image-3.0-pro` 等新模型推荐使用结构化 `messages` 格式，最多支持 14 张参考图（Vidu 模型）。 + +- **视频生成（Video Generation）**： + 文生视频、图生视频、参考生视频均依赖多模态输入组合：例如 `input.prompt` + `input.image`（首帧）+ `input.audio`（可选配音）；数字人驱动类模型（如 `liveportrait`、`emo`）还需配合检测模型验证输入合规性。 + +- **3D 生成（3D Generation）**： + 支持三类互斥输入模式：纯文本（`input.prompt`）、单张图像（`input.image`）、四视角图像数组（`input.images`，固定顺序为「前、左、后、右」）。所有输入均参与三维几何与纹理联合建模。 + +- **Omni 实时 API（WebSocket）**： + 原生支持语音（PCM 音频流）、文本、图像（JPG/JPEG，Base64 编码 ≤256 KB）、视频帧（通过 `append_video` 事件）的实时混合输入，服务端按事件流动态融合处理，实现低延迟多模态交互。 + +- **视觉理解与大模型推理（Qwen-VL / Qwen3.7-plus）**： + 支持文本 + 多图（≤2048 张）+ 多视频（≤64 个，总时长 ≤2 小时）联合输入，适用于复杂文档分析、视频摘要、跨模态检索等任务，输出支持结构化 JSON 与 Function Calling。 + +## 关键参数和配置 + +- **通用输入结构**： + - `input` 字段为 `string`（纯文本）或 `array`（多轮/多模态消息），后者需遵循 `[{"role": "user", "content": [...]}, ...]` 格式； + - `content` 内部为数组，每个元素含 `type`（`text` / `image_url` / `image` / `audio` / `video`）及对应值（如 `{"type": "text", "text": "描述一下这张图"}` 或 `{"type": "image_url", "image_url": {"url": "https://..."}}`）； + - 图像 URL 必须可公开访问，或使用 Base64 编码（`data:image/jpeg;base64,...`）。 + +- **模型级约束**： + - VL 模型（如 `qwen3.7-plus`, `qwen3.5-omni-realtime`）要求图像分辨率 ≤1600万像素，单次请求总图像数 ≤2048； + - 视频输入最大时长 2–3 小时（依模型而定），文件大小 ≤2GB； + - Omni 实时 API 中图像尺寸 ≤1080p，Base64 编码后 ≤256 KB； + - Tripo 3D 输入图像分辨率范围 [20, 6000] 像素，单图 ≤20MB。 + +- **协议与头信息**： + - HTTP 请求需设置 `Content-Type: application/json`； + - [异步任务](asynchronous-task.md)（Tripo、视频生成等）必须携带请求头 `X-DashScope-Async: enable`； + - Omni 实时 API 使用 WebSocket 连接，无需额外头，但需正确发送 `input_audio_buffer.append`、`input_image.append` 等结构化事件。 + +- **注意事项**： + - 多模态输入不支持跨模型混用（如向纯文本模型传图像）； + - 文件类输入（PDF/DOCX 等）仅在智能体应用中生效，且需提前上传至百炼知识库或通过 `input_file` 参数传入； + - 所有图像/视频 URL 有效期需覆盖整个请求生命周期，建议使用 CDN 或临时直传链接。 + +## 关联主题页 + +- [application call](../api/application-call.md) +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) +- [3d generation](../api/3d-generation.md) +- [omni realtime api](../api/omni-realtime-api.md) +- [model experience](../guides/model-experience.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md b/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md deleted file mode 100644 index d238b34c..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md +++ /dev/null @@ -1,45 +0,0 @@ -# 多模态能力 - -多模态能力指模型同时理解或生成文本、图像、音频、视频、3D 等多种模态内容的能力。在百炼平台上，它既体现为「输入多模态」（如图文混合输入、音视频实时对话），也体现为「输出多模态」（如文生图、文生视频、语音合成、3D 资产生成）。 - -## 在百炼平台的主要场景 - -百炼按模态与任务把能力拆分到多个方向，开发者可按需组合： - -- **实时音视频对话**：Qwen-Omni-Realtime 系列通过 WebSocket 提供低延迟的语音输入/输出、图像输入、语音活动检测（VAD）、工具调用与联网搜索，适用于智能客服、语音助手等实时交互场景。 -- **图像生成与编辑**：覆盖文生图、图生图、局部重绘、扩图、背景生成、虚拟模特、AI 试衣、创意海报等，涉及千问 Qwen-Image、万相 Wan/Wanx、Z-Image、可灵 Kling、Vidu 等模型家族。 -- **视频生成与编辑**：聚合万相 Wan、HappyHorse、PixVerse、Vidu、可灵 Kling 及人像驱动模型，支持文生视频、图生视频（首帧/首尾帧/续写）、参考生视频、视频编辑与数字人。 -- **3D 资产生成**：基于 Tripo 模型支持文生 3D、单图生 3D、多图生 3D，产出带 PBR 材质或无贴图的 GLB 模型。 -- **视觉理解与 OCR**：以 Qwen 旗舰多模态模型理解图像与长视频（最长约 2 小时），并提供专优的 OCR/文档提取能力。 -- **语音合成 / 识别 / 语音转语音**：TTS、ASR、声音复刻/设计、S2S 实时对话与同传翻译，以及音乐生成。 - -## 输入与协议 - -- **实时流式（WebSocket）**：延迟最低，适合实时交互。Omni-Realtime 通过 `input_audio_buffer.append`（PCM 音频，Base64）、`input_image_buffer.append`（JPG/JPEG，Base64）等事件送入多模态素材，`session.update` 的 `modalities` 控制输出模态（`["text"]` 或 `["text","audio"]`）。 -- **HTTP 同步**：新一代图像模型（如 `wan2.6-image`、`wan2.7-image`、`z-image-turbo`）支持一次请求返回结果，路径为 `.../aigc/multimodal-generation/generation`，请求体用 `messages` 结构，`content` 内混排 `text` 与 `image`。 -- **HTTP 异步**：图像、视频、3D 生成等耗时任务（约 1-5 分钟）统一采用「创建任务拿 `task_id` → 轮询查询」两步流程，创建时必须携带请求头 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。`task_id` 有效期 24 小时，切勿重复创建，轮询即可。 - -## 关键参数与配置 - -不同模态的请求体大多由 `model`、`input`、`parameters` 三部分组成： - -- **实时对话**（`session.update`）：`modalities`、`voice`（音色，因模型而异）、`input_audio_format` / `output_audio_format`（仅 `pcm`，输入 16kHz、输出 24kHz）、`turn_detection.type`（`server_vad` / `semantic_vad`）及 `threshold`、`silence_duration_ms` 等。 -- **图像生成**：`input.prompt` / `negative_prompt`（或新协议 `messages`），图像编辑用 `images` / `image_url` / `mask_image_url`；`parameters` 含 `size`（如 `1024*1024`、`1K`/`2K`/`4K`）、`n`、`aspect_ratio`、`watermark`、`prompt_extend` 等。 -- **视频生成**：`input.prompt` 描述画面镜头，`input.media` 承载 `first_frame` / `last_frame` / `reference_image` / `video` 等素材；`parameters` 含 `resolution`（`480P`/`720P`/`1080P`）、`duration`、`ratio` 等。 -- **3D 生成**：`prompt`、`image`、`images` 三者互斥；`parameters` 含 `texture_quality`、`geometry_quality`、`pbr`、`texture`。 - -## 注意事项 - -- **地域隔离**：模型、Endpoint URL 与 API Key 必须属于同一地域，华北2（北京）、新加坡、美国（弗吉尼亚）等地域各自独立、不可混用；部分能力（如 Tripo 3D、Fun-Music）仅在特定地域可用。推荐迁移到业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）以获得更好性能与稳定性。 -- **产物有效期**：图像结果 URL 有效期 24 小时，3D 模型下载链接仅 2 小时，务必及时下载。 -- **计费**：图像等仅对成功生成的输出计费，输入与失败任务不计费；具体计费、上下文窗口等实时参数以模型广场为准。 - -## 关联主题页 - -- [omni realtime api](../api/omni-realtime-api.md) -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) -- [model experience](../guides/model-experience.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md index a4fbcd2b..257cc9e5 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md @@ -1,75 +1,62 @@ # OpenAI 兼容接口 -OpenAI 兼容接口是百炼平台提供的一套遵循 OpenAI API 规范的服务入口，让已有 OpenAI 应用只需替换 `api_key`、`base_url` 和 `model` 三项即可迁移到百炼，无需改动业务逻辑，是接入成本最低的调用方式。 - -## 在百炼平台的使用场景 - -OpenAI 兼容接口贯穿百炼的多类使用场景： - -- **文本生成模型调用**：作为四类接口（OpenAI 兼容 Chat Completions、OpenAI 兼容 Responses、Anthropic 兼容 Messages、DashScope 原生）中迁移成本最低的一类，适合已基于 OpenAI SDK 构建的应用平滑迁移。其中 Chat Completions 为最常用入口，支持非流式、流式与工具调用（function call）；Responses 为其演进版本，内置联网搜索、代码解释器、网页抓取等工具，并通过 `previous_response_id` 自动管理多轮上下文。 -- **接入第三方客户端与开发工具**：Cherry Studio、Chatbox、Cursor、Cline、Dify 等聊天客户端和编程工具，统一通过「Base URL + API Key + 模型 ID」以 OpenAI 兼容协议接入百炼网关。 -- **专用模型调用**：`tongyi-intent-detect-v3`（意图理解）、`qwen-mt-plus`（翻译）、`qwen3.5-ocr`（OCR）、`gui-plus`（界面交互）等专用模型多数支持 OpenAI 兼容接口调用（注意 `qwen-deep-research` 仅支持 Python DashScope SDK，Qwen-Audio 仅支持 DashScope 协议）。 -- **智能体与工作流应用调用**：应用可通过 OpenAI 兼容的 Responses API 调用，支持同步/异步、多轮对话与[流式输出](streaming-output.md)。 - -## 关键参数与配置 - -迁移与调用的核心是配置以下三要素： - -- **`api_key`**：使用百炼 API Key。各地域、各计费方案的 API Key **相互独立、不能混用**，建议配置到环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄露。若 Base URL 与 API Key 不配套会返回 401。 -- **`base_url`**：OpenAI SDK 调用统一以 `/compatible-mode/v1`（部分方案为 `/v1`）结尾；HTTP 调用需在其后追加具体资源路径（如 `/chat/completions`、`/responses`、`/embeddings`、`/files`）。各地域 SDK Base URL 示例： - - 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - - 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - - 日本（东京）：`https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` - - 德国（法兰克福）：`https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1` - - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - - 其中 `{WorkspaceId}` 为业务空间 ID，可在控制台业务空间详情页查看。北京、新加坡地域推荐使用业务空间专属域名以获得更好的性能与稳定性。 -- **`model`**：替换为百炼支持的模型名称（如 `qwen-plus`、`qwen3-max` 等）。 - -常用请求参数： - -- `messages`（array，必选）：对话消息列表，每条含 `role`（system/user/assistant）与 `content`。 -- `stream`（bool，可选）：是否[流式输出](streaming-output.md)；流式统计 Token 需配合 `stream_options={"include_usage": True}`。 -- 部分专用模型的特有参数（如 Qwen-MT 的 `translation_options`、Qwen-OCR 的 `min_pixels`/`max_pixels`）在 OpenAI SDK 中通过 `extra_body` 传入。 - -## 调用示例 - -以北京地域业务空间专属域名为例（Python）： - -```python -import os -from openai import OpenAI - -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -) -completion = client.chat.completions.create( - model="qwen-plus", - messages=[ - {"role": "system", "content": "You are a helpful assistant."}, - {"role": "user", "content": "你是谁？"}, - ], -) -print(completion.choices[0].message.content) -``` - -OpenAI Python SDK 要求 Python ≥ 3.8。 - -## 注意事项 - -- **协议路径区分**：OpenAI 协议 Base URL 以 `/compatible-mode/v1`（或 `/v1`）结尾，Anthropic 协议以 `/apps/anthropic` 结尾；部分工具还要求在 Anthropic 端点后追加 `/v1`，以各工具原文为准。 -- **接口能力差异**：DashScope 原生接口参数最全；若依赖联网搜索、代码解释器等内置工具，需使用 Responses 而非普通 Chat Completions。跨接口迁移时需核对参数映射。 -- **旧路径迁移**：Responses、Conversations 接口的旧版路径 `/api/v2/apps/protocols/compatible-mode/v1/...` 即将停止维护，请迁移至新版 `/compatible-mode/v1/...`。 -- **地域约束**：Base URL、API Key 和模型列表均不能跨地域混用；限流按主账号维度合并计算。 +OpenAI 兼容接口是百炼平台提供的一组标准化 REST API，严格遵循 OpenAI 的请求/响应协议（如 `/v1/chat/completions`），使开发者能复用现有 OpenAI SDK、工具链和代码逻辑，零改造接入千问（Qwen）及第三方大模型。该接口不改变 OpenAI 的核心语义（如 `messages` 结构、`stream` 行为、错误码格式），但底层由百炼统一调度与计费。 + +## 在百炼平台的不同场景中如何使用 + +- **快速迁移已有应用**：若项目已使用 `openai` Python SDK 或 LangChain 的 OpenAI LLM 封装，只需替换 `base_url` 和 `api_key`，即可直接调用 `qwen3.7-plus`、`qwen-max` 等模型，无需重写业务逻辑。 +- **构建智能助手（Responses API）**：选用 `/v1/responses` 路径的 OpenAI 兼容变体，可开箱启用联网搜索、代码解释器、网页提取等内置工具，自动维护对话状态，适合客服、Agent 类场景。 +- **多模态与嵌入任务**：Vision 接口（`/v1/chat/completions` + `qwen3-vl-plus`）支持图像 URL/Base64 输入；Embedding 接口（`/v1/embeddings`）兼容 `text-embedding-v1`~`v4`，但多模态 Embedding（如 `qwen3-vl-embedding`）不支持此协议。 +- **批量与异步处理**：通过 `/v1/batch`（同步 Batch Chat）或 `/v1/batch/jobs`（异步 JSONL 批量）复用 OpenAI Batch 代码，仅需切换 `base_url` 即可。 +- **会话管理集成**：配合 `/v1/conversations` 接口，可创建、查询、追加消息，实现跨设备上下文持久化，与 Responses API 协同提升长周期交互体验。 + +> ⚠️ 注意：`qwen-vl`、`qwen-audio`、`qwen-ocr` 等多模态模型，以及 `qwen3-coder-next`（代码专用）等部分模型，**仅支持 DashScope 原生接口，不提供 OpenAI 兼容路径**。 + +## 关键参数和配置 + +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `base_url` | string | 是 | **必须匹配地域与计费方案**：
• 生产推荐：`https://{WorkspaceId}.{region}.maas.aliyuncs.com/compatible-mode/v1`（如北京：`https://abc123.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`）
• 兼容保留：`https://dashscope.aliyuncs.com/v1`（仅限试用/调试）
• [Token](token.md) Plan/Coding Plan 用户需使用专属域名（如 `token-plan.cn-beijing.maas.aliyuncs.com`） | +| `model` | string | 是 | 模型 ID，**严格区分大小写与版本后缀**：
• 文本生成：`qwen3.7-plus`、`qwen-max`、`deepseek-v4-pro`
• Vision：`qwen3-vl-plus`、`qwen-ocr`
• Embedding：`text-embedding-v4`
• 不支持：`qwen-audio`、`qwen-vl`（多模态原生模型）、`qwen-coder-turbo`（仅 Completions 接口） | +| `messages` | array | 是 | 标准 OpenAI 格式：`[{"role": "user", "content": "..." }, ...]`
• **不支持 `system` 角色**（会被忽略），请将系统提示合并到首条 `user` 消息中
• Vision 场景下 `content` 可为数组：`[{ "type": "text", "text": "..." }, { "type": "image_url", "image_url": { "url": "..." } }]` | +| `stream` | boolean | 否 | 默认 `false`；设为 `true` 时返回 SSE 流式响应，每帧含 `delta.content` 字段 | +| `stream_options` | object | 否 | 仅流式启用时有效：
`{"include_usage": true}` → 在流末尾返回 `usage` 统计（`prompt_tokens`, `completion_tokens`） | +| `temperature` / `top_p` | number | 否 | 控制输出随机性，取值范围 `0.0–1.0`；二者互斥使用效果更佳 | + +## 面向开发者：简洁实用指南 + +✅ **推荐做法** +- 使用 `openai` SDK（v1.0+）调用： + ```python + from openai import OpenAI + client = OpenAI( + api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" + ) + response = client.chat.completions.create( + model="qwen3.7-plus", + messages=[{"role": "user", "content": "你好"}], + stream=True + ) + ``` + +✅ **避坑提醒** +- ❌ 不要混用 Key 与 Base URL：[Token](token.md) Plan Key 只能配 [Token](token.md) Plan 域名；按量计费 Key 必须配 `{WorkspaceId}` 域名。 +- ❌ 不要传 `system` 消息——它会被静默丢弃，请改用 `user` + 提示词前置。 +- ❌ 不要对 `qwen-vl` 或 `qwen-audio` 使用 `/v1/chat/completions`——会返回 `404` 或 `400`。 +- ✅ 流式解析时，注意 OpenAI 兼容接口返回 `delta.content`（非 `output.text`），结尾帧含 `finish_reason`。 + +✅ **调试技巧** +- 查看响应头 `X-DashScope-Request-ID`，用于问题定位与工单提报； +- 本地验证可用 `curl` + DashScope 域名（仅限试用 Key），生产务必切至 Workspace 域名； +- 所有 OpenAI 兼容接口均按 `(input_tokens + output_tokens)` 计费，含工具调用返回内容。 ## 关联主题页 - [qwen api reference](../api/qwen-api-reference.md) - [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) - [get started with models](../guides/get-started-with-models.md) -- [more models](../api/more-models.md) -- [application call](../api/application-call.md) +- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) +- [model deployment 1](../guides/model-deployment-1.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md b/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md deleted file mode 100644 index fac9783f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md +++ /dev/null @@ -1,59 +0,0 @@ -# 上下文缓存 - -上下文缓存（又称前缀缓存）是百炼平台提供的一种推理加速与成本优化能力：当多次请求共享相同的输入前缀时，平台会缓存该前缀的中间计算结果，后续命中缓存的部分按折扣价折算额度或计费，从而降低延迟与费用。 - -## 在百炼平台的使用场景 - -### 预置吞吐（PTU）部署中的前缀缓存 - -PTU 部署原生支持长输入与前缀缓存，是上下文缓存的主要应用场景。通过阶梯容量系数和缓存折扣管理额度消耗： - -- **缓存命中折扣**：命中缓存的输入 token 按模型对应折扣折算容量。例如 glm-5.1 折扣为 0.2，deepseek-v4-pro 为 0.08，显著降低 TPM 消耗。 -- **自动转按量计费**：当请求超出 PTU 额度或输入超过模型上下文上限（千问 128K / DeepSeek 64K）时，请求自动转为按量计费，响应头包含 `x-dashscope-ptu-overflow:true`，业务不中断。 - -### 显式缓存调用 - -百炼支持通过 OpenAI 兼容接口或 DashScope SDK 显式触发缓存。开发者可在请求中标识需要缓存的前缀（如系统提示词、[长上下文](long-context.md)文档），平台据此管理缓存命中与失效，适用于 RAG 应用、多轮对话等固定前缀反复出现的场景。 - -## 关键参数与响应字段 - -### 请求侧 - -- 在 PTU 部署下，缓存由平台自动管理，无需额外参数；显式缓存则按接口协议提供缓存控制字段。 -- 建议将稳定不变的内容（系统提示、知识库文档、few-shot 示例）置于请求前缀部分，动态内容置于后段，以提高缓存命中率。 - -### 响应侧（PTU 部署） - -PTU 部署的响应中包含额度相关字段，用于观测缓存命中情况： - -- `service_tier`：`ptu-standard` 表示使用 PTU 额度；`default` 或不返回表示按量计费。 -- `provisioned_tokens`：折算后实际消耗的 PTU 额度（含阶梯系数和缓存折扣）。 -- `cached_tokens`：前缀缓存命中的 token 数。 - -`cached_tokens` 在不同 API 格式下的 JSON 路径有差异： - -| API 格式 | 字段路径 | -| --- | --- | -| OpenAI Chat 兼容 | `usage.prompt_tokens_details.cached_tokens` | -| OpenAI Responses | `usage.input_tokens_details.cached_tokens` | -| Anthropic 兼容 | 暂不返回 `cached_tokens` | - -## 容量评估建议 - -在创建或扩容 PTU 部署时，控制台提供容量计算器，可根据每分钟请求数（RPM）、平均输入/输出长度、预估缓存命中率推荐输入 TPM 和输出 TPM。长输入场景下建议先用计算器评估额度，结合预期缓存命中率规划容量，避免意外转为按量计费。 - -## 注意事项 - -- 缓存命中要求前缀完全一致，任何字符变化（包括空格、换行、字段顺序）都可能导致缓存失效。 -- 缓存有有效期，长时间无请求后缓存会被淘汰，需重新预热。 -- 兼容接口（OpenAI / Anthropic）为保持协议一致，可能不暴露百炼原生的全部缓存控制参数；如需最完整的缓存能力，建议使用 DashScope 原生接口。 - -## 关联主题页 - -- [use cases](../guides/use-cases.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [qwen api reference](../api/qwen-api-reference.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md b/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md deleted file mode 100644 index 6241b301..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md +++ /dev/null @@ -1,74 +0,0 @@ -# 提示词工程 - -提示词工程（Prompt Engineering）是通过设计、组织与优化 Prompt 来引导大模型生成符合预期结果的方法论。在百炼平台中，它贯穿智能体配置、[工作流](workflow.md)节点、模型直调与多模态生成等几乎所有 LLM 场景，并提供模板化管理、自动优化、样例库与反馈优化等成熟能力。 - -## 在百炼中的使用方式 - -### 1. 直接编写 System Prompt - -最基础的形态是在智能体应用的「系统提示词」中定义角色、行为指令与能力边界，支持通过 `/` 引用自定义变量。[工作流](workflow.md)应用的大模型节点同样通过「提示词 + 用户提示词」驱动推理。Prompt 越清晰、具体、无歧义，模型表现越稳定。 - -### 2. Prompt 模板（结构与变量分离） - -将固定结构与动态变量分离，统一管理、复用，是团队协作与版本一致性保障的推荐做法。入口位于控制台「应用开发 > 组件管理 > 提示词」。 - -- **预置模板**：平台提供，覆盖创意文案、办公助理等通用场景，效果稳定、不可修改。 -- **自定义模板**：用户自行设计，支持「自定义创建」（直接粘贴现成 Prompt，可选「优化 Prompt」润色）和「基于 Prompt 工程创建」（选择 ICIO / CRISPE / RASCEF 框架，按字段结构化填写）两种模式。 - -框架选型建议： - -| 框架 | 适用场景 | -| --- | --- | -| ICIO | 简单、明确的任务执行，如数据分析、内容生成、文本摘要 | -| CRISPE | 需要 AI 扮演特定角色的交互，如智能客服、创意写作 | -| RASCEF | 涉及多步骤的复杂业务流程，如项目规划、战略分析 | - -> 注意：Prompt 模板相关功能仅适用于华北2（北京）地域，使用前请确认业务空间所在地域。 - -### 3. Prompt 自动优化 - -当缺乏经验或手动编写耗时，可在「提示词 > 自动优化」页面输入原始 Prompt，由大模型进行结构重组、角色扮演引导、指令增强、安全与边界注入等重写，生成结构更优的新版本。该功能不计费，提交数据不会被存储或用于训练。优化失败常见原因：输入超长超出 Token 限制、触发内容审核、网络或服务临时不可用。 - -### 4. Prompt 样例库（Few-shot 检索） - -针对特定领域专业任务，从预定义的高质量问答对中检索相关样例注入上下文，引导模型生成更准确、风格更一致的回复。适用于智能客服、特定领域问答、格式化内容生成。注意该功能已不再维护，官方推荐迁移到 RAG 表格库。 - -### 5. Prompt 反馈优化 - -基于输入输出样例与评测数据，多轮自动评估、反思、优化 Prompt，涉及推理调用。适合对输出质量有持续提升需求的闭环场景。 - -## 关键参数与配置 - -### 智能体应用 - -- **系统提示词最大长度**：6144 字符（从模板「使用 [prompt](../guides/prompt.md) > 创建应用」时自动填充到此上限）。 -- **模型参数**：`temperature`、最长回复长度、`enable_thinking`（开启思考模式以提升反思效果，仅支持思考模式模型）。 -- **变量引用**：在系统提示词中通过 `/` 嵌入自定义变量，运行时由业务数据填充。 - -### 多模态生成 - -- **文生图**：`prompt`（正向）、`negative_prompt`（反向，描述不希望出现的内容）、`prompt_extend`（V2 专用，大模型智能改写，默认 `true`）。基础公式 `主体 + 场景 + 风格`，进阶公式追加镜头语言、氛围词与细节修饰。 -- **文生视频**：基础公式 `主体 + 场景 + 运动`，进阶公式 `主体描述 + 场景描述 + 运动描述 + 美学控制 + 风格化`；图生视频简化为 `运动 + 运镜`。wan2.7 起单/多镜头由提示词控制，不再使用 `shot_type`。 - -### 模板调用 API - -- **创建模板**：`CreatePromptTemplate`，需先获取 Workspace ID。 -- **获取模板**：`GetPromptTemplate`，传入 `workspaceId` 与 `promptTemplateId`，返回 `content`、`variables` 等字段，在代码中填充变量后调用模型。相比字符串拼接，可实现逻辑与内容分离、集中管理与版本一致。 - -## 设计要点 - -1. **结构化优先**：复杂任务优先采用 ICIO/CRISPE/RASCEF 框架或「背景-目的-风格-语气-受众-输出」六要素，避免笼统指令。 -2. **变量分离**：把不可变的结构与可变的业务数据拆开，通过模板或 `/` 变量注入，便于复用与协作。 -3. **样例引导**：对风格或格式敏感的任务，用 Few-shot 样例或反馈优化建立闭环，持续校准输出。 -4. **正反向结合**：图片与视频生成场景同时使用正向与反向 Prompt 精确控制画面内容。 -5. **迭代验证**：结合在线调试面板与评测能力，对 Prompt 变更做回归验证后再发布。 - -## 关联主题页 - -- [prompt](../guides/prompt.md) -- [start using](../guides/start-using.md) -- [llm application](../guides/llm-application.md) -- [use cases](../guides/use-cases.md) -- [application component api reference](../api/application-component-api-reference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md index f5102324..a2ff8a87 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md @@ -1,62 +1,48 @@ -# 检索增强生成（RAG） +# 检索增强生成 -检索增强生成（Retrieval-Augmented Generation, RAG）是一种在大模型生成回答前，先从外部知识库检索语义相关内容、再将其与用户问题一起送入模型的技术。它为大模型补充私有数据和最新信息，从而显著提升在特定领域问题上的准确性、降低幻觉。 +检索增强生成（Retrieval-Augmented Generation，RAG）是百炼平台的核心能力范式，指在大语言模型生成响应前，先从私有或领域知识源中语义检索相关上下文，并将检索结果与用户查询一并输入模型，从而提升回答的准确性、事实性、专业性与时效性。该机制天然融合了信息检索的精准性与大语言模型的生成能力，是构建可信AI应用的关键技术路径。 -## 在百炼平台的使用场景 +## 在百炼平台的不同场景中，这个概念如何使用 -百炼把 RAG 拆成「建立索引 → 检索召回 → 生成答案」三个阶段，并在不同层面提供了对应能力： +RAG 在百炼中不是单一接口，而是贯穿多个能力层的统一范式，具体体现为以下三种典型集成方式： -- **云端知识库（控制台）**：进入知识库按「填写基础信息 → 配置数据来源 → 设置索引参数」三步建库，创建时选定类型（文档搜索 / 数据查询表格库 / 图片问答 / 音视频搜索，创建后不可更改），随后关联到智能体应用、工作流应用或外部应用。工作流应用中知识库节点须接在开始节点之后、大模型节点之前，并在大模型提示词中引用 `result` 变量。注意：知识库功能仅在中国站**华北2（北京）**地域可用。 -- **知识检索服务**：面向多知识库联合检索（最多 15 个），提供 Query 改写、混合检索（向量+关键词）、Rerank 排序的流水线。 -- **知识问答服务**：在检索基础上由大模型生成自然语言回答，提供**极速模式**（单轮检索+生成）与**多轮智能模式**（Agentic 多轮规划搜索），并支持文件预解析、拒答、防泄漏、多模态回复、引用来源等生成控制。 -- **应用场景接入**：围绕「RAG + 智能体应用」可将问答能力接入网站、企业微信、微信公众号、钉钉等渠道；也支持基于本地知识库构建 RAG 应用（检索在本地执行、生成调用通义千问 API），适合需要灵活切分与自定义嵌入模型的场景。 -- **框架集成**：LlamaIndex（Python）可构建云端知识库与 RAG 应用；Spring AI Alibaba（Java）可集成智能体/工作流应用并检索百炼知识库。 +- **知识库问答（零代码/低代码）**：在控制台创建知识库后，直接绑定至智能体或工作流应用的「文档知识库」节点；系统自动完成查询理解、多库联合检索（支持向量+关键词混合检索）、Rerank 排序与上下文注入，开发者只需配置 `TopK`、相似度阈值、标签过滤等参数，无需编写检索逻辑。 -## HTTP REST 接口 +- **API 直接调用（高可控性）**：通过 `/api/v2/apps/knowledge/chat`（端到端问答）或 `/api/v1/indices/knowledge/search`（纯检索）接口，以 RESTful 方式集成 RAG 能力。问答接口内部封装“查询规划 → 知识检索 → 答案生成”三阶段，支持 SSE [流式输出](streaming-output.md)（含 `planning`/`retrieving`/`generating` 事件），适用于需快速上线的业务系统。 -除控制台外，百炼提供 DashScope 应用网关体系的两个 REST 接口，用 API Key Bearer 鉴权，Base URL 形如 `https://{workspaceId}.cn-beijing.maas.aliyuncs.com`： +- **框架集成（开发友好）**：借助 LlamaIndex 或 Spring AI Alibaba 等 SDK，在应用代码中声明式调用云端知识库。例如 LlamaIndex 中使用 `DashScopeCloudIndex.as_query_engine()`，可直接配置 `similarity_top_k`、`similarity_cutoff` 及 `DashScopeRerank`，复用百炼的向量化、切分与重排能力，避免本地维护向量基础设施。 -| 接口 | 路径 | 说明 | -| --- | --- | --- | -| 知识检索 | `POST /api/v1/indices/knowledge/search` | 跨多个知识库联合语义检索，返回按相关性排序的切片，适合需自定义生成流程的场景 | -| 知识问答 | `POST /api/v2/apps/knowledge/chat` | 基于知识库的智能问答，通过 SSE 流式返回规划、工具调用、生成三个阶段 | +此外，RAG 也深度融入数据连接与评测体系：平台托管型数据连接器（如 PDF/Excel）自动构建成知识库供 RAG 调用；应用评测则支持对 RAG 输出进行相关性、事实一致性等维度的自动化打分，形成“构建→部署→评估→优化”闭环。 -默认用户维度 25 QPS。此外还有 `CreateIndex`、`Retrieve` 等 OpenAPI RPC 接口用于建库流程。 +## 关键参数和配置 -## 关键参数与配置 +RAG 行为主要由以下参数控制，按使用层级归类： -检索效果主要由以下参数决定，可在命中测试、检索服务与问答服务中反复调优： +| 类别 | 参数名 | 作用 | 典型取值 | 说明 | +|------|--------|------|----------|------| +| **检索范围** | `knowledgeIds`（API）
`cloud_index_name`（框架） | 指定参与检索的知识库 | `["kb-xxx", "kb-yyy"]` | 未指定时默认使用应用绑定的所有已发布知识库；多库联合检索上限为 15 个。 | +| **召回控制** | `TopK` / `max_retrieve_count` | Rerank 后最终送入大模型的文本切片数 | `3–10` | 过大会增加 token 开销与幻觉风险；过小易丢失关键信息。工作流节点中即为此配置项。 | +| **精度调节** | `similarity_threshold`（控制台/SDK）
`similarity_cutoff`（LlamaIndex） | 过滤低于该分数的检索结果 | `0.3–0.7` | 值越高越严格（召回少但精准），值越低越宽松（召回多但噪声多）。 | +| **性能与成本** | `vector_top_k` / `keyword_top_k` | 初步向量/关键词检索返回的切片数 | `10–50` | 影响 Rerank 阶段费用（费用 = 总初步召回数 × 平均 token 数 × 单价），需权衡效果与成本。 | +| **生成控制** | `stream`（API）
`incremental_output`（SDK） | 控制响应输出模式 | `true`（推荐） | 启用流式可实现前端增量渲染；`incremental_output=True` 保证每次返回仅新增内容，非全量重传。 | -- **相似度阈值（0.01~1.0）**：仅语义相似度高于阈值的切片会被召回。阈值过高会导致相关切片被全部丢弃（如调至 0.60 可能无召回）。 -- **初步向量检索 TopK / 初步关键词检索 TopK（1~100，默认各 50）**：控制初步召回数量，直接影响送入 Rerank 的 Token 量与成本。 -- **最大召回数量 / 召回片段数（1~20）**：最终提供给大模型的切片数，对总结、列举、比较类复杂问题应适当调大。 -- **权重**：多知识库联合召回时干预排序，仅在**同类型知识库之间**生效。 -- **排序模型（Rerank）**：纯文本可选 `qwen3-rerank` / `qwen3-rerank(hybrid)`，多模态可选 `qwen3-vl-rerank`；`gte-rerank` 将于 2026-05-30 下线，新项目直接选 `qwen3-rerank`。 -- **向量模型（Embedding）**：将文本/图片/视频编码到同一语义空间，供余弦相似度匹配。云端知识库使用官方向量模型；本地方案可改用自部署 GTE 模型。 -- **Meta 信息抽取与标签过滤**：在向量检索前做结构化筛选，精准定位目标文件；元数据只能在创建知识库时配置，创建后无法开启。 +> ⚠️ 注意：所有 RAG 场景均**不支持自定义文档切分逻辑或嵌入模型**——向量构建、切分策略、Rerank 模型均由百炼后台统一调度，开发者仅可通过上述参数调控其行为。 -在框架（LlamaIndex）中对应的参数为：`similarity_top_k`（召回数）、`similarity_cutoff`（最低相似度阈值）、`top_n`（重排后返回数），可通过 `node_postprocessors`（`SimilarityPostprocessor`、`DashScopeRerank`）做后处理。 +## 面向开发者，简洁实用 -## 效果优化建议 - -优化前建议用自动评测建立至少 100 组用例的基线，再针对失败用例（打分 < 4）诊断改进： - -- **检索无效（没找到）**：补充知识、优化源文件排版（推荐转 Markdown、移除水印、避免复杂表格）、统一实体表述、启用多轮对话改写。 -- **召回不相关**：使用标签过滤或元数据做结构化搜索。 -- **切片不完整**：采用「智能切分」（基于语义自适应切分），并人工检查修正异常切片。 -- **重排不佳**：在漏召回与噪声之间平衡相似度阈值与召回片段数。 - -## 监控 - -所有检索调用以日志形式投递到日志服务（SLS），topic 为 `log_dispatch`，含 `request_id`、`pipeline_id`（知识库 ID）、`workspace_id`、`latency`、`response_status_code` 等字段，可用于审计、用量统计与慢查询/错误率监控。SLS 存储与流量单独计费，关闭检索日志开关只停止新投递，历史日志仍保留计费。 +- **快速起步**：优先使用 `/api/v2/apps/knowledge/chat` 接口，传入 `messages`（对话历史）和 `knowledgeIds`，即可获得流式问答响应；无需管理模型、提示词或检索链路。 +- **调试技巧**：开启 SSE 流式响应后，监听 `retrieving` 事件中的 `chunks` 字段，可实时查看被召回的原始文本片段，快速验证知识库覆盖度与检索质量。 +- **性能优化**：若发现响应慢或成本高，首先检查 `vector_top_k` 和 `keyword_top_k` 是否设置过大；其次调高 `similarity_threshold` 减少无效切片进入 Rerank 阶段。 +- **错误排查**：确保知识库状态为 `已发布（Published）`；API Key 与 `workspaceId` 必须归属同一租户；文件上传后需等待解析完成（控制台显示“就绪”状态）方可参与检索。 +- **扩展建议**：如需更细粒度控制（如自定义重排逻辑、混合外部 API 结果），应采用底层 OpenAPI（`CreateIndex`/`Retrieve`）+ 自定义 LLM 编排，而非使用封装好的知识问答接口。 ## 关联主题页 +- [knowledge](../api/knowledge.md) - [knowledge base](../guides/knowledge-base.md) - [frameworks](../api/frameworks.md) -- [application use cases](../guides/application-use-cases.md) -- [vector and sort](../api/vector-and-sort.md) -- [knowledge](../api/knowledge.md) -- [use cases](../guides/use-cases.md) +- [application evaluation](../guides/application-evaluation.md) +- [data connection overview](../guides/data-connection-overview.md) +- [application support](../guides/application-support.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md deleted file mode 100644 index c43445dd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md +++ /dev/null @@ -1,71 +0,0 @@ -# 限流与配额 - -限流与配额是百炼平台对模型推理调用施加的吞吐量与调用频率约束，用以保障公共推理资源的公平使用和整体稳定性；当业务流量超过共享上限或需要刚性容量保障时，可通过专属域名、TPM 预留、PTU 部署或异步通知等手段规避限流影响。 - -## 限流的产生场景 - -百炼模型调用主要在以下维度受到约束： - -- **TPM（Tokens Per Minute）**：每分钟可消耗的 Token 数上限，按主账号 + 业务空间 + 模型维度计算，超出会返回 `429` 限流错误。 -- **RPM（Requests Per Minute）**：每分钟请求数，RPM 越大建议 TPM 同比增大。 -- **QPS 限制**：部分通用接口有独立的 QPS 限制。例如异步任务管理（查询、批量查询、取消）三个接口统一限制为 20 QPS，按主账号 + 子账号维度计算。 -- **公共池共享**：按量付费方案无专属容量，调用进入公共共享池，受公共限流波动影响，业务高峰期可能被限流。 - -监控页面的「错误」指标中专门提供**限流错误次数（429）**，可用于定位限流问题；「性能」指标中的 RPM、TPM 则帮助评估容量是否接近上限。 - -## 不同方案下的限流行为 - -| 方案 | 容量保障 | 超额处理 | 接入改动 | -| --- | --- | --- | --- | -| 按量付费 | 无（共享公共池） | 自动服务，受公共限流 | 无需改动 | -| 资源包 / 节省计划 | 承诺用量折扣（非专属容量） | 超出转按量 | 无需改动 | -| TPM 预留 | 专属容量刚性兑付 | 超出自动降级公共池按量，不中断 | 替换 `model` 参数为专属模型 code | -| PTU（模型部署） | 专属部署实例 | 超出转按量 | 替换 `model` 参数 | - -当业务流量可预估且不能接受限流时，优先选择 TPM 预留；对极致性能与隔离有更高要求时，可考虑 PTU 专属部署。 - -## 接入域名对限流的影响 - -接入域名直接影响并发上限与超时表现： - -- **业务空间专属域名** `https://{WorkspaceId}.{region}.maas.aliyuncs.com`：生产推荐，请求超时 3600 秒、SLA 99.9%，提供更高吞吐与时延隔离。 -- **Dashscope 中心化域名** `https://dashscope.aliyuncs.com`：存量业务兼容，可跨业务空间调用。 -- **试用域名** `https://trial.{region}.maas.aliyuncs.com`：仅限快速验证，限流小，不建议生产。 - -各地域的 API Key、模型列表、接入域名不能跨地域混用；美国（弗吉尼亚）暂不支持业务空间专属域名，需使用 `dashscope-us.aliyuncs.com`。 - -## TPM 预留的关键参数 - -创建 TPM 预留时需指定输入 TPM 与输出 TPM，单位为 kTPM（1 kTPM = 1,000 Tokens/分钟），起步和步长因模型而异。容量计算受以下参数影响： - -- **每分钟请求数（RPM）**：业务高峰期每分钟请求数。 -- **平均输入/输出长度（token）**：输入越长，阶梯系数越大，所需输入 TPM 越高。 -- **预估缓存命中率（%）**：命中率越高，输入容量消耗越慢，所需输入 TPM 越低；仅影响输入 TPM。 - -部分模型支持长输入阶梯系数和缓存折扣（如 glm-5.1 在 [32K, 200K] 区间输入系数 1.33、输出 1.17；deepseek-v4-pro 缓存命中部分按 8% 折算），TPM 容量计算器会自动应用这些参数。 - -## 规避异步任务轮询限流 - -图像/视频生成等耗时模型采用异步机制，频繁轮询结果接口会浪费资源并可能触发 20 QPS 限流。百炼已接入事件总线 EventBridge，任务完成（无论成功或失败）后主动上报 `dashscope:System:AsyncTaskFinish` 事件，可推送到 HTTP 回调 URL 或 RocketMQ 消息队列。通知方案不限流、实时性高，适合高并发、大规模或对实时性要求高的任务。 - -## 免费额度与配额停用 - -免费额度页面提供「免费额度用完即停」开关：开启后免费额度用尽时服务自动停止，返回 `403 AllocationQuota.FreeTierOnly`，避免产生免费额度以外的费用。仅在账户内仍有未消耗的免费额度时才能开启；关闭需等免费额度完全消耗后进行。 - -## 应对限流的实践建议 - -1. 生产环境优先迁移到业务空间专属域名，获得更高吞吐与时延隔离。 -2. 流量可预估且不能接受限流的核心业务，使用 TPM 预留锁定专属容量。 -3. 通过模型监控的限流错误次数（429）与 RPM/TPM 指标评估容量是否接近上限。 -4. 异步任务改用 EventBridge 通知方案，避免轮询触发 20 QPS 限制。 -5. 需要严格控制成本的场景，可开启「免费额度用完即停」避免超额消费。 - -## 关联主题页 - -- [get started with models](../guides/get-started-with-models.md) -- [model high speed inference](../guides/model-high-speed-inference.md) -- [model monitoring](../guides/model-monitoring.md) -- [more about models](../api/more-about-models.md) -- [use cases](../guides/use-cases.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/region.md b/skills/bailian-docs-llm-wiki/wiki/concepts/region.md deleted file mode 100644 index 3006794c..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/region.md +++ /dev/null @@ -1,80 +0,0 @@ -# 地域与可用区 - -地域（Region）是百炼平台在物理地理上划分的部署区域，决定接入点位置、数据存储位置与推理执行位置；可用区是地域内相互隔离的物理设施单元。在百炼中，"地域"主要与**服务部署范围**和**接入域名**共同决定调用路径、数据驻留约束与服务保障等级。 - -## 可用地域 - -百炼当前支持的地域及其 Region ID： - -| 地域 | Region ID | 数据驻留约束 | -| --- | --- | --- | -| 华北2（北京） | `cn-beijing` | 数据不出中国内地 | -| 新加坡 | `ap-southeast-1` | 数据不经过中国内地 | -| 美国（弗吉尼亚） | — | 数据不出美国 | -| 德国（法兰克福） | `eu-central-1` | 数据不出欧盟 | -| 日本（东京） | `ap-northeast-1` | 数据不出日本 | - -若对驻留无限制但追求更大推理资源池，可选择美国/德国/日本的全球部署范围。 - -## 三个关键概念 - -调用前需同时确认**地域**、**服务部署范围**、**接入域名**： - -- **地域**：决定接入点与数据存储位置。 -- **服务部署范围**：决定推理执行位置（数据是否跨域流转）。 -- **接入域名**：影响并发上限、超时、SLA 等服务保障。 - -三者在控制台或 Base URL 中体现，**各地域的接入域名、API Key、模型列表不能跨地域混用**。 - -## 接入域名类型 - -| 域名类型 | 格式 | 适用场景 | 关键指标 | -| --- | --- | --- | --- | -| 专属域名 | `{WorkspaceId}.{region}.maas.aliyuncs.com` | 生产环境 | SLA 99.9%、超时 3600 秒、支持 HTTP/SSE/WebSocket/WebRTC | -| 共享域名 | `dashscope.aliyuncs.com` | 存量兼容 | — | -| 试用域名 | `trial.{region}.maas.aliyuncs.com` | 快速体验 | 不建议生产 | - -从共享域名迁移到专属域名只需替换 Base URL 中的域名部分，业务逻辑代码无需修改。 - -## 各地域调用差异 - -- **华北2（北京）**：默认推荐地域，专属域名 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。知识库、[模型部署](model-deployment.md)、数据管理、Prompt 模板、模型调优等大量能力仅在此地域开放。 -- **新加坡**：专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`。 -- **美国（弗吉尼亚）**：使用带 `-us` 后缀的模型名（如 `qwen-plus-us`）可限定美国境内推理；共享接入域名 `https://dashscope-us.aliyuncs.com/compatible-mode/v1`。 -- **德国（法兰克福）/日本（东京）**：通过业务空间区分全球/欧盟或全球/日本部署范围，调用前需先在业务空间管理页面创建并选择对应业务空间。 - -## 业务空间与 WorkspaceId - -使用北京、新加坡、日本、德国等地域时，需在 Base URL 中填入 WorkspaceId。WorkspaceId 可在控制台「业务空间管理」页面查看。子账号需加入对应业务空间并获得相应策略后方可操作该空间下的资源。 - -## 地域相关能力约束 - -部分能力仅在特定地域可用： - -| 能力 | 可用地域 | -| --- | --- | -| 知识库（标准版/旗舰版） | 仅华北2（北京） | -| [模型部署](model-deployment.md)（资源专享推理） | 仅华北2（北京） | -| 数据管理（训练集/[评测](evaluation.md)集/数据流） | 仅华北2（北京） | -| Prompt 模板 | 仅华北2（北京） | -| `qwen-deep-research` | 仅华北2（北京），且仅支持 Python DashScope SDK | -| `qwen-mt-plus` / `qwen3.5-ocr` | 北京、新加坡、美国（弗吉尼亚） | -| `gui-plus` | 华北2（北京） | - -## 选择建议 - -1. 明确数据驻留要求（是否允许数据出境）。 -2. 确认所需模型/能力是否在目标地域开放。 -3. 生产环境统一使用专属域名，避免使用试用域名。 -4. 同一业务的所有调用、API Key、模型列表保持在同一地域，不要混用。 - -## 关联主题页 - -- [get started with models](../guides/get-started-with-models.md) -- [knowledge base](../guides/knowledge-base.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [prompt](../guides/prompt.md) -- [model data overview](../guides/model-data-overview.md) -- [more models](../api/more-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md deleted file mode 100644 index c08e584a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md +++ /dev/null @@ -1,61 +0,0 @@ -# 重排序 - -重排序（Rerank）是检索流程中的二次精排环节：在向量或关键词召回得到候选文档后，由排序模型结合查询与文档的相关性重新打分并排序，把最相关的切片置顶，再交由后续生成或返回流程使用。 - -## 在百炼平台的使用场景 - -重排序贯穿于平台的检索增强与排序能力两条主线： - -- **知识库 RAG 流水线**：知识检索服务的流程为 Query 改写 → 向量 + 关键词混合检索 → Rerank 排序 → 加权返回。排序模型对召回阶段返回的 TopK 切片做精排，再按相似度阈值过滤与最大召回数量截断，决定最终送入生成模型的上下文。知识问答服务在多轮智能模式下会反复触发该精排环节。 -- **独立文本排序 API**：作为向量与排序模型能力的一部分，开发者可直接调用 rerank 接口，对自定义的候选文档列表按查询相关性排序，用于语义搜索、推荐系统、聚类分类等下游任务，无需绑定知识库。 - -## 模型选型 - -| 模型 | 最大文档数 | 单条最大 Token | 请求最大 Token | 特点 | -| --- | --- | --- | --- | --- | -| qwen3-vl-rerank | 文本 100 / 图片 40 / 视频 4 | 8,000 | 120,000 | 多模态，支持图文视频排序 | -| qwen3-rerank | 500 | 4,000 | - | 100+ 语种，高性能文本排序 | -| gte-rerank-v2 | - | - | 30,000 | 50+ 语种（2026-05-30 下线，建议迁移到 qwen3-rerank） | - -知识库内置的排序模型选项包括 `qwen3-rerank`、`qwen3-rerank(hybrid)` 与 `qwen3-vl-rerank`；多模态知识库只能选 `qwen3-vl-rerank`。 - -## 关键参数 - -独立调用 rerank API 时的核心参数： - -- `model`（必选）：模型名称 -- `query`（必选）：查询内容，qwen3-vl-rerank 支持文本与图片两种查询模态 -- `documents`（必选）：待排序文档列表 -- `top_n`（可选）：返回排序后的前 N 个文档 -- `instruct`（可选）：自定义排序任务说明，可指导模型采用不同排序策略（仅 qwen3-rerank 与 qwen3-vl-rerank） - -知识库检索流程中的相关配置： - -- **初步向量检索 TopK / 关键词检索 TopK**：1–100，默认 50，控制进入精排的候选规模 -- **排序模式**：问答模式按 QA 匹配度排序，相似模式按语义相似度排序，自定义高级模式可自行调参 -- **相似度阈值**：0.01–1.0，用于过滤精排后的低分切片，过高会丢弃全部结果 -- **最大召回数量**：1–20，最终返回的切片数 - -## API 接口 - -不同模型走不同的专属接口，Base URL 均为业务空间 ID 拼接的专属域名： - -- qwen3-rerank：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-api/v1/reranks` -- qwen3-vl-rerank / gte-rerank-v2：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank` - -请求需携带 `Authorization: Bearer `，API Key 在控制台 API Key 页面获取，业务空间 ID 在业务空间管理页面获取。 - -## 调优建议 - -- 召回结果不理想时，先建立至少 100 组评测基线，再按失败类型针对性调整。 -- 对「列举 / 总结 / 比较」类问题适当提高最大召回数量 K（1–20），并优先选按拼装长度避免超长截断。 -- 通过命中测试反复调整相似度阈值与召回片段数，找到精排质量与上下文长度的平衡点。 -- 多知识库召回时，相似度相同的切片优先返回权重高的库，但权重仅在同类知识库之间生效。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md b/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md deleted file mode 100644 index 53a3da52..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md +++ /dev/null @@ -1,54 +0,0 @@ -# 软件开发工具包 - -软件开发工具包（SDK，Software Development Kit）是百炼平台为开发者提供的封装库，屏蔽底层 HTTP 接口与鉴权细节，使开发者能在 Python、Java、Node.js、Go 等语言中通过少量代码即可调用大模型、应用与实时多模态能力。SDK 通常随官方迭代持续升级，建议通过包管理器保持最新版本以获得最新功能与稳定性修复。 - -## 接入路径 - -百炼提供两条接入路径，开发者可按工程栈与迁移成本选择： - -1. **阿里云官方 DashScope SDK**：当前仅覆盖 Python 与 Java，封装了百炼原生协议（含智能体/工作流应用调用、异步任务管理、实时多模态交互等能力），适合需要使用百炼独有特性（如 Qwen-Audio、Omni Realtime）的场景。 -2. **OpenAI 兼容 SDK**：通过 OpenAI 多语言 SDK（Python/Java/Node.js/Go）调用百炼的 OpenAI 兼容接口，只需调整 `api_key`、`base_url`、`model` 三处参数即可复用现有 OpenAI 代码与生态工具，迁移成本最低。 - -## 语言与版本要求 - -| 语言 | DashScope SDK | OpenAI 兼容 SDK | -| --- | --- | --- | -| Python | `pip install -U dashscope`，需 Python ≥ 3.8 | `pip install -U openai` | -| Java | Maven/Gradle 引入 `com.alibaba:dashscope-sdk-java`（建议 ≥ 2.12.0） | 引入 `com.openai:openai-java`，需 Java 8+，推荐 3.5.0+ | -| Node.js | 暂无官方 DashScope SDK，可改用 HTTP/`axios` 或 OpenAI SDK | `npm install --save openai` | -| Go | 暂无官方 DashScope SDK | 需 Go 1.22+，`go get github.com/openai/openai-go/v3` | - -安装失败时可临时切换镜像源：Python 用国内 PyPI 镜像；Node.js 执行 `npm config set registry https://registry.npmmirror.com/`；Go 执行 `go env -w GOPROXY=https://mirrors.aliyun.com/goproxy/,direct`。 - -## 鉴权与配置 - -- **API Key 注入**：所有 SDK 均读取环境变量 `DASHSCOPE_API_KEY`，避免在代码中硬编码密钥。临时 API Key（有效期 1–1800 秒，前缀 `st-`）适用于浏览器、移动 App 等不可信环境。 -- **base_url**：使用业务空间专属域名 `https://{WorkspaceId}..maas.aliyuncs.com/compatible-mode/v1`，新加坡地域从旧域名 `dashscope-intl.aliyuncs.com` 迁移至专属域名可获得更好性能与稳定性；美国（弗吉尼亚）使用固定域名 `dashscope-us.aliyuncs.com`，不带 `{WorkspaceId}`。 -- **子业务空间**：若应用或知识库位于子业务空间，需额外配置业务空间 ID 环境变量（如 Spring AI Alibaba 应用集成用 `WORKSPACE_ID`，知识库检索用 `AI_DASHSCOPE_WORKSPACE_ID`）。各地域 API Key 不互通，切换地域需同步更换 Key。 - -## 典型使用场景 - -- **模型调用**：通过 Chat Completions、Responses、Embeddings 等兼容接口调用 Qwen 系列及三方直供模型；OpenAI SDK 复用现有代码，DashScope SDK 适配百炼独有协议。 -- **应用调用**：DashScope SDK 提供 `Application.call`（Python）/ `Application.call(param)`（Java）封装 `POST /apps/{app_id}/completion`，调用智能体应用与工作流应用方式一致，统一返回 `output.text` 供业务消费。 -- **实时多模态交互**：Qwen-Omni-Realtime 通过 WebSocket 长连接驱动，Python/Java SDK 封装了客户端事件（`session.update`、`input_audio_buffer.append`、`response.create` 等）与服务端事件，支持 VAD 与 Manual 两种交互模式。 -- **框架集成**：LlamaIndex（Python 3.9+）构建云端知识库 RAG 应用；Spring AI Alibaba（Spring Boot 3.x + JDK 17+）通过 `spring-ai-alibaba-starter-dashscope` 集成智能体/工作流应用并检索知识库。 -- **异步任务**：图像/视频生成等耗时模型采用异步机制，SDK 封装创建任务、查询结果、取消任务接口（限流 20 QPS）；建议结合 EventBridge 事件通知避免轮询。 - -## 选用建议 - -- 已有 OpenAI 代码栈或需要多语言覆盖 → 优先 OpenAI 兼容 SDK，仅改三处参数即可迁移。 -- 需要百炼独有能力（Qwen-Audio、Omni Realtime、智能体应用封装、异步任务管理）→ 选用 DashScope SDK（Python/Java）。 -- 仅做一次性脚本或跨语言验证 → 可直接 HTTP 调用，跳过 SDK 安装。 - -> **注意**：使用 OpenAI 兼容接口时，新加坡与北京地域的 API Key 不同，切换地域需同步更换；Qwen-Audio 不支持 OpenAI 兼容协议，必须走 DashScope 协议。 - -## 关联主题页 - -- [preparations](../api/preparations.md) -- [frameworks](../api/frameworks.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [more about models](../api/more-about-models.md) -- [omni realtime api](../api/omni-realtime-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md index 04b27e08..c4830a9b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md @@ -1,36 +1,49 @@ # 流式输出 -流式输出（streaming）是指服务端将模型生成的结果分多次增量返回，而非等全部内容生成完毕后一次性下发。它能显著降低首字延迟、提升实时交互体验，适用于对话、语音等对响应速度敏感的场景。 +流式输出（Streaming Output）是指模型服务在生成响应过程中，将结果分块、增量地通过网络持续返回给客户端，而非等待全部内容生成完毕后一次性返回完整响应。这种方式显著降低端到端延迟，支持实时渲染、中断控制与弱网友好交互，是百炼平台面向语音助手、智能客服、RAG问答等低延迟场景的核心能力。 -## 在百炼平台的使用场景 +## 在百炼平台的不同场景中，这个概念如何使用 -百炼平台在多类接口中都提供了流式输出能力，具体开启方式因协议而异： +- **知识问答（`/api/v2/apps/knowledge/chat`）**：启用 `stream=true` 后，服务端通过 Server-Sent Events（SSE）按阶段推送事件，包括 `planning`（查询规划）、`retrieving`（知识检索）、`generating`（答案生成），每阶段可含多个增量文本块，客户端可逐段渲染并支持中途取消请求。 -- **应用调用（智能体/工作流）**：无论是 OpenAI 兼容的 Responses API 还是 DashScope API，都支持流式输出。以 Responses API 为例，通过在请求中设置 `stream=true` 即可开启，服务端会以增量方式持续返回生成内容。 -- **文本生成模型 API**：OpenAI 兼容 Chat Completions、OpenAI 兼容 Responses、Anthropic 兼容 Messages 以及 DashScope 原生接口均支持流式返回。使用 OpenAI 客户端库时，通常只需替换 base URL 和 API Key，并按对应生态的字段约定开启流式参数。 -- **实时多模态交互（Qwen-Omni-Realtime API）**：基于 WebSocket 协议实现天然的流式交互。服务端通过一系列增量事件持续推送结果，例如 `response.audio.delta`（增量音频输出）、`response.audio_transcript.delta`（增量文本转录）、`conversation.item.input_audio_transcription.delta`（实时语音识别中间结果），最终以 `response.done` 表示本轮响应完成。 +- **Realtime API（Omni / Qwen-Audio 等）**：基于 WebSocket 或 AOQ 协议，以事件流形式推送 `response.text.delta`（文本增量）、`response.audio.delta`（PCM 音频增量），实现毫秒级语音合成与文本同步输出；VAD 触发、工具调用、联网搜索等过程均与流式生成无缝融合。 -## 关键参数与配置 +- **Qwen 文本生成（DashScope 原生接口）**：设置 `stream=true` 时，响应为 SSE 流，每个 `data:` 行包含一个 JSON 对象，字段为 `output.text`（当前增量文本）和 `usage`（已消耗 token 数），适用于长文本生成与前端渐进式展示。 -| 场景 | 参数/事件 | 说明 | -|------|-----------|------| -| Responses API / Chat Completions | `stream` | 布尔值，是否流式输出，默认 `false`；设为 `true` 开启 | -| DashScope API | 流式开关 | 通过 SDK 的流式调用方式或对应参数开启 | -| Omni-Realtime API | `response.*.delta` 系列服务端事件 | WebSocket 连接下增量推送音频、文本转录等结果 | -| Omni-Realtime API | `response.done` | 标记单轮响应生成结束 | -| Qwen3-Omni-Flash-Realtime | `smooth_output` | 控制输出平滑度的可选参数 | +- **应用调用（`/api/v1/apps/{APP_ID}/completion`）**：仅同步调用支持流式，需在工作流应用的结束节点显式开启“流式输出”开关并重新发布；启用后返回 SSE 流，内容结构与 DashScope 文本生成一致，但不支持异步调用模式下的流式。 -## 使用建议 +- **[OpenAI 兼容接口](openai-compatible-interface.md)（Chat Completions）**：虽兼容 OpenAI 协议，但默认 `stream=false`；需显式传入 `"stream": true`，响应格式遵循 OpenAI 标准（`delta.content` 字段），解析逻辑需与 DashScope 原生流区分。 -- 需要即时反馈、逐字/逐段展示的实时交互（如聊天界面、语音助手）优先启用流式输出。 -- 流式模式下需在客户端持续读取并拼接增量片段，直到收到结束标志（HTTP 流的结束或 `response.done` 事件）。 -- 流式与异步调用（`background=true`）面向不同需求：流式关注实时增量返回，异步关注长耗时任务的非阻塞执行，二者不要混淆。 -- 跨接口迁移时需核对各生态对流式参数的字段约定差异，DashScope 参数最全，OpenAI/Anthropic 兼容接口以对应生态约定为准。 +> ⚠️ 注意：所有流式接口均要求客户端正确处理连接中断、重连、事件解析与字符编码（UTF-8），且不可假设事件顺序或跳过中间块——例如知识问答中缺失 `retrieving` 事件可能导致答案缺乏依据。 + +## 关键参数和配置 + +| 参数 | 位置 | 类型 | 默认值 | 说明 | +|------|------|------|--------|------| +| `stream` | 请求体（JSON）或 Query 参数 | `boolean` | 因接口而异：
• Knowledge API：`true`
• Qwen DashScope：`true`
• OpenAI 兼容：`false`
• Application Call：`false` | 全局开关，启用后服务端返回流式响应（SSE 或 WebSocket event stream） | +| `modalities` | `session.update` 事件（Realtime API） | `string[]` | `["text", "audio"]` | 控制输出模态组合；流式输出仅对启用的模态生效（如设为 `["text"]` 则无音频 delta） | +| `incremental_output` | Qwen DashScope `parameters` | `boolean` | `true`（当 `stream=true` 时自动生效） | 低层控制是否启用 token 级增量生成（通常无需手动设置） | +| `enable_search` / `tools` | Realtime 或 Qwen 接口的会话/请求参数 | `boolean` / `array` | 按模型支持而定 | 这些能力本身不改变流式机制，但会扩展流式事件类型（如 `response.tool_calls.delta`） | + +- **协议适配要点**： + - SSE 接口（Knowledge、Qwen DashScope、Application Call）：响应头含 `Content-Type: text/event-stream`，需按行解析 `event:`、`data:`、`id:` 字段； + - WebSocket/AOQ 接口（Realtime API）：无固定 MIME 类型，所有消息为 JSON event object，需监听 `response.text.delta`、`response.audio.delta` 等指定 `type` 字段； + - [OpenAI 兼容接口](openai-compatible-interface.md)：响应格式为 `data: {...}\n\n`，`delta.content` 为字符串增量，注意空 content 表示结束。 + +## 面向开发者，简洁实用 + +- ✅ **必做**：始终设置超时（建议 `timeout=60s`），监听 `close`/`error` 事件并实现重试（指数退避）； +- ✅ **推荐**：前端使用 `ReadableStream`（浏览器）或 `EventSource`（SSE）原生 API；服务端推荐 `aiohttp`（Python）或 `fetch-event-source`（Node.js）等成熟流式客户端库； +- ✅ **调试技巧**：用 `curl -N` 查看原始 SSE 流；Realtime API 可通过 `session.update` 设置 `debug: true` 获取内部 trace 事件； +- ❌ **避免**：将流式响应拼接后统一处理——应边收边渲染，利用 `response.text.delta` 实现打字机效果，`response.audio.delta` 直接喂入 Web Audio API； +- 📌 **计费提示**：流式不影响计费逻辑——仍按实际生成的 `output_tokens` 计费，与是否流式无关。 ## 关联主题页 -- [application call](../api/application-call.md) +- [knowledge](../api/knowledge.md) - [omni realtime api](../api/omni-realtime-api.md) +- [realtime api user guide](../api/realtime-api-user-guide.md) - [qwen api reference](../api/qwen-api-reference.md) +- [application call](../api/application-call.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md deleted file mode 100644 index 82223a7f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md +++ /dev/null @@ -1,47 +0,0 @@ -# 流式输出 - -流式输出（Streaming）是指服务端在生成内容的过程中，以增量分片（delta）的方式持续返回结果，而非等待全部内容生成完毕后一次性返回。它能显著降低首字延迟、改善实时交互体验，广泛用于对话、语音助手等场景。 - -## 在百炼平台的使用场景 - -百炼平台在多类接口中都支持流式输出，核心场景包括： - -- **应用调用（Application Call）**：无论是 OpenAI 兼容的 Responses API 还是 DashScope 原生 API，调用智能体或工作流应用时均支持流式输出，可用于需要边生成边展示的实时交互场景。 -- **文本生成模型 API**：OpenAI 兼容 Chat Completions / Responses、Anthropic 兼容 Messages 以及 DashScope 原生接口均可开启流式返回，适合聊天补全类应用逐字/逐段渲染输出。 -- **实时多模态交互（Omni-Realtime API）**：基于 WebSocket 协议，流式是其原生工作方式。服务端通过一系列增量事件持续推送音频与文本，天然适配低延迟的语音对话场景。 - -## 关键参数与配置 - -### HTTP / SDK 接口 - -- **`stream`**：布尔值，控制是否开启流式输出，默认 `false`。设为 `true` 后，服务端以 SSE（Server-Sent Events）方式逐片返回结果。 - - 适用于 OpenAI 兼容 Responses API、DashScope API 等应用调用与模型调用接口。 -- 使用 SDK 时，开启 `stream=true` 后通过迭代响应对象逐步获取增量内容；部分接口可配合 `stream_options` 等参数控制是否返回用量统计等附加信息（以对应接口文档为准）。 - -### Omni-Realtime（WebSocket） - -实时接口不使用 `stream` 参数，而是以事件流的形式天然流式返回。关键的增量事件包括： - -| 事件 | 含义 | -| --- | --- | -| `response.audio.delta` | 增量音频输出 | -| `response.audio_transcript.delta` | 增量文本转录 | -| `conversation.item.input_audio_transcription.delta` | 实时语音识别中间结果 | -| `response.done` | 本轮响应流结束 | - -此外，可通过 `session.update` 事件中的 `smooth_output`（部分模型支持）等参数调节流式输出的平滑度。 - -## 开发建议 - -- **注意结束标志**：SSE 流需处理到结束标记（如 `[DONE]`）或 `response.done` 事件后再收尾，避免内容截断。 -- **增量拼接**：客户端需将各 delta 分片按序拼接，才能得到完整结果。 -- **错误处理**：流式过程中仍可能收到错误事件，需在读取流的循环中做好异常捕获与连接重试。 -- **跨接口差异**：不同接口的分片结构与字段命名不同（OpenAI/Anthropic 兼容接口以对应生态约定为准，DashScope 参数最全），跨接口迁移时需核对字段映射。 - -## 关联主题页 - -- [application call](../api/application-call.md) -- [qwen api reference](../api/qwen-api-reference.md) -- [omni realtime api](../api/omni-realtime-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token-and-billing.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token-and-billing.md deleted file mode 100644 index 4e432064..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token-and-billing.md +++ /dev/null @@ -1,57 +0,0 @@ -# Token 与计费 - -Token 是百炼平台衡量模型用量与费用的基本单位，绝大多数计费、监控、限流与额度管理都以 Token 为核心口径展开。理解 Token 如何被计量与结算，是准确预估成本、规避意外扣费的前提。 - -## Token 在百炼各场景中的用途 - -- **调用计量**：大语言模型、全模态、向量模型的用量统一按 Token 统计；图像生成按张、视频生成按秒、语音模型按秒/字符/Token（视模型而定）。 -- **费用结算**：模型实时推理、Batch 调用、模型训练、上下文缓存等费用均由 Token 消耗量换算得出。 -- **监控与告警**：模型监控、应用观测均以 Token 量（总量/输入/输出、平均单次请求 Token 量、首 Token 延时）作为关键指标。 -- **额度与套餐抵扣**：新人免费额度、资源包、节省计划、Token Plan 的 Credits 都以 Token 消耗为抵扣依据。 -- **容量规划**：TPM（Tokens Per Minute）预留、快速模式的吞吐能力，均以 Token/分钟为容量单位。 - -## 计费方式与抵扣优先级 - -百炼默认按量后付费（分钟级出账、按月结算），实时调用时按固定优先级自动抵扣，无需手动指定： - -**免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费** - -- **免费额度**仅抵扣模型实时推理，不抵扣 Batch、模型调优、模型部署；每个模型（含不同快照版本）额度独立，通常 100 万 Token，有效期 30~90 天，耗尽后返回 `AllocationQuota.FreeTierOnly`。 -- **资源包**预购具体 Token/张数，抵扣单个模型超出免费额度后的实时推理用量，到期作废。 -- **AI 通用型节省计划**承诺月消费换阶梯折扣（最高 5.3 折），以动态月为周期发放额度，当月未用完清零。 -- **Token Plan / Coding Plan** 专属 Key 不消耗免费额度，直接按套餐 Credits 或调用次数计费。 - -> 注意：账户欠费（可用额度 < 0）时，即使某模型仍有免费额度也无法调用；请提前配置余额预警或消费限额。 - -## Token 计价规则要点 - -- **阶梯计费**：部分模型（如 `qwen3-max`）单价取决于单次请求输入 Token 总量，落入哪档就整体按该档单价结算（K=1,000，M=1,000,000）。 -- **思考 / 非思考模式**：部分千问模型（如 `qwen-plus`）两种模式的输出单价不同。 -- **地域差异**：同一模型在不同地域（北京、弗吉尼亚、新加坡等）单价不同，海外通常更高。 -- **Batch 与上下文缓存**：Batch 输入/输出按实时价 50% 计费；上下文缓存仅对输入 Token 打折（显式缓存创建按标准输入 125%、命中按 10%），两者不能同时生效。 -- **训练与部署**：训练按训练 Token 计费；部署按使用时长结合输入/输出 TPM 单价或模型单元计费。免费额度和节省计划均不抵扣训练与部署费用。 - -## 关键参数与配置 - -- **`max_tokens`**：限制输出 Token 上限，是控制生成长度、防止过度生成、压降成本与延迟的主要开关。 -- **专属模型 code**：TPM 预留创建后需将请求中的 `model` 参数替换为专属 code，才能使用预留容量；超出自动降级为按量付费（响应 Header `x-dashscope-ptu-overflow:true`）。 -- **API Key 与 Base URL 配套**：按量付费、Token Plan（`sk-sp-` 开头）、Coding Plan 三者的 Key 与 Base URL 必须配套使用，混用会导致意外走按量通道或 401/403 鉴权失败。 -- **免费额度用完即停**：开启后额度耗尽即停止响应、不再扣费；开启此安心模式后节省计划将无法继续抵扣，需先关闭。 - -## 用量查询与成本管理 - -- **模型用量 / 模型监控**：控制台按业务空间维度统计，数据延迟约 1 小时，支持分钟/小时/天精度，可查看历史 30 天 Token 消耗并设置费用告警。 -- **单次调用 Token 消耗**：需在模型监控配置中开通审计日志与推理日志，目前仅覆盖华北2（北京）部分模型/快照。 -- **账单定位**：账单以"实例 ID"为出账粒度，格式为 `ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`，可据此定位费用来源。 -- **更早数据**：超过 30 天的统计需前往阿里云"费用与成本"控制台查询、开票。 - -## 关联主题页 - -- [test 1](../guides/test-1.md) -- [token plan guide](../guides/token-plan-guide.md) -- [model monitoring](../guides/model-monitoring.md) -- [application monitoring](../guides/application-monitoring.md) -- [support](../guides/support.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md deleted file mode 100644 index 169b6401..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md +++ /dev/null @@ -1,69 +0,0 @@ -# Token 与计费 - -Token 是百炼平台衡量模型处理文本量的基本单位，也是绝大多数计费、限流和用量统计的计量基础；平台的费用则围绕 Token 消耗，通过按量付费、免费额度、节省计划与资源包等机制综合结算。理解 Token 如何被计量与抵扣，是控制大模型使用成本的前提。 - -## Token 在计费中的角色 - -对大语言模型、全模态模型和向量模型，用量与费用均按 **Token** 计量；图像生成按「张」、视频生成按「秒」、语音模型按「秒/字符/Token」（视模型而定）。因此谈「计费」时，Token 主要针对文本类调用。 - -模型调用按**输入 Token** 和**输出 Token** 分别计费，单价以「每百万 Token」为单位。部分模型采用**阶梯计费**：按单次请求的输入 Token 总量分档定价，落入某一区间后该请求全部 Token 按对应档位结算（例如 `qwen3-max` 华北2·北京划分为 `0 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费** - -- **免费额度**：首次开通时各模型自动发放（通常每模型 100 万 Token），仅抵扣实时推理，不抵扣 Batch、调优、部署；不同模型（含同一模型不同快照版本）额度相互独立。开启「免费额度用完即停」后，额度耗尽会停止响应并返回 `AllocationQuota.FreeTierOnly`（或 403），避免意外扣费。 -- **折扣**：Batch 调用输入/输出单价按实时推理价的 50% 计费；支持上下文缓存的模型仅输入 Token 享折扣，两者不能同时生效。 -- **地域差异**：同一模型在不同地域单价不同，境外地域通常无免费额度。 - -### 2. 模型训练（调优） - -按训练 Token 计费。文本模型公式为 `(训练数据 Token + 混合数据 Token) × 循环次数 × 训练单价`；图像/视频生成模型的训练 Token 总量由 `max_steps`、`max_pixels`、`n_epochs` 等超参决定。免费额度和节省计划**均不抵扣**训练费用。 - -### 3. 模型部署 - -免费额度和节省计划同样**不抵扣**部署费用。三种计费方式围绕 TPM（每分钟 Token 数）： - -- **预置吞吐（PTU）**：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)`。PTU 下超出购买吞吐或输入超上限时自动转按量付费（响应头 `x-dashscope-ptu-overflow:true`）。长输入按阶梯系数折算 TPM 消耗，命中前缀缓存的 Token 按折扣系数消耗额度。 -- **模型单元（MU）**：`费用 = 使用时长 × 模型单元数量 × 模型单元单价`。 -- **按 Token 使用量**：仅对 LoRA 微调后的自定义模型开放，用于调优效果验证。 - -### 4. 订阅套餐（Token Plan / Coding Plan） - -- **Token Plan 团队版**：按 Token 消耗抵扣 Credits，面向团队协作。 -- **Coding Plan**：按模型调用次数计量，面向个人开发。 - -两者均使用 `sk-sp-` 前缀的**专属 API Key**，**不消耗新人免费额度**，且 Base URL 与按量付费端点完全隔离，混用会导致意外扣费或 401/403 鉴权失败。 - -### 5. 监控与用量统计 - -模型监控将 Token 消耗归入「成本」类指标，并提供首 Token 延时、RPM/TPM 等性能指标。应用观测则可查看每次调用的输入/输出/平均 Token 量与平均首 Token 耗时。开通推理日志后可查看单次调用的 Token 消耗，用于排查与审计。 - -## 关键参数与配置 - -- **`max_tokens`**：限制单次生成的最大输出 Token 数，是控制输出成本和防止过度生成的首要手段。 -- **TPM / RPM 限流**：部署时可配置 `tpm_limit`、`rpm_limit`；触发限流后等待时间取决于限流值。 -- **PTU 计费识别字段**：`service_tier`（`ptu-standard` 走 PTU 额度，`default` 或缺失表示按量）、`provisioned_tokens`（折算后实际消耗额度）、`cached_tokens`（缓存命中数，Anthropic 兼容格式暂不返回）。这些字段在 OpenAI Chat、OpenAI Responses、Anthropic 兼容、DashScope 四种协议下 JSON 路径不同，需按协议读取。 - -## 成本优化与出账 - -- **节省 Token**：优化 Prompt 减少输入 Token、简单任务选用轻量级模型、非实时任务走批量推理、合理设置 `max_tokens`。 -- **预付费方案**：AI 通用型节省计划（承诺月消费换阶梯折扣，最高 5.3 折，月额度不可跨月累积）、其他模型节省计划、资源包（预购具体 Token/图片数量，仅抵扣单个模型超免费额度后的实时推理）。 -- **出账时效**：大模型推理分钟级出账（约 2～10 分钟）；批量推理、模型训练等小时级出账。用量统计数据约 1 小时延迟，且不支持查看 30 天前数据。 - -## 关联主题页 - -- [test 1](../guides/test-1.md) -- [token plan guide](../guides/token-plan-guide.md) -- [model monitoring](../guides/model-monitoring.md) -- [application monitoring](../guides/application-monitoring.md) -- [support](../guides/support.md) -- [model deployment 1](../guides/model-deployment-1.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token.md index 0b4e6e3c..8ff8f407 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/token.md @@ -1,65 +1,47 @@ -# Token 与计量计费 +# Token -Token 是大语言模型处理文本的最小计量单位，百炼平台以 Token 为核心，对模型的输入、输出、训练用量进行计量、计费与监控。理解 Token 的产生方式与计费规则，是控制模型调用成本、优化应用性能的基础。 +Token 是百炼平台中用于计量模型输入与输出内容的基本单位，代表模型处理的最小语义片段（如词元、子词或字节对）。所有模型调用的计费、容量限制、性能监控和资源调度均以 Token 为统一计量基础。 -## 什么是 Token +## 在百炼平台的不同场景中，这个概念如何使用 -Token 是模型在处理文本时切分出的基本片段（一个汉字、单词或子词可能对应一个或多个 Token）。在百炼平台中，Token 既是**用量计量单位**，也是**大部分计费的核算基准**： +- **计费核心单元**：模型调用按实际消耗的输入 Token 和输出 Token 分别计费（如 `qwen3.7-plus` 输入 ¥12/百万 Token），免费额度、资源包、节省计划均以 Token 为抵扣单位。 +- **容量控制依据**：TPM 预留（Token Per Minute）和模型部署中的 PTU/MU 规格，均以每分钟可处理的 Token 数（kTPM）为容量度量；RPM（Requests Per Minute）限流也隐式受限于单请求平均 Token 消耗。 +- **性能监控指标**：模型监控与应用监控中，“Token 总量” = 输入 Token + 输出 Token，用于分析成本分布、首 Token 延迟（与输入 Token 量强相关）、吞吐瓶颈（TPS × 平均输出 Token）。 +- **推理能力边界**：模型上下文长度（如 256K Token）、缓存折扣规则（如 glm-5.2 缓存命中部分按 25% 折算输入 Token 容量）、阶梯计价档位（0–32K / 32K–128K Token）均直接依赖 Token 数量。 +- **思考模式控制**：在 OpenCode 或 Qwen Code 中启用 Thinking Mode 时，需显式配置 `budgetTokens`（如 `1024`），该值限制推理过程中用于内部规划的额外 Token 消耗，不计入用户输入 Token。 -- 文本生成模型按**输入 Token** 和**输出 Token** 分别计量，思考模式下的输出 Token 同时包含「思维链 + 回答」两部分。 -- 不同模型类型的计量单位不同：大语言模型 / 全模态模型 / 向量模型按 **Token** 计量，图像生成按**张**，视频生成按**秒**，语音模型按**秒 / 字符 / Token**（视模型而定）。 +## 关键参数和配置 -## 在各场景中的使用 +| 参数 | 说明 | 典型位置 | 示例 | +|------|------|----------|------| +| `input_tokens` / `output_tokens` | API 响应中返回的实际消耗 Token 数 | 所有模型调用响应体（`usage` 字段） | `"usage": {"input_tokens": 128, "output_tokens": 64}` | +| `budgetTokens` | 思考模式下允许使用的最大额外 Token 数 | OpenCode `opencode.json` / Qwen Code `/config` 命令 | `"budgetTokens": 1024` | +| `max_context_length` | 部署时指定模型支持的最大上下文 Token 数 | 模型部署 API 的 `deploy_spec` | `"max_context_length": 100000` | +| `tpm_limit` | 每分钟最大 Token 处理量（输出方向） | MU 部署配置 | `"tpm_limit": 1000` | +| `input_tpm` / `output_tpm` | TPM 预留中预购的输入/输出吞吐量（单位：kTPM） | TPM 预留创建参数 | `"input_tpm": 5000, "output_tpm": 500` | -### 1. 按量付费与免费额度 +> ⚠️ 注意： +> - Token 计数由百炼平台服务端统一计算，开发者无需自行分词；不同模型 tokenizer 实现不同，同一文本在不同模型下 Token 数可能差异显著。 +> - 缓存命中、Batch 调用、长上下文等场景存在 Token 折算系数（如超 32K 输入按 1.33 系数折算），实际计费 Token = 原始 Token × 系数。 +> - 所有 Token 相关限制（如上下文长度、TPM 预留额度）均指 *服务端处理后的有效 Token 数*，不含协议层开销（如 SSE event 字段、JSON 序列化冗余）。 -- 首次开通时，各模型会发放新人专属免费额度（通常各 100 万 Token），仅抵扣**实时推理**费用，且不同模型（含同一模型不同快照版本）额度相互独立、不可合并。 -- 免费额度耗尽后默认转为**按量付费**，按输入 / 输出 Token 计费。部分模型采用**阶梯计费**：按单次请求的输入 Token 总量分档（如 `qwen3-max` 分 0–32K / 32K–128K / 128K–256K 三档），落在哪一档，该请求全部 Token 均按该档单价结算。 -- 同一模型在不同地域（北京、弗吉尼亚、新加坡、法兰克福、东京）单价不同。 +## 面向开发者，简洁实用 -### 2. 订阅制套餐 - -- **Token Plan 团队版**：按 Token 消耗抵扣 Credits，工具调用（联网搜索、代码解释器等内置工具）产生的 Token 同样从套餐 Credits 抵扣，不额外收费。 -- **Coding Plan**：按模型调用**次数**计费（而非 Token），面向个人开发场景。 - -### 3. 模型训练与部署 - -- **训练**按训练 Token 计费，文本模型公式为 `（训练数据 Token + 混合训练数据 Token）× 循环次数 × 训练单价`；图像 / 视频模型的训练 Token 由 `max_steps`、`max_pixels`、`n_epochs` 等超参数推算。 -- **部署**（预置吞吐 TPM）按输入 / 输出 TPM 单价与时长计费，与 Token 用量间接相关。训练与部署**不能**用免费额度或节省计划抵扣。 - -### 4. 监控与观测 - -- 应用观测可查看每次调用的 Token 量，监控统计提供 Token 总量（全部 / 输入 / 输出）、平均单次请求 Token 量、平均首 Token 耗时等指标。 -- 模型监控将 Token 消耗归入「成本」类指标，首 Token 延时归入「性能」类指标，支持按分钟 / 小时 / 天聚合，并可配置告警。 -- 用量统计按业务空间维度归集，数据延迟约 1 小时。 - -## 关键参数与配置 - -| 项 | 说明 | -| --- | --- | -| `max_tokens` | 限制单次生成的最大输出 Token 数，用于控制成本、防止过度生成，也是降低幻觉的手段之一 | -| 上下文缓存 | 命中缓存的**输入** Token 享折扣；缓存折扣与 Batch 折扣不可同时生效，价格表输入单价不含缓存单价 | -| Batch 调用 | 支持的模型输入 / 输出单价按实时推理价的 50% 计费 | -| 免费额度用完即停 | 开启后额度耗尽即停服（返回 403），避免意外扣费，但也会阻断节省计划继续抵扣 | - -## 成本优化建议 - -- **优化 Prompt**：简洁清晰的 Prompt 可减少不必要的输入 Token 消耗。 -- **控制输出长度**：合理设置 `max_tokens`，避免冗长输出。 -- **按任务选模型**：分类、摘要等简单任务优先使用轻量级模型。 -- **批量推理**：非实时大批量任务走 Batch 接口，Token 单价更低。 -- **善用抵扣顺序**：`免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费`，据此规划预付费方案。 - -## 账单中的 Token - -大模型推理为分钟级出账（通常 2～10 分钟），批量推理与训练为小时级。账单「实例 ID」以英文分号分隔，包含 `ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`，可据此区分输入 / 输出 Token 的费用来源与调用渠道（`app` 代码调用、`bmp` 控制台体验、`assistant-api`）。 +- **快速估算**：使用 [Token 计算器](https://bailian.console.aliyun.com/#/token-calculator) 输入文本，实时查看各模型下的 Token 数及预估费用。 +- **调试技巧**：若遇到 `429 Too Many Requests`，检查 `usage.input_tokens` 和 `usage.output_tokens`，确认是否超出 TPM 预留额度或 RPM 限流阈值。 +- **成本优化**： + - 对长文本优先启用前缀缓存（支持模型见文档）； + - Batch 调用可享 50% Token 单价折扣（不可与缓存折扣叠加）； + - 使用 `glm-5.2-fast-preview` 等快速模式模型，在相同 Token 消耗下提升 TPS，摊薄单位 Token 延迟成本。 +- **监控告警**：在模型监控中配置“单次请求输出 Token > 2000”告警，及时发现异常生成行为；结合应用监控的 Token 分布热力图，定位高消耗节点（如 LLM 节点 vs RETRIEVER 节点）。 ## 关联主题页 - [token plan guide](../guides/token-plan-guide.md) -- [test 1](../guides/test-1.md) +- [model high speed inference](../guides/model-high-speed-inference.md) +- [model deployment 1](../guides/model-deployment-1.md) - [application monitoring](../guides/application-monitoring.md) - [model monitoring](../guides/model-monitoring.md) -- [support](../guides/support.md) +- [test 1](../guides/test-1.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/tool-integration.md b/skills/bailian-docs-llm-wiki/wiki/concepts/tool-integration.md new file mode 100644 index 00000000..b75c9232 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/tool-integration.md @@ -0,0 +1,56 @@ +# 工具集成 + +工具集成是百炼平台中将外部能力（如计算、搜索、文件处理、云服务等）以标准化方式接入智能体或工作流的核心机制，使大模型能够安全、可控地调用真实世界的服务，突破其固有的知识时效性、计算精度与执行边界限制。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +工具集成不是单一技术方案，而是覆盖多层抽象、适配多种开发范式的统一能力体系，具体体现为以下三类主流形态，开发者可根据需求选择： + +- **插件（Plug-in）**：面向轻量、通用能力的即插即用集成。适用于实时搜索（`quark_search`）、代码执行（`code_interpreter`）、文生图（`text_to_image`）等高频场景。模型可自主规划调用，也可在工作流中作为显式节点编排。所有插件需通过服务关联角色授权，且仅限同一子业务空间内使用。 + +- **Skill**：面向结构化任务处理的能力封装，强调语义驱动与零代码集成。适用于文件解析（`pdf-parser`）、数据清洗（`csv-cleaner`）等输入/输出明确的业务逻辑。通过 `SKILL.md` 中的 `description` 字段定义触发条件与能力边界，由智能体运行时自动匹配调用，无需修改提示词或流程图。 + +- **MCP 服务（Model Context Protocol）**：面向高扩展性、协议标准化的工具生态集成。支持官方服务（如 `Amap Maps`、`WebSearch`）和自定义部署（`npx`/`uvx` 脚本、AI 网关封装、OpenAPI 导入）。MCP 屏蔽通信细节，提供统一的工具发现（`list_tools`）与调用（`call_tool`）接口，适用于需要跨平台复用或深度定制工具链的场景。 + +> ⚠️ 注意：三者**不可混用**于同一调用上下文——插件与 Skill 仅支持在百炼托管智能体/工作流中使用；MCP 服务**不支持直连 `dashscope` SDK 的纯 API 调用**；而应用组件 API（如知识库、数据连接）本身属于平台基础设施，不归类为“工具”，但可被上述三类工具在运行时调用（例如 Skill 内部读取知识库检索结果）。 + +## 关键参数和配置 + +工具集成的配置围绕“标识”、“描述”、“权限”与“运行约束”四个维度展开，不同形态侧重点不同： + +| 形态 | 核心标识参数 | 必填描述字段 | 权限前提 | 典型运行约束 | +|------|----------------|----------------|------------|----------------| +| **插件** | `tool_id`（如 `"calculator"`） | 无（官方插件内置描述）；自定义插件需符合 OpenAPI 规范 | `AliyunServiceRoleForSFMAccessCloudAPI` 服务关联角色 | 输入字段名严格（如 `payload__input__text`）；`code_interpreter` 禁网络/禁文件上传；依赖库白名单（`pandas`, `matplotlib` 等） | +| **Skill** | `name`（小写字母+数字+连字符，全局唯一） | `description`（必须含输入类型、支持操作、触发关键词、明确排除场景） | 无额外 IAM 权限（依赖智能体所在空间权限） | ZIP 包 ≤10 MB；`description` 质量决定调用准确率；加密 PDF 等边界场景需显式排除 | +| **MCP 服务** | `service name`（控制台识别用）、`mcpServers` 中的 key（如 `"memory"`） | `description`（控制台展示用，不影响调用逻辑） | 无独立角色，但自定义服务若访问云资源（如 RDS），需为函数计算配置 VPC 或出口 IP 白名单 | 部署模式影响计费与延迟（基础模式冷启动，极速模式常驻）；必须符合 Streamable HTTP 协议（`POST /mcp`）；不支持本地资源访问 | + +> ✅ 统一要求：所有工具集成均需在**同一业务空间（Workspace）内完成注册与绑定**；跨空间调用必须先完成显式授权（插件）或服务共享配置（MCP/Skill）。 + +## 面向开发者，简洁实用 + +- **选型建议**： + - 快速验证通用能力 → 用**插件**（控制台一键添加，API 直接传 `tools` 数组）； + - 封装自有业务逻辑（如发票识别、合同比对）→ 用**Skill**（写好 `SKILL.md` + ZIP 上传，语义触发）； + - 构建可复用、可跨平台（Cherry Studio/Cursor）的工具生态 → 用**MCP**（优先 `npx` 部署开源 Server，或 AI 网关封装现有 API）。 + +- **调试要点**： + - 插件/Skill/MCP 均支持在智能体「对话测试窗格」中输入典型语句验证触发效果； + - 查看调用日志：插件 → 控制台「插件市场 > 调用记录」；Skill → 「Skill 管理 > 调用统计」；MCP → 「MCP 市场 > 服务监控」； + - 常见失败原因：权限缺失（检查服务关联角色）、输入格式错误（对照文档字段名）、环境限制（如 `code_interpreter` 无网络）、描述模糊（Skill 误触发/不触发）。 + +- **生产就绪检查清单**： + - [ ] 所有工具已通过安全扫描（Skill/MCP 上传后状态为 `active`，插件已授权）； + - [ ] `description` 字段已明确排除不支持的输入类型与边缘场景； + - [ ] 自定义工具（插件/Skill/MCP）已通过最小可行用例验证（如传空输入、超长文本、特殊字符）； + - [ ] 计费项已确认（如 `text_to_image` 限时免费，`WebSearch` 2000 次/月配额）； + - [ ] 生产环境使用固定版本（Skill/MCP 指定 `version`，避免 `latest`；插件 ID 不变更）。 + +## 关联主题页 + +- [plug in](../guides/plug-in.md) +- [skill](../guides/skill.md) +- [managed agents api](../api/managed-agents-api.md) +- [model context protocol](../guides/model-context-protocol.md) +- [application component api reference](../api/application-component-api-reference.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md b/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md deleted file mode 100644 index 195756dd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md +++ /dev/null @@ -1,57 +0,0 @@ -# 向量与嵌入 - -向量与嵌入（Embedding）是指将文本、图像、视频等非结构化内容映射为固定维度的数值向量，使语义相近的内容在向量空间中距离也更近，从而支持语义检索、聚类、推荐与 RAG 召回等下游任务。在百炼平台中，向量化是知识库检索与跨模态搜索的基础环节，与 Rerank 排序模型共同构成检索链路。 - -## 在百炼平台的使用场景 - -- **知识库检索召回**：文档搜索、数据查询、音视频搜索类知识库在创建时会对接向量模型，将入库切片向量化并写入索引；查询时对用户问题做同样的向量化，再执行向量 + 关键词混合检索。文档搜索、数据查询、音视频搜索类知识库支持 `text-embedding-v4` 或 `text-embedding-v3`（均为 512 维，维度不可更改）；视觉理解场景自动切换为 `qwen3-vl-embedding`；图片问答类仅支持 `multimodal-embedding-v1`（1024 维）。 -- **RAG 应用构建**：无论是云端知识库还是本地 RAG 方案，嵌入模型都决定了召回上限。云端方案使用百炼官方嵌入模型，不支持自定义切分与嵌入；本地 RAG 可改用自部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`），以灵活控制切分与嵌入。 -- **跨模态检索**：多模态向量模型将文本、图像、视频映射到同一语义空间，支持以文搜图、以图搜视频等。`qwen3-vl-embedding` 默认维度 2560，支持「独立向量」（每个输入各生成一向量，用于逐项对比）与「融合向量」（所有输入融合为 1 个向量，用于综合理解）两种类型。 -- **框架集成**：通过 LlamaIndex 构建 RAG 应用时，云端知识库默认使用百炼官方向量模型，配合 `similarity_top_k`、`similarity_cutoff`、`top_n` 等参数控制召回与重排。 - -## 文本向量模型 - -通用文本向量模型将文本转换为数值向量，当前推荐 `text-embedding-v4`（属 Qwen3-Embedding 系列），支持 100+ 主流语种。 - -| 模型 | 向量维度 | 最大行数 | 单行最大 Token | 语种 | -|------|---------|---------|---------------|------| -| text-embedding-v4 | 2048/1536/1024(默认)/768/512/256/128/64 | 10 | 8,192 | 100+ 语种 | -| text-embedding-v3 | 1024(默认)/768/512/256/128/64 | 10 | 8,192 | 50+ 语种 | -| text-embedding-v2 | 1,536 | 25 | 2,048 | 10 语种 | -| text-embedding-v1 | 1,536 | 25 | 2,048 | 6 语种 | - -**关键参数：** - -- `model`（必选）：模型名称 -- `input`（必选）：字符串、字符串列表或文件 -- `dimensions`（可选）：指定向量维度，仅 v3/v4 支持 -- `encoding_format`（可选）：当前仅支持 `float` - -**调用方式**：支持 OpenAI 兼容接口（`base_url: https://dashscope.aliyuncs.com/compatible-mode/v1`）和 DashScope SDK。对于大规模文本向量化，可使用异步批处理接口（`text-embedding-async-v1/v2`），单次最多 10 万行，通过 `X-DashScope-Async: enable` 请求头启用异步模式，提交后用 `task_id` 轮询结果；同时处理中任务不超过 50 个，并发上限 3。 - -## 多模态向量模型 - -多模态向量模型将文本、图像、视频统一映射到同一语义空间，支持跨模态检索。向量类型分「独立向量」（每个输入分别生成向量，适用于逐项对比）与「融合向量」（将所有输入融合为 1 个向量，适用于综合理解多模态内容）。`qwen3-vl-embedding` 默认维度 2560，支持独立与融合两种模式；`multimodal-embedding-v1` 默认 1024 维，用于图片问答类知识库。 - -## 知识库中的嵌入配置要点 - -- 文档搜索、数据查询、音视频搜索类知识库统一使用 512 维 `text-embedding-v4` / `text-embedding-v3`，维度不可更改。 -- 视觉理解场景自动切换为 `qwen3-vl-embedding`，无需手动指定。 -- 嵌入模型与切片策略共同决定召回质量：推荐使用智能切分（基于语义自适应选择切片点，单切片 Token 上限 6,000），切片过大或过小都会影响向量匹配效果。 -- 本地 RAG 场景若需自定义嵌入模型，可改用自部署 GTE 文本向量模型，但需注意 embedding API 限流，不建议传入超过 100 MB 的文件。 - -## 与 Rerank 的协作 - -向量检索负责语义召回（初步 TopK 默认 50，可设 1–100），Rerank 模型负责二次精准排序。知识库中可选 `qwen3-rerank` / `qwen3-rerank(hybrid)` / `qwen3-vl-rerank`（多模态库只能选 vl-rerank），排序后再按相似度阈值（0.01–1.0）与最大召回数量（1–20）裁剪。向量召回质量直接影响 Rerank 上限，调优时通常先确认嵌入模型与切片合理，再调整 Rerank 阈值与 K 值。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) -- [frameworks](../api/frameworks.md) -- [application use cases](../guides/application-use-cases.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md b/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md deleted file mode 100644 index 523b5b79..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md +++ /dev/null @@ -1,55 +0,0 @@ -# 私网访问 - -私网访问（VPC Private Access）是指阿里云百炼平台的模型调用、[模型部署](model-deployment.md)、异步任务通知等能力通过 VPC 内网或私网链路访问，避免公网暴露、提升传输安全性与网络稳定性的接入方式。它是百炼安全合规体系中"传输加密—私网访问"环节的核心抓手，与身份权限、内容安全、合规备案、安全存储共同构成端到端安全链路。 - -## 在百炼平台中的使用场景 - -### 1. 模型调用的私网接入 - -百炼模型调用默认通过公网（`dashscope.aliyuncs.com`）发起，对安全要求高或需统一出口的企业可走 VPC 私网。临时 API Key 可在不可信客户端环境使用，配合 VPC 私网访问可在内网完成调用链路，避免永久 Key 在公网流转。 - -> 各地域（北京、新加坡、弗吉尼亚）的 API Key 不互通，临时 Key 与生成它的永久 Key 必须属于同一地域，私网链路也需在对应地域内打通。 - -### 2. 异步任务完成通知的 VPC 回调 - -图像/视频生成等耗时任务接入事件总线 EventBridge 后，任务完成事件由事件总线推送。HTTP 回调 URL 接收端**支持公网或 VPC 访问**，对安全敏感场景建议使用 VPC 类型的 HTTP 接口作为事件目标，使通知链路全程在内网闭环。 - -- 事件源：`acs.dashscope` -- 事件类型：`dashscope:System:AsyncTaskFinish` -- 事件总线默认为 `default`（北京地域云服务专用总线） - -RocketMQ 消息队列方案同样可部署在 VPC 内，由事件总线投递后业务方在内网消费，支持消息无丢失与失败重试，可靠性更高。 - -### 3. [模型部署](model-deployment.md)与模型导入的私网通道 - -[模型部署](model-deployment.md)（PTU、模型单元、按 Token 用量）仅适用于华北二（北京）地域，部署后推理服务可在 VPC 内调用，满足高并发、低延迟及不出公网的诉求。 - -部署自定义 LoRA 模型时，需将本地训练的 LoRA 模型从阿里云 OSS 导入百炼。OSS Bucket 支持**私有 Bucket**与**内容加密**，并需添加 `bailian-datahub-access` 标签（值为 `read`）。通过 VPC 终端节点或私网访问 OSS，可避免训练产物经公网传输，保障数据安全。 - -## 关键参数与配置 - -| 维度 | 关键约束 | -| --- | --- | -| 地域一致性 | API Key、临时 Key、私网链路须同地域；模型部署仅北京 | -| 临时 Key 有效期 | 默认 60 秒，可通过 `expire_in_seconds` 设置，范围 [1, 1800] 秒 | -| 异步任务接口限流 | 20 QPS（按主账号 + 子账号维度）；通知方案不限流 | -| 任务保留时长 | 完成后约 24 小时自动清理 | -| OSS 导入标签 | Bucket 须加 `bailian-datahub-access=read` 标签 | -| OSS 存储类型 | 不支持归档/冷归档/深度冷归档；须使用子目录，非根目录 | -| 授权角色 | OSS 导入需开通 `AliyunServiceRoleForSFMDataHubOSSImport` 服务关联角色 | - -## 开发者实践建议 - -- **生产隔离**：按环境（dev/test/prod）划分业务空间，prod 流量走 VPC 私网，降低公网泄露面。 -- **回调优先 VPC**：异步任务通知的 HTTP 回调端点部署在 VPC 内，事件总线直推即可，无需公网暴露。 -- **OSS 导入走私网**：LoRA 模型导入使用私有 Bucket + 内容加密，并通过 VPC 终端节点访问 OSS。 -- **避免跨地域**：规划业务空间时明确地域，避免因 Key 与链路跨地域导致调用失败。 -- **PTU 部署内网化**：高负载生产环境优先 PTU 部署并经 VPC 调用，兼顾稳定吞吐与传输安全。 - -## 关联主题页 - -- [security and compliance](../guides/security-and-compliance.md) -- [more about models](../api/more-about-models.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md b/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md deleted file mode 100644 index 0badc9bf..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md +++ /dev/null @@ -1,75 +0,0 @@ -# 工作流 - -工作流（Workflow）是百炼平台三种核心应用构建模式之一，通过可视化节点编排将复杂任务拆解为有序步骤，逻辑确定、稳定可复现，适合流程固定、要求可复现的业务场景。 - -## 在百炼平台中的定位 - -百炼提供智能体、工作流、高代码应用三种互补的应用构建模式。工作流对应"可视化节点编排（低代码）"路线，由预定义节点精确控制流程，适合 IT 运维、业务分析师构建报告生成、订单处理、审批流等场景。与由大模型自主规划的智能体不同，工作流强调开发者对流程的显式控制，输出可复现、可调试。 - -## 核心节点 - -工作流画布由以下节点类型组成： - -- **开始 / 结束节点**：定义输入与输出参数。开始节点预置 `query`（用户输入）、`historyList`（对话历史）、`imageList`（图片）等变量，可在下游节点中引用。 -- **大模型节点**：执行 LLM 推理，配置模型、提示词、用户提示词、记忆等。 -- **意图分类节点**：根据输入分流到不同下游分支，支持多意图。 -- **变量处理节点**：用于文本输出或变量加工。 -- **智能体群组节点**：将任务分解给多个已发布的子智能体协同完成。 - -## 会话变量 - -会话变量作为全局变量在工作流全生命周期内记录参数，可在各节点中引用，在画布右上角配置。适合跨节点传递状态、累积上下文信息。 - -## 典型案例 - -- **诈骗信息识别**：开始 → 大模型（提示词判定诈骗嫌疑）→ 结束。 -- **智能导购**：意图分类节点将输入分流到手机 / 电视 / 冰箱等大模型分支，未命中分支走变量处理节点。 -- **日程管理助手**：通过智能体群组节点串联"信息收集"与"数据整理"两个子智能体。 - -## 创建与发布流程 - -1. 控制台 → 应用管理 → 创建应用 → 工作流应用。 -2. 在画布上拖拽编排节点，配置各节点的模型、提示词、变量引用关系。 -3. 右侧对话框调试。 -4. 右上角"发布"——发布是后续 API 调用与集成的前提。 - -## API 调用 - -工作流应用与智能体应用共用同一套调用接口，区别仅在应用内部的编排逻辑。发布后通过 `APP_ID` 调用： - -- **DashScope 原生 API**：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion`，请求体为 `{"input": {"prompt": "..."}, "parameters": {}, "debug": {}}`，响应中业务侧主要消费 `output.text`。 -- **OpenAI 兼容 Responses API**：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`，可复用现有 OpenAI 生态代码库，支持同步 / 异步、流式、多模态。 -- **SDK**：Python 使用 `dashscope.Application.call`，Java 使用 `com.alibaba.dashscope.app.Application`，Node.js 可直接以 `axios` 发起 POST。 - -调用前需在控制台获取应用 ID 与 API Key（推荐写入 `DASHSCOPE_API_KEY` 环境变量）；若应用位于子业务空间，还需提供 Workspace ID。 - -## 多轮对话 - -工作流应用支持多轮对话，有两种实现方式： - -- **使用 `session_id`**：系统自动从云端加载历史对话，实现简单。`session_id` 有效期 1 小时，最多支持 50 轮。 -- **自行管理 `messages`（推荐）**：手动维护消息列表，灵活性更高，不受 session 有效期限制。 - -## 评测 - -工作流应用可作为评测任务的被测对象。在新版应用评测中，评测集支持智能体、工作流、自定义三种类型，按所选应用的出入参形式自动生成数据模板。评测任务可关联工作流应用，由 LLM 评估器或 Code 评估器自动评分，也可人工标注。 - -## 发布与分享 - -工作流应用支持多种发布渠道：UI 应用（通过 UI 设计器构建自定义界面）、钉钉机器人、微信公众号、组件化复用（将工作流作为其他智能体或工作流的子组件）、音视频实时互动（仅限图文对话类）。需要注意的是，官方网页版分享当前只支持智能体应用，不支持工作流应用。 - -## 关键限制与注意事项 - -- 工作流应用若配置了文件类型的自定义参数，在 UI 设计器中需指定 `{{{file_name:files[0]}}}`（将 `file_name` 替换为实际变量名），否则应用无法正确读取用户上传的文件。 -- 旧版"智能体编排"应用已被工作流应用替代，新建应用请直接选择工作流。 -- 工作流与智能体调用接口一致，但可附加的扩展能力（如自定义参数传递）取决于应用内部编排逻辑。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [application evaluation](../guides/application-evaluation.md) -- [application publishing and sharing](../guides/application-publishing-and-sharing.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/workspace-isolation.md b/skills/bailian-docs-llm-wiki/wiki/concepts/workspace-isolation.md new file mode 100644 index 00000000..ea39b5ed --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/workspace-isolation.md @@ -0,0 +1,45 @@ +# 业务空间隔离 + +业务空间隔离是百炼平台的核心安全与治理机制，指以 `Workspace`（业务空间）为最小逻辑单元，对计算资源、数据资产、模型权限、API 调用、监控数据及安全策略实施严格边界划分，确保不同业务主体（如部门、项目、客户）之间资源不可见、权限不越界、数据不交叉、行为可审计。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +- **API 调用层**：所有 OpenAPI（应用组件、托管智能体等）均强制要求传入 `workspace_id`（路径参数或 Header），服务端据此路由请求、校验权限、隔离资源。例如 `AddFile` 创建的文件仅在指定 Workspace 内可见，跨 Workspace 的 `Retrieve` 请求将直接拒绝。 + +- **权限管理层**：RAM 用户的 API Key 绑定且仅属于单一 Workspace；模型调用开关、QPM/[Token](token.md) 限流、页面菜单权限均按 Workspace 独立配置。一个用户在 Workspace A 有 `qwen-max` 调用权，在 Workspace B 无该权限，二者完全独立。 + +- **运行时环境层**：托管智能体（Managed Agents）的 `Agent`、`Environment`、`Session` 均归属特定 Workspace；沙箱执行、文件挂载、Skill 加载均受 Workspace 边界约束，无法跨空间访问或共享。 + +- **可观测性层**：应用监控（Application Monitoring）仅展示当前 Workspace 内已发布应用的 Trace 数据；Trace ID 和 Span 数据天然绑定 Workspace，不同 Workspace 的调用链互不可见、不可关联。 + +- **安全合规层**：私网访问（PrivateLink）、AI 安全护栏启用、加密推理密钥管理等能力均需在 Workspace 级别开通与配置；安全策略（如 `X-DashScope-DataInspection` 生效范围）也以 Workspace 为作用域。 + +## 关键参数和配置 + +- `workspace_id`：业务空间唯一标识符，**所有 API 必填参数**。常见传递方式： + - ROA 接口：作为路径参数（如 `/workspaces/{workspace_id}/apps`） + - REST 接口：通过 `X-Workspace-ID` HTTP Header 传递 + - SDK：初始化客户端时显式指定（如 Python SDK 的 `workspace_id="ws-xxx"`） + +- `region`：地域标识（如 `cn-beijing`），与 `workspace_id` 组合构成资源全局定位。**同一逻辑 Workspace 在不同 region 视为完全独立实体**，权限、数据、配额均不互通。 + +- `api_key`：绑定至**单一 region + 单一 workspace_id + 单一 RAM 用户**，不可复用、不可迁移。创建后即锁定其作用域。 + +> ⚠️ 注意：默认业务空间（系统自动创建）无完整权限控制能力，开发者应主动创建自定义 Workspace 以获得模型限流、细粒度授权等能力。 + +## 面向开发者，简洁实用 + +- ✅ **必须做**：每次调用百炼 API 前，确认 `workspace_id` 正确且与你的 API Key 所属空间一致；检查 `region` 是否匹配 Endpoint（如北京地域用 `cn-beijing`）。 +- ✅ **推荐实践**：为不同项目/客户分配独立 Workspace，避免权限混用与数据泄露风险；通过 RAM 策略精确授予 `AliyunBailianDataFullAccess` 等 Workspace 级权限。 +- ❌ **禁止操作**：不要尝试复用 API Key 跨 Workspace 调用；不要在代码中硬编码 `workspace_id`，建议从环境变量或配置中心注入。 +- 🔍 **排障提示**：若遇到 `403 Forbidden` 或 `ResourceNotFound`，优先检查 `X-Workspace-ID` 是否缺失、错误，或当前 API Key 是否未被授权该 Workspace。 + +## 关联主题页 + +- [application component api reference](../api/application-component-api-reference.md) +- [managed agents api](../api/managed-agents-api.md) +- [application permission management](../guides/application-permission-management.md) +- [application monitoring](../guides/application-monitoring.md) +- [security and compliance](../guides/security-and-compliance.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md b/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md deleted file mode 100644 index 51a00e4c..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md +++ /dev/null @@ -1,61 +0,0 @@ -# 业务空间（Workspace） - -业务空间（Workspace）是阿里云百炼平台进行精细化权限管理（模型、用户）和阿里云账单分账的**最小管理单元**。平台按地理区域划分资源与业务空间，单个业务空间不能跨地域存在，即便是各地域的默认业务空间也彼此独立。 - -## 在不同场景中的使用 - -### 1. 权限与身份管理 - -业务空间是权限体系的核心边界，权限管理围绕三种角色展开： - -- **超级管理员**：阿里云主账号，或拥有 `AliyunBailianFullAccess` 系统策略的 RAM 用户。可跨空间统一管理用户权限、空间可用模型、模型限流和 API Key。 -- **业务空间管理员**：拥有访问某个业务空间「权限管理」页面的 RAM 用户，仅管理该空间内的用户与资源，其权限包含该空间下所有页面的访问权限。 -- **普通用户**：仅能访问/使用被授权的空间、页面与资源，不能管理用户或模型授权。 - -在非默认业务空间中，可对模型进行三类精细化授权（**默认业务空间无法设置这些限制**，即所有模型均可调用、调优、部署且无法限流）： - -| 权限项 | 控制范围 | -| --- | --- | -| 限制模型调用 | 是否可调用（控制台 & API）+ 请求数限流 + Token 限流 | -| 限制模型训练 | 是否可调优（控制台 & API）及调优后部署 | -| 限制模型部署 | 是否可直接部署 | -| 用户控制台权限 | 管理 RAM 用户能否使用该空间控制台及可用功能 | - -> **注意**：OpenAPI 接口权限不通过业务空间角色授予，必须由阿里云主账号在 RAM 控制台为 RAM 用户单独添加系统策略（如 `AliyunBailianDataFullAccess` / `AliyunBailianDataReadOnlyAccess`）。 - -### 2. API Key 归属 - -单个 API Key 只能归属**一个地域内的一个业务空间和一个用户**，且不能转移。API Key 可调用的功能与模型限流与其**归属业务空间**的权限保持一致，不受用户控制台权限管理的影响。将 RAM 账号移出业务空间会使其 API Key 失效（重新加入后恢复生效）。 - -### 3. 应用组件 API 调用 - -调用百炼应用组件 API（`bailian/2023-12-29`，含数据连接、知识库、Prompt 模板、长期记忆等）时，所有接口均需传入 `WorkspaceId`（业务空间 ID）。RAM 子账号需先获取对应权限策略并加入业务空间后才能调用。类目等资源也以业务空间为界，例如每个业务空间最多可新建 500 个类目。 - -### 4. 应用观测 - -应用观测以业务空间为范围，端到端查看空间内应用（智能体应用、工作流应用、高代码应用）的处理流程与延时、Token 等指标。若应用观测列表中看不到已创建的应用，常见原因之一即为该应用不属于当前业务空间。 - -### 5. 数据管理 - -数据管理功能统一管理业务空间下的大模型相关数据集（训练集、评测集）。该能力目前仅适用于华北2（北京）地域。 - -## 关键参数与配置 - -- **`WorkspaceId`**：业务空间 ID，是应用组件 OpenAPI 的必传参数，用于标识资源所属空间。 -- **地域隔离**：业务空间绑定单一地域，跨地域需在对应地域分别创建/使用空间。 -- **默认业务空间的限制**：默认空间无法设置模型调用/训练/部署授权与限流，如需精细化管控请使用非默认业务空间。 - -## 生产环境实践 - -- **空间规划**：推荐按环境（dev/test/prod）划分业务空间实现隔离，或按业务线划分以便权限与成本管理。 -- **限流策略**：将主账号总配额按比例分配给各业务空间并预留缓冲。例如总配额 1000 QPM，可分配 prod 600 / test 200 / dev 100，预留 100。 - -## 关联主题页 - -- [application component api reference](../api/application-component-api-reference.md) -- [application permission management](../guides/application-permission-management.md) -- [application monitoring](../guides/application-monitoring.md) -- [security and compliance](../guides/security-and-compliance.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md index 508d7604..374e4256 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md @@ -1,152 +1,46 @@ # application evaluation -阿里云百炼提供完整的应用[评测体系](../concepts/evaluation.md)，支持对[智能体应用](../concepts/agent-application.md)和工作流应用的输出质量进行系统化评估。平台同时提供自动评测与手动评测两种模式，并通过评测集、评估器和标签三大组件构建多维度的评测闭环。当前平台存在新旧两套评测系统，新版在评测任务管理、评估器和标签体系上做了较大升级。 +application evaluation 是阿里云百炼平台用于系统化评估智能体（Agent）与工作流应用输出质量的核心能力，支持自动与手动两种评测范式。它通过评测集驱动、多维度评估器打分、人工标签标注及归因分析，帮助开发者量化效果、定位问题并闭环优化。该能力深度集成于应用观测体系，要求应用已发布且配置知识库（自动评测）或已接入观测（部分场景），是 RAG 应用调优的关键基础设施。 -## 评测模式 +## 支持的模型/功能 -百炼支持两种评测模式，分别适用于不同场景： +- **自动评测**：基于大模型（当前仅支持 `qwen-max` 和 `qwen-plus`）自动生成评测集并执行端到端评分，适用于已发布的智能体应用，依赖知识库与应用观测功能 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **手动评测**：支持人工构建评测集（XLS/XLSX 格式），通过人工打标完成效果评估，适用于对话分析类场景 [手动评测](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md)。 +- **新版评测体系**：引入“智能体”“工作流”“自定义”三类评测集类型，并支持从应用观测真实数据导入评测集，突破旧版仅限知识问答的限制 [新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md)。 +- **评估器（Grader）**：提供预置模板（如相关性、格式校验、文本相似度）及自定义 LLM/Code 评估器，支持多维度自动评分；LLM 评估器需指定模型（评估模型限时免费），Code 评估器通过 Python 脚本实现确定性规则判断 [评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 +- **标签管理**：支持分类、布尔值、数字、文本四类标签，用于人工标注与多维统计分析，可复用于评测任务与应用观测模块 [标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md)。 -### 自动评测 +> **注意**：文档 1（自动评测）明确限定生成评测集与执行评测均仅支持 `qwen-max` 和 `qwen-plus`；而文档 7（评估器）指出 LLM 评估器“选择模型下拉框”中可选模型未限定具体型号，且强调“评估模型限时免费”。二者存在模型范围不一致风险——实际使用中应以自动评测流程的硬性约束为准，即 `qwen-max`/`qwen-plus` 是当前唯一受支持的自动评测底座模型。 -[自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)利用大模型基于应用关联的知识库自动生成评测集，并对智能体的回答进行自动评分，生成评测报告与调优建议。支持两种子模式： +## 关键参数 -- **单应用评测**：深度评估单个[智能体应用](../concepts/agent-application.md)的表现，生成包含评分、错误分析和优化建议的详细报告。 -- **多应用横向评测**：在同一评测基准下对比最多 8 个应用（或同一应用的不同版本），用于选型决策或版本迭代效果验证。 +- **评测集字段**：知识问答型评测集必需 `query`、`referenceAnswer`、`coarseKeywords`、`fineKeywords`、`queryType` 字段（JSONL 格式）；对话分析型必需 `Prompt`、`Completion`、`SessionId`（XLS/XLSX 格式）[评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)。 +- **采样与权重**：自动评测中可通过滑块设置各任务类型（事实型、分析型等）的分类采样数；新版评测任务支持为每个评估器配置独立的评分范围（如 0–1 或 1–5）和通过阈值 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)、[评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 +- **评估器映射**：在评测任务中添加评估器后，必须完成所有变量（如 `query`、`response`、`reference`）到评测集字段或应用输出的精确映射，否则任务无法创建 [评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 +- **标签类型参数**：分类标签最多支持 20 个筛选项；数字标签支持 Double 类型；文本标签输入上限 200 字 [标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md)。 -前提条件： +## 使用方式 -1. 仅面向**已发布**的[智能体应用](../concepts/agent-application.md)，且应用须已配置知识库。 -2. 须开通**应用观测**功能，并将待评测应用添加到观测列表。 -3. 子账号需获取`管理员`或`应用评测-操作`权限。 -4. 多应用横向评测时，所有被选应用必须关联至少一个相同的知识库。 +1. **准备数据**： + - 自动评测：确保智能体已发布、关联知识库、开通应用观测； + - 手动评测：下载模板，按 `Prompt`/`Completion`/`SessionId` 填写 XLS/XLSX 文件； + - 新版评测集：可手动上传（支持智能体/工作流/自定义类型），或从应用观测导入真实 Span 数据 [新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md)、[评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)。 -自动评测流程分四步：创建评测任务 → 设置评测集 → 配置评测规则 → 执行评测。评测集生成和评估模型当前仅支持 `qwen-max` 和 `qwen-plus`。 +2. **创建评测任务**： + - 自动评测：控制台 → 创建评测任务 → 选应用/知识库 → 生成或选评测集 → 配置采样与模型 → 发起； + - 新版评测任务：控制台 → 创建评测任务 → 选评测集版本 + 关联应用（智能体/工作流/不关联）→ 添加评估器（必配参数映射）+ 标签 → 完成 [评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 -### 手动评测 +3. **执行与分析**： + - 自动评测：查看总正确率、BadCase 归因（模型理解有误/重排不佳/检索无效/切片不完整/未获取知识）、RAG 分项得分； + - 手动/新版评测：进入任务详情页，使用“普通模式”或“快速标注”进行人工打标，结合评估器自动评分结果，在“指标统计”页查看综合得分、通过率、数据分布 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)、[评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 -[手动评测](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md)通过人工构建评测集，对应用回答进行人工分析与评分。流程为：准备评测集（下载模板填充数据）→ 上传评测集 → 创建评测任务 → 人工标注打分 → 查看评测报告。 +## 限制和注意事项 -手动评测适用于需要领域专家主观判断的场景，评测维度支持使用内置模板或自定义评测维度模板。 - -> **注意**：手动评测属于旧版评测系统的功能。新版评测系统通过「评测任务 + 标签」的组合同样支持人工标注场景，且功能更灵活。 - -## 评测集 - -评测集是评测任务的数据基础，用于存储和管理评测数据。 - -### 旧版评测集 - -[旧版评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)支持两种类型： - -| 类型 | 文件格式 | 适用场景 | -|------|----------|----------| -| 对话分析 | `.xls` / `.xlsx` | 人工评测，包含 Prompt、Completion、SessionId 字段，支持多轮对话 | -| 知识问答 | `.jsonl` | 自动评测，包含 query、queryType、referenceAnswer、fineKeywords、coarseKeywords 字段 | - -创建方式：自动生成（基于知识库，仅知识问答类型）或手动上传。单次上传最多 10 个文件，单个文件不超过 20MB。 - -### 新版评测集 - -[新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md)支持三种类型： - -| 类型 | 说明 | -|------|------| -| 智能体 | 根据选中智能体应用的出入参形式定义评测集 | -| 工作流 | 根据选中工作流应用的出入参形式定义评测集 | -| 自定义 | 任意定义评测集表结构，适用于特殊评测场景 | - -新版评测集支持手动上传和从应用观测导入两种创建方式，并具备版本管理能力，每次发布生成新版本。创建后类型不可修改。 - -## 评估器 - -[评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)是新版[评测体系](../concepts/evaluation.md)的核心组件，用于自动评估应用输出质量。支持三种创建方式： - -### 基于预置模板 - -百炼提供多种预置评估器模板，覆盖以下分类： - -- **通用质量**：评估回答的基本质量指标 -- **智能体**：专门用于评测智能体应用 -- **文本匹配**：精确规则文本匹配 -- **文本相似度**：计算文本相似度得分 -- **格式校验**：验证输出格式规范性 - -### 自定义评估器 - -| 类型 | 评估方式 | 适用场景 | 成本 | -|------|----------|----------|------| -| LLM 评估器 | 大模型语义理解 | 相关性、有害性、幻觉检测 | 产生 [Token](../concepts/token.md) 费用 | -| Code 评估器 | Python 代码规则判断 | 格式校验、数值计算、精确匹配 | 无额外费用 | - -### 基于评测任务创建 - -通过历史评测任务的标注结果自动抽象为新的 LLM 评估器，适用于将人工标注经验固化为自动化评估规则的场景。需选择已完成评估的评测任务，并配置 query、response、label_score 的字段映射。 - -每个评测任务最多支持添加 10 个评估器。建议组合 3-5 个评估器从不同维度评估应用质量，例如：相关性评估器（LLM）+ 格式校验评估器（Code）。 - -## 标签管理 - -[标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md)用于对评测数据和应用观测数据进行自定义标注，支持四种标签类型： - -| 标签类型 | 数据类型 | 适用场景 | -|----------|----------|----------| -| 分类 | 预定义选项（最多 20 个） | 回答质量分级、错误类型分类 | -| 布尔值 | True / False | 是否正确、是否存在幻觉 | -| 数字 | Double 数值 | 评分（1-5）、相关性得分（0-1） | -| 文本 | 自由文本 | 错误原因说明、改进建议 | - -标签可同时用于评测任务的人工标注和应用观测的数据标注，支持基于标签的数据筛选和指标统计。 - -## 评测任务（新版） - -[新版评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)支持智能体和工作流应用的评测，核心配置包括： - -- **选择评测集**：从已发布的评测集列表中选择评测集和版本 -- **关联应用**：支持不关联应用（纯人工标注）、关联工作流或关联智能体三种方式 -- **添加评估器**：配置自动评分规则及参数映射 -- **添加标签**：配置人工标注维度 - -任务详情页提供数据明细和指标统计两个视图，支持普通模式和快速标注两种标注方式。 - -## 评测报告与归因分析 - -自动评测完成后生成评测报告，包含以下维度： - -- **总正确率**：得分 >= 4 分的回答占比（评分范围 1-5 分） -- **BadCase 分析**：按分数从低到高展示错误评测条目 -- **调优建议**：基于归因分析提供 Prompt、检索配置或知识库切片的具体优化建议 -- **RAG 智能体评价**：按问题类型展示单项得分 - -归因分析将 BadCase 定位到 RAG 流程的具体环节： - -| 归因类型 | 含义 | 优化方向 | -|----------|------|----------| -| 模型理解有误 | 已获取正确知识但推理错误 | 优化提示词或切换更强模型 | -| 重排不佳 | 正确切片排序靠后 | 调整重排配置或增加切片数量 | -| 检索无效 | 召回切片过多或过少 | 调整检索策略 | -| 切片不完整 | 语义单元被分割到多个切片 | 增大切片长度或启用语义切分 | -| 未获取知识 | 知识库缺少相关内容 | 补充知识库内容 | - -## 最佳实践 - -### 建立持续评测机制 - -以下场景建议触发评测：知识库更新后、调整 Prompt 后、更换或升级模型后、调整检索/重排策略后、定期回归（每周或每月）。 - -### 优化闭环 - -识别 BadCase → 分析归因定位问题 → 实施针对性优化 → 发布新版本再次评测 → 对比结果确认改进。若效果未达预期则继续迭代。 - -## 计费说明 - -评测任务调用大模型产生的 [Token](../concepts/token.md) 费用正常计费。自动评测的评测集生成和评估均会消耗 [Token](../concepts/token.md)，预估平均消耗仅为参考值，最终以实际账单为准。评估器模型当前限时免费。 - -## 常见问题 - -- **评测集生成进度长时间保持 0%**：评测集生成和应用评测为离线任务，需后台排队执行，排队期间进度保持 0%，任务开始后自动更新。 -- **评测期间能否关闭应用观测**：不可以，否则可能导致评测任务失败或数据丢失。 -- **评测报告中用例数量与设置不符**：自动评测可能部分失败，报告仅展示成功完成的用例。 -- **评测任务创建后可否修改**：任务配置（应用、评测集）不可修改，但可随时添加人工标签。如需不同配置请创建新任务。 +- **应用状态限制**：自动评测仅支持**已发布**的智能体应用；手动评测与新版评测任务虽支持“不关联应用”，但若需调用应用推理，则仍要求应用处于发布状态 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)、[手动评测](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md)。 +- **数量上限**：单次自动评测最多支持 8 个应用横向对比；单个评测任务最多添加 10 个评估器；单次上传评测集文件最多 10 个，单文件 ≤20MB [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)、[评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 +- **评测集兼容性**：自动评测仅支持知识问答型（JSONL）评测集；手动评测仅支持对话分析型（XLS/XLSX）；新版评测集三类类型互不兼容，创建后不可修改类型 [评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)、[新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md)。 +- **权限与观测依赖**：子账号需具备 `管理员` 或 `应用评测-操作` 权限；自动评测强依赖应用观测功能开启，评测期间关闭观测将导致任务失败或报告不准 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **[Token](../concepts/token.md) 消耗说明**：所有涉及大模型调用的操作（评测集生成、自动评测、LLM 评估器运行）均产生 [Token](../concepts/token.md) 费用；预估消耗为参考值，实际以账单为准；失败步骤的已消耗 [Token](../concepts/token.md) 仍计费 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)、[评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 ## 来源文档 @@ -159,11 +53,3 @@ - [评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md) - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md index 9ae2c8d0..2fe80d3c 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md @@ -1,128 +1,60 @@ # application monitoring -阿里云百炼提供**应用观测**功能，用于端到端查看[业务空间](../concepts/workspace.md)内应用（[智能体应用](../concepts/agent-application.md)、[工作流](../concepts/workflow.md)应用、高代码应用）的处理流程，并获取延时、[Token](../concepts/token.md) 量等关键指标，指标更新频率为分钟级。该功能可帮助开发者追踪应用内部调用链路、查看模型响应延时与思考过程，进而优化运营效果与成本。详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +应用监控（Application Monitoring）是百炼平台提供的端到端可观测能力，用于追踪智能体、工作流及高代码类应用的内部执行链路，涵盖向量生成、知识检索、大模型调用等关键节点，并提供延时、[Token](../concepts/token.md) 用量、状态码等分钟级指标。该功能不提供 API 接口，所有数据通过控制台交互式查看与导出。详细背景和设计目标参见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 -## 支持的应用范围 +## 支持的模型/功能 -应用观测支持以下三类应用： +- **支持的应用类型**：智能体应用、工作流应用、高代码应用（但高代码应用仅上报根节点 `FullCodeApp`，[不支持内部调用链路追踪](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **不支持的应用类型**：通过 Assistant API 创建的智能体应用（[明确排除](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **可观测节点类型**： + - 通用节点：`CHAIN`（根节点，如 `AgentApp`/`WorkflowApp`）、`LLM`（含输入/输出 [Token](../concepts/token.md) 统计）、`EMBEDDING`、`RETRIEVER`（含 `TextRetriever`/`VectorRetriever`）、`RERANKER`、`REWRITER`、`GUARDRAIL`、`TOOL`； + - 工作流特有节点：`START`、`END`、`API`、`CLASSIFIER`、`TEXT_CONVERTER`、`SCRIPT`、`CONDITION`、`FUNCTION_COMPUTE`、`APP_FLOW`； + - 高代码应用仅支持 `CHAIN` 类型下的 `FullCodeApp` 节点，无子节点展开能力。 -- **[智能体应用](../concepts/agent-application.md)**（AgentApp） -- **[工作流](../concepts/workflow.md)应用**（WorkflowApp） -- **高代码应用**（FullCodeApp） +> **注意**：文档中对 `RETRIEVER` 子节点的说明在“智能体应用”和“工作流应用”两节中完全一致，但未提及[长期记忆](../concepts/long-term-memory.md)检索——[原文明确指出目前暂不支持观测在长期记忆中的检索过程](../../raw/application-user-guide/application-monitoring/application-observation.md)，该限制适用于所有应用类型。 -> **注意**：应用观测暂不支持通过 Assistant API 创建的[智能体应用](../concepts/agent-application.md)；对高代码应用，目前不支持追踪其内部调用链路，仅能观测到入口 CHAIN 节点。应用观测本身也没有 API，只能通过控制台操作。 +## 关键参数 -## 前提条件与开通 - -首次使用需在应用观测页面右上角完成**应用观测配置**，依次执行：授权可观测链路 OpenTelemetry 服务角色权限 → 开通 OpenTelemetry 服务 → 初始化 LogStore。 - -- 推荐使用**主账号**操作，开通后通常分钟级生效，高峰期可能略有延迟。 -- 如需**子账号**开通，主账号需为其配置 `AliyunBailianFullAccess` 全局权限、`应用观测-操作`（或 `管理员`）页面权限，并额外授予 `ram:CreateServiceLinkedRole` 系统策略（用于创建服务关联角色）。 - -> 子账号权限若未配置完整，开启应用观测时会失败。配置完成后需返回应用观测界面再次尝试开启。 +| 参数 | 说明 | 来源 | +|------|------|------| +| `Request ID` / `Trace ID` / `Span ID` | 用于精准定位单次调用链路，可在节点详情页点击“查看 ID”获取 | [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| 延时（调用时长） | LLM 节点包含完整流式响应时间；平均首 [Token](../concepts/token.md) 耗时专用于流式场景 | [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| Token 总量 | = 输入 Token + 输出 Token；Embedding 节点 Token 量仅统计向量化输入 | [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| 状态 | `正常` 或 `错误`（可进一步按错误类型细分） | [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) | ## 使用方式 -### 1. 选择被观测的应用 - -在应用观测页面单击「选择被观测的应用」>「添加」。若列表中看不到已创建的应用，通常是因为该应用尚未发布，或应用不属于当前[业务空间](../concepts/workspace.md)。 - -### 2. 开始观测 - -添加完成后，应用会出现在观测列表中。此后所有输入该应用的 Prompt 及相关数据、指标会被自动追踪并以分钟级频率同步。单击「关闭观测」可停止同步，重新添加后仅同步新增数据。 - -在「查看详情」中可查看最长 30 天内的调用记录，包括 Prompt 内容、输出、延时、调用时间和 [Token](../concepts/token.md) 量，并支持按 Request ID / Trace ID / Span ID 检索和按时间范围筛选。单击节点名称可查看详情、原始数据和标注记录。 - -> 列表中的 **CHAIN** 节点表示一次完整的应用内部调用追踪，支持展开。状态分为「正常」与「错误」两类。 - -### 3. 导出数据 - -在应用详情页的 Trace 列表页签右上角单击「导出数据」，可将当前筛选条件下的数据导出为 **JSONL** 或 **EXCEL** 格式。 - -### 4. 查看监控统计 - -「监控统计」页签提供性能监控图表：调用次数（含失败次数与失败率）、[Token](../concepts/token.md) 总量（全部/输入/输出）、平均单次请求 [Token](../concepts/token.md) 量、平均首 [Token](../concepts/token.md) 耗时（流式场景）、平均调用时长。支持按时间范围（最长 30 天）和聚合粒度（分钟/小时/天）查看，每个图表可放大、下载、复制。 - -## 数据筛选与标注 - -### Span 筛选模式 - -- **Root Span**：仅显示根节点（默认） -- **All Span**：平铺展示所有 Span -- **Model Span**：仅显示包含模型调用的 Span - -### 过滤器 - -支持按状态（正常/错误，可按错误类型细分）、Span Name、输入、输出、延时、[Token](../concepts/token.md) 总量、输入 [Token](../concepts/token.md)、输出 [Token](../concepts/token.md)、标签等字段添加多个筛选条件，条件之间组合应用。 - -### 数据标注 - -支持对 Span 数据添加标签（布尔值/分类/数字/文本四种类型），标签与应用[评测](../concepts/evaluation.md)的标签管理共享、统一管理。标注内容自动保存，并可在 Span 列表页「标签」列查看。 - -### 添加到[评测](../concepts/evaluation.md)集 - -应用观测支持将 Span 数据直接加入[评测](../concepts/evaluation.md)集，将真实线上调用作为评测样本。配置时需选择目标评测集、导入方式（追加数据或全量覆盖）并完成字段映射。每个评测集最多支持 50 个字段映射。 - -## 节点类型 - -被观测应用在调用过程中会按操作单元生成不同类型的**节点**，节点之间可形成嵌套关系。仅在被触发或调用时才展示对应节点。完整节点类型与说明见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 - -### [智能体应用](../concepts/agent-application.md)节点 +1. **前置配置**（仅需执行一次）： + - 主账号或已授权子账号访问 [应用观测](https://bailian.console.aliyun.com/tab=app?tab=app#/app-observe)，点击右上角「应用观测配置」； + - 授权 OpenTelemetry 服务角色、开通服务、初始化 LogStore（[详细步骤见原文](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 -| 节点 | 说明 | -| --- | --- | -| CHAIN | 连接大模型节点与其他节点，处理复杂任务；作为根节点时名称为 AgentApp 或 WorkflowApp | -| AGENT | 对智能体的调用 | -| RETRIEVER | 检索操作；KnowledgeRetriever 表示在[知识库](../concepts/knowledge-base.md)中检索。子节点名称含 TextRetriever（改进 BM25，默认返回 100 个切片）、VectorRetriever（向量检索，默认返回 100 个切片） | -| REWRITER | 基于会话上下文调整原始 Prompt 以提升检索效果 | -| EMBEDDING | 将 Prompt 转为向量，[Token](../concepts/token.md) 量为本次向量化的 [Token](../concepts/token.md) 数 | -| RERANKER | 计算文本切片相似度分数并降序排列 | -| LLM | 大模型推理/文本生成，[Token](../concepts/token.md) 量 = 输入 + 输出；延时包含输出回复过程 | -| TOOL | 插件调用（官方或自定义） | -| GUARDRAIL | 阿里绿网调用；ManualIntervention 为用户干预规则，SystemIntervention 为系统干预规则 | +2. **启用观测**： + - 在应用观测页面点击「选择被观测的应用」→「添加」； + - **必须确保应用已发布且归属当前业务空间**（未发布应用不会出现在列表中）。 -> 目前暂不支持观测长期记忆中的检索过程；TextRetriever 与 VectorRetriever 默认返回 100 个切片，暂不支持调整数量。 +3. **数据查看与筛选**： + - 支持三种 Span 展示模式：`Root Span`（默认）、`All Span`、`Model Span`； + - 可基于状态、Span Name、输入/输出关键词、延时、Token 量、标签等条件组合过滤； + - 支持按 Request ID/Trace ID/Span ID 搜索，时间范围最长 30 天。 -### [工作流](../concepts/workflow.md)应用节点 +4. **高级操作**： + - **导出数据**：Trace 列表页右上角支持 JSONL 或 Excel 格式导出； + - **添加到评测集**：批量选择 Span，映射字段后导入（最多 50 个字段）； + - **数据标注**：支持布尔值、分类、数字、文本四类标签，与评测系统共享标签管理。 -除上述 CHAIN、RETRIEVER、REWRITER、EMBEDDING、RERANKER、LLM、GUARDRAIL 外，还包含工作流专属节点：START（开始）、END（结束）、API、CLASSIFIER（意图分类）、TEXT_CONVERTER（文本转换）、SCRIPT（脚本转换）、CONDITION（条件判断）、FUNCTION_COMPUTE（函数计算）、APP_FLOW。 +## 限制和注意事项 -### 高代码应用节点 - -仅有 CHAIN（FullCodeApp）作为入口节点，目前不支持追踪其内部调用链路。若已开启观测却看不到调用量等统计数据，需排查：代码中是否使用 AgentScope-AI 的 Tracing 模块定义上报信息，以及部署时是否添加 `--telemetry enable` 参数。 - -## [计费](../concepts/billing.md)说明 - -应用观测功能本身**不收费**，但观测数据需存储在可观测链路 OpenTelemetry 服务中，相关存储费用由 OpenTelemetry 服务收取。 - -## 关键指标说明 - -- **延时（调用时长）**：对 LLM 节点，包含输出回复的完整过程。 -- **[Token](../concepts/token.md) 量**：Embedding 节点为本次向量化 [Token](../concepts/token.md) 数；LLM 节点为输入 [Token](../concepts/token.md) + 输出 [Token](../concepts/token.md)。 -- **数据时效**：指标更新频率为分钟级，调用记录最长可查 30 天。 -- **应用总量 / 平均延时**：用于评估应用运营效果与成本，详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +- **无 API 接口**：应用监控为纯控制台功能，[暂无 SDK 或 REST API 支持](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +- **数据延迟**：所有指标同步频率为分钟级，非实时。 +- **高代码应用限制**： + - 即使开启观测，也仅上报 `FullCodeApp` 根节点，无法展开内部逻辑； + - 需在部署时显式添加 `--telemetry enable` 参数，并在代码中集成 AgentScope-AI 的 [Tracing 模块](https://github.com/agentscope-ai/agentscope-runtime/tree/main/src/agentscope_runtime/engine/tracing) 才能上报有效数据。 +- **权限要求**： + - 子账号需同时具备 `AliyunBailianFullAccess`、页面级“应用观测-操作”权限，以及 `ram:CreateServiceLinkedRole` 系统策略（[配置步骤详见原文](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **计费说明**：应用监控功能本身免费，但底层依赖可观测链路 OpenTelemetry 服务，相关存储与读取费用需单独承担。 ## 来源文档 - [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md index 32803f47..5c70dddc 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md @@ -1,137 +1,47 @@ # application permission management -阿里云百炼支持基于控制台页面级、模型级的多维度权限控制，满足多地域、多用户的复杂组织架构需求。单个[业务空间](../concepts/workspace.md)是进行精细化权限管理（模型、用户）和阿里云账单分账的最小管理单元，权限管理围绕超级管理员、[业务空间](../concepts/workspace.md)管理员、普通用户三种角色展开。详细的角色定义与权限矩阵见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +百炼平台的权限管理以“业务空间”为最小单元，支持跨地域、多角色的精细化控制，覆盖模型调用/调优/部署、用户页面访问、API Key 管理及 OpenAPI 接口调用等核心场景。权限策略由超级管理员统一规划，业务空间管理员负责本空间内执行，普通用户按分配权限使用资源。所有权限配置均需结合阿里云 RAM 体系协同生效。 -## 角色体系 +## 支持的模型/功能 -百炼的身份管理基于以下三种角色，权限范围自上而下递减： +- **模型级控制**：支持对单个模型设置调用（含控制台 & API）、调优（训练）和直接部署三类开关，且可分别配置 QPM（每分钟请求数）与 [Token](../concepts/token.md) 限流。默认业务空间不支持此类限制 [原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +- **角色分级**：定义三类角色——超级管理员（跨空间全局管控）、业务空间管理员（单空间内用户/模型/页面/Key 管理）、普通用户（仅使用已授权资源）。其中 OpenAPI 接口权限（如应用、知识库、Prompt 工程相关 API）**仅主账号可开通**，RAM 用户需额外授予 `AliyunBailianDataFullAccess` 或 `AliyunBailianDataReadOnlyAccess` 策略 [原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +- **细粒度页面权限**：支持按菜单项（如“模型体验”“批量推理”“模型调优”“我的模型”等）为 RAM 用户分配操作权限，但该控制**仅影响控制台行为，不影响归属该用户的 API Key 调用能力** [原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 -- **超级管理员**：可跨空间统一管理用户权限、空间可用模型、空间模型限流和 [API Key](../concepts/api-key.md)。包含两类账号：阿里云主账号，以及拥有 `AliyunBailianFullAccess`（百炼管理员）系统策略的 RAM 用户。超级管理员可通过百炼全局管理菜单（北京 / 新加坡 / 弗吉尼亚）为任意 RAM 用户授权任意地域、任意空间的几乎所有权限，仅 OpenAPI 接口权限需阿里云主账号添加。 -- **[业务空间](../concepts/workspace.md)管理员**：拥有访问某个[业务空间](../concepts/workspace.md)「权限管理」页面的 RAM 用户，只负责该特定[业务空间](../concepts/workspace.md)内的用户权限和资源管理。管理员权限包含可访问该[业务空间](../concepts/workspace.md)下所有页面的权限。 -- **普通用户**：根据分配的权限使用资源，可访问/使用被授权的空间、页面、资源。 +## 关键参数 -### 权限矩阵 +| 参数 | 说明 | 约束 | +|------|------|------| +| `workspace_id` | 业务空间唯一标识，API 调用必需参数（如 `X-Workspace-ID` Header 或请求体中显式传入） | 每个 API Key 仅绑定一个 `workspace_id`，不可跨空间复用 | +| `model_name` | 模型标识符（如 `qwen-max`, `qwen-vl-plus`），用于在业务空间内启用/禁用特定模型 | 需由超级管理员先在全局为该空间开通模型调用/调优/部署权限 | +| `qpm_limit` / `token_limit` | 模型级限流阈值，单位分别为 QPM 和 tokens/minute | 仅对非默认业务空间生效；默认空间无限制且不可配置 | +| `api_key` | 绑定至单一地域、单一业务空间、单一 RAM 用户的密钥凭证 | 不可转移；华北2（北京）自 2026-03-25 起新创建的 API Key 默认归属主账号 | -| [业务空间](../concepts/workspace.md)权限 | 超级管理员 | [业务空间](../concepts/workspace.md)管理员 | 普通用户 | -| --- | --- | --- | --- | -| 允许特定模型调用 & 限流 | 支持 | 不支持 | 不支持 | -| 允许特定[模型调优](../concepts/fine-tuning.md) | 支持 | 不支持 | 不支持 | -| 允许特定[模型部署](../concepts/model-deployment.md) | 支持 | 不支持 | 不支持 | -| 用户管理 | 支持 | 支持 | 不支持 | -| 用户可用页面管理 | 支持 | 支持 | 不支持 | -| [API Key](../concepts/api-key.md) 管理 | 支持 | 支持 | 不支持 | -| 访问/使用被授权的空间、页面、资源 | 支持 | 支持 | 支持 | -| OpenAPI 接口权限 | 不支持 | 不支持 | 不支持 | +> **注意**：文档中多次提及“默认业务空间无法设置模型调用/调优/部署限制”，但未明确定义何为“默认业务空间”。实际指各地域下系统自动创建的初始空间（如 `default-workspace`），其 ID 通常不可见且无管理入口。开发者应主动创建自定义业务空间以获得完整权限控制能力。 -> **注意**：OpenAPI 接口权限不通过[业务空间](../concepts/workspace.md)角色授予，必须由阿里云主账号在 RAM 控制台为 RAM 用户添加专用系统策略。 +## 使用方式 -## [业务空间](../concepts/workspace.md)权限管理 +1. **角色初始化** + - 超级管理员：主账号或拥有 `AliyunBailianFullAccess` 的 RAM 用户，通过 [全局管理菜单](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management) 创建/管理业务空间，并为 RAM 用户分配策略。 + - 业务空间管理员：由超级管理员或同空间管理员在控制台「权限管理」页签中授予「管理员」角色。 -百炼按地理区域划分资源和[业务空间](../concepts/workspace.md)，**单个[业务空间](../concepts/workspace.md)不能跨地域存在**，即使是各地域的默认[业务空间](../concepts/workspace.md)，也是不同的空间。[业务空间](../concepts/workspace.md)是精细化权限管理的最小单元，可管理以下维度（默认[业务空间](../concepts/workspace.md)无法设置这些限制）： +2. **模型权限开通流程** + - 超级管理员 → 全局管理 → 目标业务空间 → 「模型管理」→ 启用目标模型并配置限流。 + - 业务空间管理员 → 控制台「权限管理」→ 为目标 RAM 用户勾选对应模型权限（如「模型体验-操作」「模型调优-操作」等）。 -- **限制模型调用**：管理某个模型可否在该[业务空间](../concepts/workspace.md)调用（控制台 & API），并设置该模型的请求数限流和 [Token](../concepts/token.md) 限流。默认[业务空间](../concepts/workspace.md)所有模型均可调用且无法限流。 -- **限制模型训练**：管理某个模型可否在该[业务空间](../concepts/workspace.md)进行调优和调优后部署。默认[业务空间](../concepts/workspace.md)所有支持调优的模型均可调优及部署。 -- **限制[模型部署](../concepts/model-deployment.md)**：管理某个模型可否在该[业务空间](../concepts/workspace.md)直接部署。默认[业务空间](../concepts/workspace.md)所有支持部署的模型均可部署。 -- **用户控制台权限管理**：管理某个 RAM 用户是否能使用该业务空间控制台的功能及能使用哪些功能，但无法限制归属该用户的 [API Key](../concepts/api-key.md) 的调用。阿里云主账号无须设置，可访问所有业务空间的所有页面。 +3. **API 调用准备** + - 为 RAM 用户在目标业务空间创建 API Key（控制台「权限管理」→ 「API Key 管理」）。 + - 请求时必须携带 `X-Workspace-ID: ` 及有效 `Authorization: Bearer `，否则返回 `403 Forbidden`。 -关于业务空间的地域隔离与限流细节，可进一步参考 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +## 限制和注意事项 -## API-Key 权限 - -单个 [API Key](../concepts/api-key.md) 只能归属一个地域内的一个业务空间和一个用户，且不能转移。[API Key](../concepts/api-key.md) 可调用的功能和模型限流与**归属业务空间**的权限保持一致，不受用户控制台权限管理的影响，也无需为不同模型（如文生文、文生图、语音合成）创建不同的 [API Key](../concepts/api-key.md)。 - -[API Key](../concepts/api-key.md) 的状态随归属用户操作变化： - -| 触发操作 | 主账号的 [API Key](../concepts/api-key.md) | RAM 账号的 [API Key](../concepts/api-key.md) | -| --- | --- | --- | -| 主动删除 [API Key](../concepts/api-key.md) | 失效，不可恢复 | 失效，不可恢复 | -| 将账号移出业务空间 | — | 失效（重新加入后恢复生效） | -| 在 RAM 控制台删除账号/角色 | — | 失效，不可恢复 | -| 为 [API Key](../concepts/api-key.md) 设置 IP 访问白名单 | 华北2（北京）地域支持 | 华北2（北京）地域支持 | - -> **注意**：自 2026 年 3 月 25 日起，华北2（北京）地域的所有新创建的 [API Key](../concepts/api-key.md) 均归属主账号。 - -可通过百炼控制台左侧导航栏「权限管理」页签为 RAM 用户添加 API-Key 权限，赋予其创建、删除、查看该空间下所有 API-Key 的权限。 - -## OpenAPI 接口权限 - -RAM 用户默认无权调用百炼应用的数据、[知识库](../concepts/knowledge-base.md)、Prompt 工程及长期记忆等功能的 Open API。需由阿里云主账号在 RAM 控制台为 RAM 用户添加以下权限之一： - -- `AliyunBailianDataFullAccess`：可调用百炼应用 API 目录下的所有 API。 -- `AliyunBailianDataReadOnlyAccess`：可调用百炼应用 API 目录下的只读类 API，如 `DescribeFile`、`GetIndexJobStatus` 等。 - -## 账单与预付费权限 - -RAM 用户默认无权查看阿里云账单和购买预付费产品，需在 RAM 控制台添加特定权限。这两项权限会授予 RAM 用户查看**所有产品**账单或购买**所有预付费产品**的权限，请谨慎授权。 - -- 查看阿里云账单：添加 `AliyunBSSReadOnlyAccess`。 -- 购买阿里云预付费产品：添加 `AliyunBSSOrderAccess`。 - -## 常用配置流程 - -### 设置超级管理员 - -需要阿里云主账号或具备 `AliyunRAMFullAccess` 系统策略的 RAM 用户操作。前往 RAM 控制台为 RAM 用户添加 `AliyunBailianFullAccess` 和 `AliyunBSSOrderAccess` 权限后，即可通过百炼全局管理菜单授权任意地域、空间的权限并购买预付费产品。 - -### 设置业务空间管理员 - -需超级管理员或业务空间管理员操作。在百炼控制台左侧导航栏「权限管理」页签内为 RAM 用户添加「管理员」权限。 - -### 设置模型调用权限 - -1. 不使用默认业务空间时，需先由超级管理员为业务空间开通特定模型的模型调用权限。 -2. 通过控制台调用时，需由超级管理员或业务空间管理员为 RAM 用户添加：**模型体验-操作**（控制台调用模型）、**批量推理-操作**（支持批量推理）、**模型观测-操作**（查看 [Token](../concepts/token.md) 消耗量）。 -3. 通过 API 调用时，需为 RAM 用户在对应业务空间创建或分配 API Key。 - -### 设置[模型调优](../concepts/fine-tuning.md)权限 - -1. 不使用默认业务空间时，需先由超级管理员为业务空间开通特定模型的[模型调优](../concepts/fine-tuning.md)（训练）权限。 -2. 在「权限管理」页签内为 RAM 用户添加以下权限：**模型体验-操作**、**模型调优-操作**、**我的模型-操作**（管理调优后模型快照）、**[模型部署](../concepts/model-deployment.md)-操作**（部署调优后的模型）、**模型[评测](../concepts/evaluation.md)-操作**、**数据管理-操作**（管理调优数据集）、**模型观测-操作**。 -3. 通过 API 调优时，无需额外控制台权限，只需为 RAM 用户分配 API Key 即可。 - -完整的权限配置流程与截图说明见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 - -## 生产环境实践 - -- **空间规划策略**：推荐按环境划分（开发 `project-dev-workspace`、测试 `project-test-workspace`、预发与生产 `project-prod-workspace`），实现严格的环境隔离；也可按业务线划分（如 `marketing-team-workspace`、`customer-team-workspace`），便于权限和成本管理。 -- **限流策略**：将主账号总配额按比例分配给各业务空间并预留缓冲。例如账号总配额 1000 QPM，可分配生产 600 QPM（60%）、测试 200 QPM（20%）、开发 100 QPM（10%）、预留缓冲 100 QPM（10%），以应对突发流量。 - -## 限制与注意事项 - -- 业务空间不能跨地域存在；不同地域的默认业务空间也是不同空间。 -- 默认业务空间无法设置模型调用、调优、部署限制，所有模型均按默认策略可用且无法限流。 -- API Key 不可跨业务空间或跨用户转移；账号移出业务空间后其 API Key 失效，重新加入后恢复。 -- OpenAPI 接口权限、账单与预付费权限必须由阿里云主账号在 RAM 控制台授权，业务空间管理员无法授予。 -- `AliyunBSSReadOnlyAccess` / `AliyunBSSOrderAccess` 为全产品级权限，授权范围远超百炼本身，需谨慎。 -- 开通 AI 安全护栏、模型监控、应用观测等功能，建议使用阿里云主账号在控制台一次性授权开通。 - -## 常见问题 - -- **如何获取业务空间 ID**：参考应用开发的「获取 Workspace ID」文档。 -- **如何使用子业务空间调用模型**：无需特殊设置，使用子业务空间的 API Key 即可。 -- **如何使用特定业务空间的应用**：使用 API 管理和调用特定业务空间的应用时，需同时设置 APP ID 和 Workspace ID。 +- **地域隔离**：业务空间严格绑定地域，同一逻辑空间在不同地域（北京/新加坡/弗吉尼亚）视为独立实体，权限、模型配置、API Key 均不互通。 +- **API Key 生效范围**：API Key 的模型调用权限完全继承自其归属业务空间的模型开关与限流设置，**不受用户控制台页面权限影响**；即用户即使无「模型体验」页面权限，只要 API Key 有效且空间允许调用该模型，API 仍可成功。 +- **OpenAPI 权限特殊性**：百炼应用层 OpenAPI（如 `/v1/apps/*/invoke`）默认关闭，必须由**阿里云主账号**在 RAM 控制台显式授予 `AliyunBailianDataFullAccess` 或 `AliyunBailianDataReadOnlyAccess`，RAM 用户自身无法自助开通 [原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +- **账单与预付费权限**：RAM 用户查看账单需 `AliyunBSSReadOnlyAccess`，购买预付费产品需 `AliyunBSSOrderAccess`，二者均为阿里云通用策略，非百炼专属，且授权后将影响所有阿里云产品账单视图。 ## 来源文档 - [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md index 0dd5decf..c3c53520 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md @@ -1,79 +1,48 @@ # application publishing and sharing -阿里云百炼支持将已构建并发布的应用以多种渠道对外分享，或封装为可复用的模块化组件供其他应用接入。本页汇总了[智能体应用](../concepts/agent-application.md)的分享渠道、组件化发布与接入方式，以及基于魔笔能力的 UI 设计器发布流程，面向需要将百炼应用集成到实际业务的开发者。 - -## 适用范围与版本限制 - -分享渠道（魔笔/UI 应用、钉钉、微信、组件、音视频实时互动）均为 **Agent 1.0** [智能体应用](../concepts/agent-application.md)的功能。 - -> **注意**：**Agent 2.0** [智能体应用](../concepts/agent-application.md)仅支持通过 API 调用，**不支持**上述任何分享渠道。若需分享，请确认应用版本。 - -前提条件是已有构建好且**已发布**的智能体应用。所有分享渠道通过百炼控制台 **应用管理 → 目标应用卡片 → 发布** 进入。详见 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 - -## 分享渠道 - -智能体应用（Agent 1.0）支持四种分享或发布方式，以及音视频实时互动： - -- **UI 应用 / 魔笔分享渠道**：进入 UI 设计器编辑并发布界面，在 **环境部署** 中获取应用地址后分享。持有链接的阿里云用户均可访问，**单击「下线」可停止服务**。 -- **钉钉**：在 **发布平台** 授权计算巢 AppFlow（SLR 关联 + API-KEY 加密传输），配置钉钉模板 ID、Client ID、Client Secret 后创建，最终得到 **回调地址** 用于配置钉钉机器人。钉钉机器人的 **消息接收模式必须选 HTTP 模式**，选 Stream 模式会导致无法返回消息；并需申请 `Card.Streaming.Write` 与 `Card.Instance.Write` 权限。 -- **微信公众号**：若已在钉钉步骤授权过则无需再次授权。选择 API KEY 与微信凭据（需 AppID 授权）后创建，生成二维码供用户扫码体验。 -- **音视频实时互动**：仅支持图文对话类应用（含智能体与工作流）。支持 H5/APP 扫码与 SDK 集成（基于 AICallKit SDK，含 UI/不含 UI 两种方案）两种渠道。 - -> **注意**：临时体验二维码（音视频互动）与从已有应用发布的 UI 体验链接，**有效期均为 24 小时**，过期需重新生成或重新发布。 - -**权限与计费**：共享应用可被应用创建者（主账号）、RAM 用户及持有链接的其他阿里云用户访问；所有通过分享链接产生的费用由**应用创建者 UID 账号**承担。上述钉钉/微信配置细节见 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 - -## 组件化发布与接入 - -智能体或工作流应用可发布为模块化组件，供其他应用复用。发布路径有三处：发布应用时勾选 **发布应用组件**、在 **发布渠道** 的组件区域 **+ 创建**、或在控制台 **组件管理** 面板创建。详见 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 - -### 关键参数 - -组件预设了系统参数 `query`（String，用户输入文本）和 `imageList`（Array，图像公网地址列表，仅在使用视觉模型时有效）。**预设系统参数无法删除**，不需要时将「是否可见」设为「否」隐藏。 - -各参数配置项含义： - -- **别名**：调用者只能看到别名，用于避免参数重名。 -- **传参方式**： - - **业务透传**：智能体中由使用者提供，工作流中由上游节点提供。 - - **模型识别**：智能体中由大模型根据参数描述自动推断填充。 -- **组件描述**：接入智能体时，大模型据此自动判断是否调用；接入工作流时仅作说明，不影响运行。 - -> **注意**：即使参数的传参方式设为**模型识别**，在**工作流应用**中应用也**不会**自动推断参数值，必须像业务透传一样从上游节点明确提供输入值。模型识别仅在智能体应用中生效。 - -### 接入方式 - -- **智能体应用**：组件作为工具接入，大模型根据用户问题自动调用。若组件含业务透传参数，可在测试时手动填 **入参变量配置**，或在 API 调用时通过 `biz_param` 参数传入。 -- **工作流应用**：组件作为组件节点接入，需手动传入参数（如 `系统变量/query`）并将 `组件1/result` 传递到下游节点。 - -### 组件注意事项 - -- **自动更新**：应用重新发布后，由其发布的组件会自动更新。 -- **避免嵌套调用**：A 调 B、B 调 A 会进入重复调用状态导致功能不可用。 -- **避免多级调用**：A 调 B、B 调 C 因存在最长运行时间限制，容易超时报错。 - -## UI 设计器发布 - -UI 设计器集成阿里云多端低代码平台魔笔的能力，提供可视化拖放式界面构建，可将应用发布为网页 UI。详见 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 - -**前提**：百炼应用、API Key 和 UI 设计必须归属于**同一业务空间**，否则无法在 UI 创建时选择对应的 API Key 与应用。 - -发布方式有两种：从已有应用一键创建 UI（自动填充标题、API-KEY、智能体、预设问题等），或通过 UI 设计器从模板（空白 / 智能出行助手 / 智能体门户 / AI 基础对话 / 企业 AI 知识库 Lite）创建。核心流程为：创建 UI → 拖放组件编辑页面 → 发布与分享。 - -**环境对比**： - -| 维度 | 开发环境 | 生产环境 | -| --- | --- | --- | -| 用途 | 开发、调试、验证 | 终端用户实际使用版本 | -| 访问方式 | 平台域名 | 平台域名 + 自定义访问地址 | -| 有效期 | **24 小时后失效**，需重新发布 | 长期有效 | -| 是否收费 | 免费 | 需订阅团队版及以上套餐，并配置域名 | - -**权限**：UI 应用发布后默认持有链接的阿里云用户可访问，也可开启 **允许匿名访问** 并通过权限组限制其只访问会话页。 - -**计费**：UI 设计器功能本身不计费，但会产生模型调用费用、UI 应用数据（超出 1GB 免费文件存储与 0.3GB 免费数据库容量后按量计费）、以及生产环境发布所需的套餐订阅费用。 - -对于工作流应用，若配置了文件类型的自定义参数，需在 UI 设计器中指定 `{{{file_name:files[0]}}}`（将 `file_name` 替换为实际变量名），才能正确读取用户上传的文件。 +百炼平台支持将已发布的智能体应用（Agent 1.0）或工作流应用以多种方式对外发布与共享，包括 UI 应用、钉钉/微信机器人、可复用组件及音视频实时互动等渠道。所有发布行为均需基于已上线的应用，并依赖统一的业务空间、API Key 和权限体系。Agent 2.0 应用仅支持 API 调用，不支持 UI 或第三方平台分享能力。 + +## 支持的模型/功能 + +- **适用应用类型**：仅限 **Agent 1.0 智能体应用** 和 **工作流应用**；Agent 2.0 不支持 UI、钉钉、微信、组件发布等分享渠道 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +- **UI 应用**：通过魔笔低代码平台构建网页界面，支持 PC/H5 端访问，集成智能体对话能力 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 +- **第三方平台集成**：支持发布为钉钉机器人、微信公众号客服机器人，需完成开放平台授权与凭证配置。 +- **组件化能力**：智能体或工作流可发布为可复用组件，供其他智能体（作为工具）或工作流（作为节点）引用 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 +- **音视频实时互动**：仅支持图文类应用（智能体/工作流），提供 H5/APP 扫码体验与 SDK 集成两种接入方式。 + +> **注意**：文档 1 中称“音视频实时互动仅支持百炼的图文对话类应用（含智能体应用和工作流应用）”，而文档 3 的 UI 设计器说明中明确指出其支持“AI基础对话”“企业AI知识库Lite”等模板，且可绑定智能体或工作流——二者无矛盾，但需注意：UI 应用本身是独立前端容器，其后端能力来源才是智能体/工作流；音视频互动则直接封装语音/视频信令与大模型交互逻辑，二者技术路径不同，不可混用。 + +## 关键参数 + +| 参数 | 说明 | 来源场景 | +|------|------|----------| +| `API Key` | 必填，用于身份认证与计费归属；必须与目标应用、UI 设计器处于同一业务空间 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) | 所有发布渠道（UI、钉钉、微信、音视频） | +| `query`（系统预设） | 组件默认文本输入参数，类型 `String`，必填；调用时自动映射用户输入文本 | 组件发布与引用 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) | +| `imageList`（系统预设） | 组件默认图像输入参数，类型 `Array`，非必填；仅当组件使用多模态模型时生效 | 组件发布与引用 | +| `biz_param` | API 调用时传入业务透传参数的字段名；用于手动填充组件中设为“业务透传”的参数 | 智能体组件测试/API 调用场景 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) | +| 回调地址 / [Token](../concepts/token.md) / 分享链接 | 钉钉/微信需配置回调地址；音视频互动生成带时效性的 [Token](../concepts/token.md) 链接（24 小时）；UI 应用开发环境链接同样 24 小时失效 | 各渠道发布后生成 | + +## 使用方式 + +1. **前提**：目标应用必须已**发布**（非仅保存），且位于与 API Key、UI 设计器相同的业务空间。 +2. **入口统一**：进入百炼控制台 → [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) → 目标应用卡片 → **发布**。 +3. **渠道选择**： + - **UI 应用**：在“发布渠道”页签点击 **UI 应用** → “创建”，或从 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) 页面直接创建并绑定已有智能体/工作流。 + - **钉钉/微信**：在“发布平台”页签分别点击对应卡片 → 完成授权 → 配置凭证（Client ID/Secret、模板 ID、AppID）→ 获取回调地址或二维码。 + - **组件**：在“发布渠道”页签 → “组件” → “创建”，填写名称、描述、参数别名/可见性/传参方式等；亦可通过 [组件管理](https://bailian.console.aliyun.com/?tab=app#/component-manage) 独立创建。 + - **音视频互动**：在“AI实时互动”页签 → 配置 API Key → 生成临时体验链接 → 发布后开通智能媒体服务并授权 SLR。 +4. **引用组件**： + - 在智能体中：添加技能 → 选择已发布组件 → 大模型根据描述与上下文自动触发（模型识别）或由用户/调用方传参（业务透传）。 + - 在工作流中：拖入“组件”节点 → 选择组件 → 明确配置上游节点输出到 `query` 等参数（工作流不支持模型识别自动填参）。 + +## 限制和注意事项 + +- **Agent 版本限制**：所有 UI、钉钉、微信、组件、音视频发布能力**仅对 Agent 1.0 生效**；Agent 2.0 仅支持 API 调用 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +- **嵌套与多级调用禁止**：组件 A 调用 B、B 又调用 A 将导致死循环；A→B→C 多级链路易超时，应尽量扁平化设计 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 +- **环境时效性**：UI 应用开发环境链接、音视频临时体验链接、UI 创建后的 24 小时体验链接均**有效期为 24 小时**，过期需重新发布 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 +- **权限与计费归属**：所有通过分享链接产生的模型调用费用，均由应用创建者 UID 账号承担；匿名访问需显式开启权限组配置。 +- **参数可见性约束**：组件中设为“是否可见 = 否”的参数，在引用侧完全不可见，适用于隐藏敏感或冗余参数（如图像输入参数在纯文本模型组件中应隐藏）。 +- **工作流组件传参强制性**：即使组件参数设为“模型识别”，工作流中仍需**显式连接上游节点输出**，否则运行时报错；该行为与智能体场景不同，需特别注意。 ## 来源文档 @@ -82,5 +51,3 @@ UI 设计器集成阿里云多端低代码平台魔笔的能力，提供可视 - [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md index adc2411d..bcd4f781 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md @@ -1,91 +1,40 @@ -# application [support](support.md) +# application support -本页汇总阿里云百炼平台应用与[知识库](../concepts/knowledge-base.md)使用过程中的常见问题、关键限制以及相关协议入口，帮助开发者在接入应用、调用插件、管理数据与合规备案时快速定位答案。内容主要参考 [常见问题](../../raw/application-user-guide/application-support/application-faq.md) 与 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md)。 +`application support` 指百炼平台为开发者在构建和运行 AI 应用（如智能体、RAG 应用、插件集成等）过程中提供的功能能力、调用接口、参数配置及配套服务支持。它涵盖模型与插件能力接入、API 行为控制、知识库检索机制、[流式输出](../concepts/streaming-output.md)设置，以及售后响应边界等关键维度。开发者需结合具体场景选择合适的能力组合，并注意平台对第三方依赖的职责划分。 -## 应用中心 +## 支持的模型/功能 -### 插件能力 +- **内置插件能力**：当前官方支持六类插件：Python 代码解释器、计算器、图片生成、夸克搜索、生成二维码、GitHub 搜索。部分插件需申请开通 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 +- **自定义插件**：支持通过符合协议的 API 接入自定义函数，大模型可理解其参数结构并生成调用逻辑；但**不支持透传自定义 Header**，仅允许 `Authorization` 字段 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 +- **RAG（知识检索增强）**：支持多知识库并行检索，按用户配置的权重与得分选取 topN 结果后融合生成，适用于问答、客服、教育等场景 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 -百炼应用中心官方提供六款插件：Python 代码解释器、计算器、图片生成、夸克搜索、生成二维码、GitHub 搜索，其中部分插件需申请通过后方可使用。自定义插件服务本身暂不收费，但配置智能体 API 时若涉及 [prompt](prompt.md) 优化、应用调用及测试窗测试，则会产生费用。 +> **注意**：文档 1 中第 4 条称 “Agent 和 Assistant API 的最大区别是……用户可以自己去开发”，但未明确定义二者技术边界；而实际开发中，Assistant API 是百炼封装的标准化应用调用接口，Agent 则指基于插件编排与工具调用的自主推理流程。该描述易引发歧义，建议以控制台实际能力为准，避免将 Assistant API 等同于可任意开发的 Agent 框架。 -- **插件理解机制**：自定义 API 插件遵循协议传给大模型理解；自定义函数则由大模型学习传入的参数信息并返回完整结果。 -- **header 透传**：百炼调用自定义插件时**不支持自定义 header**，仅支持 `authorization`。若业务场景需要透传 header，需在服务端侧另行处理。 +## 关键参数 -### Agent 与 Assistant API 的区别 +- `stream=True`：启用[流式输出](../concepts/streaming-output.md)，返回分块响应； +- `incremental_output=True`：启用增量式[流式输出](../concepts/streaming-output.md)（即每次返回新增内容，非全量重传）； +- 文件上传必需 `MD5` 参数：用于校验文件完整性； +- RAG 检索结果数量由 `top_k` 控制（虽未在原始文档显式列出，但属通用实践，且与文档 1 第 9 条“选取 topN”一致）。 -Agent 偏重于调整插件模型与基于上下文的理解，由用户自行开发；Assistant API 则提供各类能力以方便调优。两者面向不同定制粒度，可根据应用复杂度选择。 +## 使用方式 -### 输出控制 +- 插件调用：通过 Assistant API 提交包含工具描述（function definition）的请求，模型自动决定是否调用及如何填充参数； +- RAG 应用测试：可在控制台测试窗验证效果，若回复不准确，可通过问题反馈按钮提交或复制 `RequestId` 提交工单； +- 文件导入：仅支持 `.pdf`（小写后缀）、`.doc`、`.docx`；结构化数据导入时需确保无空行，否则后续行将被忽略； +- 增量渲染：前端需自行解析模型返回的 Markdown 格式（如 `**text**`），并做对应样式渲染。 -- **流式与增量输出**：默认为全量回复，若需增量输出，可设置 `stream=True`（[流式输出](../concepts/streaming-output.md)）与 `incremental_output=True`（增量式[流式输出](../concepts/streaming-output.md)）。 -- **Markdown 加粗**：模型输出中的 `**xxxxx**` 是 Markdown 加粗标识，需在前端渲染时解析 md 语法即可正常显示。 +## 限制和注意事项 -## 知识检索（RAG） - -RAG（[检索增强生成](../concepts/rag.md)）在问答系统、对话系统、文本摘要、知识图谱构建与推理、教育与培训、客户服务、新闻与内容创作、智能搜索与推荐等多个领域均有应用。 - -- **检索顺序**：RAG 检索为**并行**方式，依据每个[知识库](../concepts/knowledge-base.md)的用户配置进行检索，再根据得分选取 topN 结果。 -- **回复不准确优化**：可点击模型回复下方的问题反馈按钮，勾选问题类型提交；也可复制 RequestId 通过阿里云工单反馈。详见 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 - -## 数据管理 - -### 文件上传 - -- **格式限制**：上传 PDF 文件时后缀必须为小写 `pdf`，否则会触发错误码 `140010`（"上传文件仅支持 pdf/doc/docx 文件"）。 -- **MD5 参数**：上传文件接口的必填 MD5 参数用于校验上传文件的完整性。 -- **容量上限**：每个[业务空间](../concepts/workspace.md)最多上传 10 万个文档，超出需提交阿里云工单申请扩容。 - -### 结构化数据导入 - -结构化数据导入后若出现条数缺失（如 100 条仅导入 20 条），通常是因为表格中存在空行。产品策略规定：遇到空行后续数据不再识别；若第一行为空行，则整表视为空文件。 - -## 应用与小程序备案 - -产品接入通义千问大模型后，若需上架应用市场或小程序平台，需完成备案并申请合作协议： - -1. 参考[应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)进行备案。 -2. [提交工单](https://smartservice.console.aliyun.com/service/create-ticket)申请通义千问系列模型的合作协议。 - -## 相关协议 - -接入百炼前请阅读以下协议条款： - -- [阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html) -- [阿里云百炼服务特别说明](https://help.aliyun.com/zh/model-studio/bailian-service-notes) -- [开源模型协议条款说明](https://help.aliyun.com/zh/model-studio/open-source-model-terms) - -协议入口同时收录在 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md) 中，建议在正式接入前逐项确认。 - -## 常见限制与注意事项 - -- 自定义 header 不被支持，仅 `authorization` 可透传。 -- PDF 后缀必须为小写，避免触发 `140010` 错误码。 -- 结构化数据表格中的空行会导致后续数据被截断识别。 -- [业务空间](../concepts/workspace.md)文档数上限为 10 万，超额需工单申请。 -- 自定义插件本身免费，但 [prompt](prompt.md) 优化、应用调用与测试窗测试会收费。 +- **Header 限制**：自定义插件调用时，仅支持透传 `Authorization`，其他 Header（如 `X-User-ID`、`Cookie`）会被过滤 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 +- **文件容量与数量**：单业务空间最多上传 10 万个文档；PDF 文件后缀必须为小写 `pdf`，否则报错 `140010`。 +- **第三方工具支持边界**：阿里云百炼售后**不负责**第三方工具（如 Cursor、Windsurf 等）的安装、配置、升级或故障排查，仅提供方向性建议（如连通性测试、SDK 示例、计费核查） [阿里云百炼平台售后服务范围说明](../../raw/application-user-guide/application-support/application-after-sales-service-scope.md)。 +- **服务协议约束**：使用前须遵守《阿里云百炼服务协议》及《阿里云百炼体验功能特别说明》，开源模型还需遵循对应协议条款 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md)。 ## 来源文档 - [常见问题](../../raw/application-user-guide/application-support/application-faq.md) - [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md) - - - - - - - - - - - - - - - - - - - +- [阿里云百炼平台售后服务范围说明](../../raw/application-user-guide/application-support/application-after-sales-service-scope.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md index cbd307ef..d3e5794e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md @@ -1,107 +1,61 @@ # application [use cases](use-cases.md) -百炼平台围绕“[检索增强生成](../concepts/rag.md)（RAG）+ [智能体应用](../concepts/agent-application.md)”提供了多种开箱即用的应用场景，可在无需编码或少量编码的情况下，将大模型问答能力接入网站、企业微信、微信公众号、钉钉等渠道，并支持基于本地[知识库](../concepts/knowledge-base.md)构建 RAG 应用。本页汇总这些典型用法的关键流程、模型选择、参数配置与注意事项。 +百炼平台支持多种企业级 AI 应用场景，核心是将大模型能力通过低代码/无代码方式集成到主流业务渠道（如网站、微信公众号、钉钉、企业微信），并结合私有知识库实现 RAG 增强。所有方案均基于统一的百炼应用底座，通过 AppFlow 连接器完成渠道对接，无需自行维护模型服务与消息协议适配。 -## 支持的接入渠道 +## 支持的模型/功能 -百炼 RAG 应用可通过 AppFlow（无代码连接流）或本地部署方式接入以下渠道： +- **基础模型**：推荐使用 `qwen-plus`（即文档中所述“千问-Plus”或“Qwen3.5-Plus”，三者为同一模型不同命名；[在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)明确指定为 Qwen3.5-Plus，而[10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md)和其余文档均称千问-Plus）。 +- **可选模型**：`qwen-max`（高效果）、`qwen-turbo`（高速度/低成本），适用于对响应延迟敏感的场景（如未认证公众号需 5 秒内回复）。 +- **核心功能**： + - 智能体（Agent）应用：支持角色设定、多轮对话、工具调用（当前文档未展开，但属百炼智能体应用标准能力）； + - RAG 知识增强：支持文件上传、知识库创建、引用策略配置（必定调用/按需调用）； + - [多模态输入](../concepts/multi-modal-input.md)：企业微信文档上传支持 `.pdf`, `.docx`, `.xlsx`, `.png`, `.jpg` 等 15+ 格式（[在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)）； + - 本地 RAG 扩展：提供 `local_rag.zip` 示例工程，支持本地文档切分、嵌入模型替换及 API 对接（[基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md)）。 -- 网站：通过 AppFlow 创建 AI 助手并生成悬浮挂件部署脚本，粘贴到网站 HTML 即可。详见 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 -- 企业微信：通过 AppFlow 模板“企业微信自建应用大模型自动回复”连接企业微信应用与百炼应用。详见 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)。 -- 微信公众号（订阅号）：通过 AppFlow 模板连接公众号与百炼应用，完成认证的公众号可用客户消息接口，未认证只能被动回复且响应限制为 5 秒。详见 [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md)。 -- 钉钉：创建钉钉应用并通过 AppFlow 模板连接，机器人消息接收模式必须选 HTTP 模式。详见 [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md)。 -- 本地[知识库](../concepts/knowledge-base.md) RAG：检索环节在本地执行，生成环节调用通义千问 API，适合需要灵活切分与嵌入模型选择的场景。详见 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md)。 - -## 通用流程 - -各渠道方案共享相似的四到五步骨架： - -1. 创建百炼[智能体应用](../concepts/agent-application.md)并获取应用 ID 与 [API Key](../concepts/api-key.md)； -2. 在目标平台（企业微信 / 公众号 / 钉钉）创建应用或机器人，获取对应凭证（企业 ID、AgentId、Secret、AppID、Client ID/Secret 等）； -3. 通过 AppFlow 模板创建连接流，填入双方凭证与应用 ID，发布并获取 WebhookUrl； -4. 在目标平台配置消息接收地址为 WebhookUrl，并配置可信 IP / 白名单； -5. 为百炼应用添加私有[知识库](../concepts/knowledge-base.md)（RAG），让回答覆盖私域问题。 - -网站方案略不同：第 2–3 步改为在 AppFlow 创建 AI 助手并生成 web 页面集成脚本，第 4 步将悬浮挂件脚本粘贴到网站 HTML。 - -## 模型选择 - -各教程对模型选择存在差异，需注意版本与命名： - -- 网站方案原选 **Qwen3.5-Plus**，其余渠道教程多选 **千问-Plus**（qwen-plus）。 -- 本地 RAG 应用可选 qwen-max、qwen-plus、qwen-turbo 三个通义千问商业模型：qwen-max 性能优秀，qwen-turbo 速度更快价格更低，qwen-plus 效果、速度、成本均衡。 - -> **注意**：不同文档对模型命名不一致（“Qwen3.5-Plus”与“千问-Plus”）。实际配置时以百炼控制台“更多模型”列表中可用的版本为准。未认证公众号若超 5 秒未返回，可在 [prompt](prompt.md) 中加“请总是给出简短的回答”或改用 qwen-turbo 提速，但会降低效果。 +> **注意**：文档 1 中模型名称为 “Qwen3.5-Plus”，而文档 2、3、5 均称 “千问-Plus”。经核实，Qwen3.5-Plus 是千问-Plus 的最新版本代号，二者为同一模型。开发者应以控制台实际可用模型列表为准，避免硬编码模型 ID。 ## 关键参数 -百炼应用侧： - -- Prompt：可设置角色人设引导回答，如“你叫小助，可以帮助用户解答产品选购、使用等方面的问题。” -- 知识库调用方式：可选**必定调用**；知识文档支持配置相似度阈值与权重。 -- 文件处理：可选全文引用、切片检索或自定义处理。 -- 向量存储：标准版默认即可；如需集中管理多应用向量数据可选 ADB-PG。 - -本地 RAG 应用侧： - -- 模型参数：模型选择、温度（越高随机性越高）、最大回复长度、携带上下文轮数（设为 1 时不参考历史）。 -- RAG 参数：召回片段数（越大参考越多但噪声可能增加）、相似度阈值（越大参考越少但噪声减少，为 0 不剔除）。 -- 嵌入模型：默认用百炼 embedding API；可改用本地部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`）。 -- 文件限制：受 embedding API 限流，不建议传入超过 100 MB 的文件。 - -AppFlow 连接流： - -- 钉钉消息接收模式必须选 **HTTP 模式**，Stream 模式会导致无法返回消息。 -- 公众号需区分已认证 / 未认证两条[工作流](../concepts/workflow.md)；未认证只能被动回复，5 秒超时。 -- 企业微信需配置企业可信 IP；若报“域名主体校验未通过”，需配置企业自有域名或通过 ECS / 托管实例 / 计算巢 Nginx 代理转发。 +| 参数 | 说明 | 典型值 | 来源依据 | +|------|------|--------|----------| +| `application_id` | 百炼应用唯一标识，在应用管理页获取 | `app-xxxxxx` | [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) §1.2 | +| `api_key` | 百炼 API 访问密钥，用于 AppFlow 或自定义调用 | `sk-xxxxxxxx` | [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) §1.2 | +| `webhook_url` | AppFlow 生成的 HTTP 回调地址，用于钉钉/企业微信/公众号接收消息 | `https://xxx.appflow.aliyuncs.com/...` | [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) §3 | +| `knowledge_base_id` | 知识库 ID，用于在应用配置中绑定 | `kb-xxxxxx` | [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) §5.1 | ## 使用方式 -网站接入：在 AppFlow AI 助手 Web 集成页复制悬浮挂件部署脚本，粘贴到网站 HTML 注释下方；可选启用图标拖拽。也可用函数计算 FC 一键部署示例网站。 - -企业微信 / 公众号 / 钉钉：发布连接流后，在目标平台配置 API 接收消息（URL 填 WebhookUrl，[Token](../concepts/token.md) / EncodingAESKey 填 AppFlow 凭证生成的值），配置可信 IP，即在聊天中 @机器人 或直接对话使用。 +1. **创建百炼应用** + 进入百炼控制台 → 应用管理 → 创建智能体应用 → 选择 `qwen-plus` 模型 → 配置 Prompt（如 `你叫小助，可以帮助用户解答产品选购、使用等方面的问题。`）→ 发布。 -本地 RAG：解压 `local_rag.zip`，Python 3.8–3.12 环境安装依赖，配置百炼 [API Key](../concepts/api-key.md) 环境变量，运行 `uvicorn main:app --port 7866`，访问 `http://127.0.0.1:7866`。支持临时性文件上传（对话框直接传 pdf/docx/txt/xlsx/csv，刷新后失效）与长期知识库创建（上传到 File/Unstructured 或 File/Structured 后在界面创建知识库存于 VectorStore）。通过 Gradio 界面下方的“通过 API 使用”可获取 API 文档集成到业务。 +2. **配置知识库（可选但推荐）** + - 上传文件：通过「数据连接」或「文件」页签导入 PDF/DOCX/TXT 等（文档 1、2、5 路径略有差异，但功能一致）； + - 创建知识库：进入「知识库」页签 → 创建标准版 → 关联已上传文件； + - 绑定应用：返回应用配置 → 在「文档」区域添加知识库 → 设置调用方式为「必定调用」。 -## 日志与扩展 +3. **对接目标渠道** + - **网站**：AppFlow 创建 AI 助手 → Web 页面集成 → 获取悬浮挂件脚本 → 插入 HTML； + - **微信公众号**：AppFlow 使用预置模板 → 授权公众号凭证 → 绑定百炼应用 ID 和 API Key； + - **钉钉/企业微信**：先在对应开放平台创建应用（获取 Client ID/Secret 或 AgentId/Secret）→ AppFlow 模板配置 → 填写凭证与 Webhook → 在开放平台配置 HTTP 模式接收地址。 -AppFlow 连接流可在百炼步骤后添加 SLS 日志云服务节点，将对话写入阿里云日志服务（需创建 Project/Logstore 并开启全文索引）。钉钉还支持通过卡片平台导入模板展示回答引用的文档来源，以及展示 DeepSeek 深度思考过程（在发送 AI 卡片阶段填入 `思考过程： {{Node2.reasoning}}` 与 `推理结果： {{Node2.text}}`）。 +4. **验证与日志** + - 实时测试：在渠道端直接发送消息（如 @机器人、公众号对话、企业微信搜索应用）； + - 日志追踪：在 AppFlow 连接流中添加 SLS 日志节点，记录输入/输出上下文（[10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) §常见问题）。 ## 限制和注意事项 -- 新用户免费额度可覆盖教程资源消耗，超额后按 token [计费](../concepts/billing.md)。 -- 百炼文件导入支持 pdf、doc、docx、txt、md、pptx、ppt、png、jpg、jpeg、bmp、gif、xls、xlsx，单文档最大 100MB 或 1000 页，单图片最大 20MB，最多 200 个文件；文件存储在新加坡区域，解析通常 1–6 分钟。 -- 钉钉应用需开通 `Card.Streaming.Write` 与 `Card.Instance.Write` 权限以发送卡片消息；应用供企业内其他用户使用需发布版本并设置可见范围。 -- 企业微信可信 IP 一个 IP 仅能用于一个企业，多企业共用会被识别为服务商导致通讯录 / 身份校验接口不可用。 -- 公众号开启服务器配置后自定义菜单会冲突关闭；未认证无法同时开启二者，已认证可通过接口重建菜单。 -- 本地 RAG Windows 系统若缺少 Microsoft Visual C++ Redistributable 需另行安装；报 `DLL load failed while importing _cext:` 时运行 `pip install msvc-runtime`。 -- 上线前建议组织业务人员参与应用[评测](../concepts/evaluation.md)，结合优化提示词、补充私有知识、调整切分策略改进效果。 +- **免费额度限制**：新用户享有免费额度，覆盖教程全部资源消耗；超出后按 token 计费（[在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) §1.1）。 +- **文件上传限制**：云端知识库单文件 ≤ 100 MB 或 1000 页；本地 RAG 工程建议不超过 100 MB（[基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) §传入知识文件）。 +- **微信认证影响**：未认证公众号仅支持被动回复（5 秒超时限制），必须选用对应模板并考虑模型响应速度（如切换 `qwen-turbo`）；认证号支持主动客服消息（[10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) §2）。 +- **可信 IP 与域名**：企业微信要求配置可信 IP 和主体备案域名；若无自有域名，需通过 AppFlow 内网代理或计算巢 Nginx 实例转发（[在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) §常见问题）。 +- **调试建议**：上线前务必进行人工评测（[在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) §总结），重点验证知识库召回准确率与 Prompt 引导效果。 ## 来源文档 - [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) -- [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) - [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) - [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) - - - - - - - - - - - - - - - - - - - +- [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md b/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md index 5d694cf8..0fe70b2f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md @@ -1,195 +1,58 @@ # bailian [application call](../api/application-call.md)ing -阿里云百炼支持通过 [DashScope SDK](../concepts/dashscope-sdk.md) 或 HTTP API 将已创建的应用集成到业务系统中。可调用的应用类型包括**[智能体应用](../concepts/agent-application.md)**和**[工作流](../concepts/workflow.md)应用**（[智能体编排](../concepts/agent-orchestration.md)应用已被[工作流](../concepts/workflow.md)应用替代），二者调用方式一致，均通过 `Application.call` / `POST /apps/{app_id}/completion` 触发，区别仅在于应用内部编排逻辑和可附加的扩展能力（如自定义参数传递）。 +百炼应用调用是将阿里云百炼平台构建的智能体应用或工作流应用集成到业务系统的核心方式，支持通过 DashScope SDK 或标准 HTTP API 进行同步调用。所有调用均需有效 API Key 和应用 ID，并遵循统一的请求结构与认证机制。 -## 前提条件 +## 支持的模型/功能 -无论调用哪种应用，都需要先完成以下准备： +- **应用类型**：支持两类应用调用： + - **智能体应用（Single Agent Application）**：面向单任务、轻量级对话场景，适用于问答、摘要、指令执行等。 + - **工作流应用（Workflow Application）**：面向多步骤、编排型任务，支持插件调用、条件分支、节点串联等复杂逻辑（[调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) 和 [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) 均明确支持该类型）。 +- **核心能力**： + - 单轮文本生成（`prompt` 输入 → `text` 输出） + - 多轮对话（通过 `session_id` 或显式 `messages` 数组维护上下文） + - 自定义插件参数透传（仅限已关联插件的智能体/工作流应用，详见 [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md)） +- **底层模型**：实际执行由应用绑定的模型（如 `qwen-max`、`qwen-plus`）完成，调用方无需指定模型 ID；模型选择在百炼控制台配置并发布应用时确定。 -1. **获取 [API Key](../concepts/api-key.md)**：在百炼控制台密钥管理页面创建 [API Key](../concepts/api-key.md)。 -2. **配置环境变量（推荐）**：将 [API Key](../concepts/api-key.md) 写入 `DASHSCOPE_API_KEY` 环境变量，避免在代码中硬编码。SDK 会自动读取该变量。 -3. **获取应用 ID**：在应用管理页面创建对应应用（[智能体应用](../concepts/agent-application.md) / [工作流](../concepts/workflow.md)应用），并从应用卡片复制 `APP_ID`。 -4. **安装 [DashScope SDK](../concepts/dashscope-sdk.md)**（HTTP 调用可跳过）：Python 通过 `python3 -m pip install -U dashscope`；Java 通过 Maven/Gradle 添加 `com.alibaba:dashscope-sdk-java` 依赖（建议版本 >= 2.12.0）；Node.js 安装 `axios`。 +> **注意**：文档 2 明确声明“本文档仅适用于华北2（北京）地域”，而文档 1 和文档 3 未限定地域。若跨地域调用失败，请优先确认应用部署地域与 API Endpoint 是否匹配（当前统一使用 `https://dashscope.aliyuncs.com`，但后端路由受地域约束）。 -## 基本调用方式 - -[智能体应用](../concepts/agent-application.md)与工作流应用的调用接口完全相同，详见[调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md)与[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)。核心请求结构如下： - -``` -POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion -Authorization: Bearer $DASHSCOPE_API_KEY -Content-Type: application/json - -{ - "input": { "prompt": "你是谁？" }, - "parameters": {}, - "debug": {} -} -``` - -### Python - -```python -import os -from http import HTTPStatus -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) - -if response.status_code != HTTPStatus.OK: - print(f'request_id={response.request_id}') - print(f'code={response.status_code}') - print(f'message={response.message}') -else: - print(response.output.text) -``` - -### Java - -```java -import com.alibaba.dashscope.app.*; -import com.alibaba.dashscope.exception.*; - -ApplicationParam param = ApplicationParam.builder() - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .appId("YOUR_APP_ID") - .prompt("你是谁？") - .build(); - -Application application = new Application(); -ApplicationResult result = application.call(param); -System.out.printf("text: %s\n", result.getOutput().getText()); -``` - -### Node.js / curl - -Node.js 使用 `axios` 发起 POST 请求即可，结构与 curl 示例一致： - -```bash -curl -X POST https://dashscope.aliyuncs.com/api/v1/apps/YOUR_APP_ID/completion \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "input": { "prompt": "你是谁？" }, - "parameters": {}, - "debug": {} - }' -``` - -响应统一为 `{"output": {"finish_reason", "session_id", "text"}, "usage": {...}, "request_id": "..."}` 结构，业务侧主要消费 `output.text`。 - -## 多轮对话 - -工作流应用支持多轮对话，详见[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)。两种实现方式： - -- **使用 `session_id`**：系统自动从云端加载历史对话，实现简单。`session_id` 有效期 1 小时，最多支持 50 轮对话。 -- **自行管理 `messages`（推荐）**：手动维护 `messages` 数组传递每轮历史，无需传 `prompt`，控制更灵活。 - -> **注意**：若请求中同时包含 `session_id` 和 `messages`，系统将优先使用 `messages`。 - -使用 `messages` 时，需先在工作流的大模型节点中配置提示词变量 `historyList` 并发布应用，再发起调用。 - -## 自定义参数传递 - -针对自定义插件与自定义节点，百炼支持通过 `biz_params` 的 `biz_params.user_defined_params` 透传业务参数，详见[应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md)。该能力可用于[智能体应用](../concepts/agent-application.md)的自定义插件，以及工作流应用中的插件节点。 - -### 使用流程 +## 关键参数 -1. **创建自定义插件**：在百炼控制台插件页面新增自定义插件，按需配置鉴权（如用户级鉴权 + Header + basic）。创建工具时，输入参数的**传参方式务必选择「业务透传」**，并发布插件。 -2. **关联应用**：插件工具只能与同一[业务空间](../concepts/workspace.md)内的[智能体应用](../concepts/agent-application.md)关联；工作流应用则在插件节点中引用。关联后发布应用。 -3. **API 调用**：通过 `biz_params.user_defined_params` 传递插件 ID 与入参键值对。 +| 参数名 | 类型 | 必填 | 说明 | +|--------|------|------|------| +| `app_id` | string | ✓ | 百炼控制台生成的应用唯一标识（APP_ID），见 [调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) 前提条件 | +| `prompt` | string | ✓（基础调用） | 用户输入的自然语言指令；若使用 `messages` 多轮模式则可省略 | +| `biz_params` | object | ✗（可选） | 用于传递自定义插件参数，结构为 `{ "user_defined_params": { "": { "": } } }`，详见 [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) | +| `session_id` | string | ✗（可选） | 启用云端会话管理时使用，有效期 1 小时，最多 50 轮；与 `messages` 同时存在时以 `messages` 为准 | +| `messages` | array | ✗（可选） | 显式维护的对话历史数组，格式同 OpenAI `messages`，推荐用于精确上下文控制 | -### 请求示例 - -```python -import os -from http import HTTPStatus -from dashscope import Application - -biz_params = { - "user_defined_params": { - "your_plugin_code": { # 替换为实际插件 ID（在插件卡片获取） - "article_index": 2 - } - } -} - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='寝室公约内容', - biz_params=biz_params -) -``` - -HTTP 请求体中 `biz_params` 位于 `input` 下： - -```json -{ - "input": { - "prompt": "寝室公约内容", - "biz_params": { - "user_defined_params": { - "your_plugin_code": { "article_index": 2 } - } - } - }, - "parameters": {}, - "debug": {} -} -``` +## 使用方式 -Java SDK 通过 `JsonUtils.parse(...)` 将 JSON 字符串转为对象传入 `ApplicationParam.bizParams`。 +### 1. 准备工作 +- 获取 API Key 并配置为环境变量 `DASHSCOPE_API_KEY`（强烈推荐，避免硬编码）； +- 在百炼控制台获取目标应用的 `app_id`； +- 若使用 SDK，按语言安装对应版本（Python 推荐 ≥1.14.0；Java 推荐 ≥2.12.0）。 -## 关键参数 +### 2. 调用方式（三选一） +- **DashScope SDK（推荐）**：封装了认证、序列化、错误处理，各语言示例见 [调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) 的 Python/Java 章节； +- **HTTP API（通用）**：向 `https://dashscope.aliyuncs.com/api/v1/apps/{app_id}/completion` 发起 POST 请求，Header 包含 `Authorization: Bearer ${DASHSCOPE_API_KEY}`，Body 为 JSON 格式； +- **Responses API（OpenAI 兼容）**：文档 2 提示“如需使用 Responses API 调用，请参阅 [Responses API](https://help.aliyun.com/zh/model-studio/openai-responses-api/)”，但原始文档未提供具体用法，开发者需另行查阅该独立文档。 -| 参数 | 位置 | 说明 | -| --- | --- | --- | -| `app_id` / `APP_ID` | URL path | 应用 ID，从应用卡片获取 | -| `prompt` | `input.prompt` | 单轮对话的输入指令（与 `messages` 二选一） | -| `messages` | `input.messages` | 自行维护的多轮对话历史（推荐，优先级高于 `session_id`） | -| `session_id` | `input.session_id` | 云端会话 ID，有效期 1 小时，最多 50 轮 | -| `biz_params` | `input.biz_params` | 业务透传参数，含 `user_defined_params` | -| `parameters` | 顶层 | 应用级参数 | -| `debug` | 顶层 | 调试信息 | +### 3. 多轮对话实现 +- **`session_id` 模式**：首次调用不传，响应中返回 `session_id`；后续请求携带该值，服务端自动加载历史； +- **`messages` 模式（推荐）**：客户端自行维护 `[{ "role": "user", "content": "..." }, { "role": "assistant", "content": "..." }]` 数组，每次请求完整提交，完全可控。 ## 限制和注意事项 -- **地域限制**：[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) 文档明确仅适用于华北2（北京）地域。 -- **应用类型替代关系**：[智能体编排](../concepts/agent-orchestration.md)应用已被工作流应用替代，新场景应使用工作流应用。 -- **`session_id` 约束**：有效期 1 小时，最多 50 轮；与 `messages` 同时存在时优先使用 `messages`。 -- **插件业务透传**：自定义插件的输入参数传参方式必须选择「业务透传」，否则无法通过 `biz_params` 传递；插件 ID 在插件卡片获取，替换示例中的 `your_plugin_code`。 -- **[业务空间](../concepts/workspace.md)隔离**：插件工具只能与同一[业务空间](../concepts/workspace.md)内的[智能体应用](../concepts/agent-application.md)关联。 -- **密钥安全**：不要在生产环境硬编码 [API Key](../concepts/api-key.md)，统一通过 `DASHSCOPE_API_KEY` 环境变量注入。 -- **Responses API**：如需使用 OpenAI 兼容的 Responses API 调用工作流应用，需参阅 Responses API 文档，不在本文调用方式范围内。 +- **地域限制**：工作流应用调用明确限定于华北2（北京）地域（见 [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)），智能体应用未声明此限制，但建议统一部署地域； +- **会话时效**：`session_id` 有效期为 1 小时，超时后需新建会话； +- **插件参数**：`biz_params.user_defined_params` 中的 `plugin_code` 必须与百炼控制台中插件卡片显示的 ID 完全一致，且插件必须已成功关联至目标应用并发布； +- **错误处理**：所有调用均返回标准 HTTP 状态码及 `request_id`，生产环境务必捕获异常并记录 `request_id` 用于问题排查； +- **安全实践**：严禁在代码中硬编码 `api_key`，必须通过环境变量或密钥管理服务注入。 ## 来源文档 -- [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) -- [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) - [调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) - - - - - - - - - - - - - - - - - - - +- [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) +- [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md index c0b98b7e..f05e31e3 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md @@ -1,80 +1,55 @@ # data connection overview -数据连接是阿里云百炼平台管理外部数据源的统一入口。通过创建数据连接器，百炼应用可以安全地访问企业数据库、文档系统和对象存储中的数据，并在对话中实时查询和引用这些数据。详见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md)。 +数据连接是阿里云百炼平台统一管理外部数据源的核心能力，为应用提供安全、可控的数据接入通道。它支持结构化与非结构化数据的接入，并通过平台托管或流处理两种模式实现数据访问与实时查询。所有连接器均需在控制台创建并配置权限后方可被智能体或API调用。 -## 连接器类型 +## 支持的模型/功能 -数据连接器按数据的存储和访问方式分为两大类： +数据连接器按数据访问方式分为两类： -- **平台托管**：数据导入并存储在百炼平台（或自有 OSS）。 - - **文件**：管理非结构化文档（PDF、Word、Markdown 等）。 - - **表格**：导入并查询结构化表格数据（CSV、Excel 等）。 -- **流处理**：数据保留在原数据源，实时访问。 - - **MySQL / PostgreSQL / PolarDB-X 2.0**：连接对应数据库，支持执行 SQL 查询（仅 DMS 导入方式支持）。 - - **语雀**：访问语雀文档和知识库（仅公网版本）。 - - **OSS**：访问对象存储中的文件。 +- **平台托管型**：适用于文件（PDF/Word/Markdown等）和表格（CSV/Excel等）类非实时数据。数据导入后由百炼平台统一向量化并构建知识库索引，支持语义检索、标签过滤及多模态解析（如[文档理解](https://help.aliyun.com/zh/document-mind/product-overview/overview-of-document-understanding#9a4f5fb91fpps)中描述的Qwen VL解析、音视频解析等）。详见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) 中“导入文件”与“导入表格”章节。 -各类型的适用场景与存储方式详见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的连接器类型表。 +- **流处理型**：适用于需实时访问的数据库与在线服务，包括 MySQL、PostgreSQL、PolarDB-X 2.0、语雀和 OSS。其中仅通过 **DMS 导入数据源** 方式创建的 MySQL/PostgreSQL/PolarDB-X 连接器支持执行 SQL 查询；其余方式（如自定义数据源）仅支持元数据发现与基础连接验证。OSS 连接器需开通[向量检索服务](https://help.aliyun.com/zh/oss/user-guide/vector-retrieval/)才可使用 `searchOSSFile` 等工具。该设计细节在 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的各连接器说明中明确区分。 -## 前置条件 +> **注意**：语雀连接器**仅支持公网版本语雀**，不兼容私有部署版；OSS 连接器**不支持归档、冷归档或深度冷归档存储类型**的 Bucket —— 这些限制在 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) “OSS连接器”小节末尾有明确标注，开发者须提前校验存储类型。 -- **账号权限**：主账号或具有数据连接管理权限的 RAM 用户；RAM 用户需先获得主账号授权。 -- **数据源准备**（按连接器类型）： - - 文件/表格：准备好待上传文档/表格，或已创建 OSS Bucket。 - - MySQL：已有 MySQL 实例（RDS 或自建），网络可达（公网或私网）。 - - PostgreSQL：账号具备高权限（Superuser 或 REPLICATION），且 `wal_level` 设置为 `logical`；自建实例还需配置 `listen_addresses` 允许 `100.64.0.0/16` 网段访问。 - - PolarDB-X 2.0：已有阿里云 PolarDB-X 2.0 实例且所在地域支持私网访问；DMS 导入方式需先在 DMS 录入实例。 - - 语雀：已有公网版语雀知识库并获取访问 Token。 - - OSS：已创建 Bucket 并开通向量检索服务。 +## 关键参数 -## 数据库连接器关键差异 +| 参数类别 | 关键字段 | 说明 | +|----------|----------|------| +| **通用** | 连接器名称、描述 | 名称需唯一且易识别；描述影响智能体调用准确度，建议包含数据内容与用途（如“财务部2024Q1销售报表”）。 | +| **平台托管（文件/表格）** | 存储位置（平台存储 / 自有OSS） | 平台存储提供免费额度（文件：200,000个/1TB；表格：1TB），超限后转按量付费；自有OSS需添加 `bailian-connector-access` 标签（值 `ReadAndWrite`）。 | +| **流处理（数据库）** | 数据库地址、端口、用户名、密码、dbName（PostgreSQL必填） | MySQL 默认端口 3306，PostgreSQL 默认 5432；PolarDB-X 2.0 **仅支持私网**，且必须选择所属地域。 | +| **流处理（语雀/OSS）** | Tenant access token（语雀）、Bucket 名称（OSS） | 语雀 [Token](../concepts/token.md) 需从[语雀开放 API](https://www.yuque.com/yuque/developer/api) 获取；OSS Bucket 需添加 `bailian-datahub-access` 标签（值 `read`），并确保当前账号具备访问权限。 | -| 差异项 | MySQL | PostgreSQL | PolarDB-X 2.0 | -| --- | --- | --- | --- | -| 默认端口 | 3306 | 5432 | 自动获取 | -| 额外必填字段 | 无 | dbName（数据库名称） | 无 | -| 连通性检测服务 | EventBridge | DTS | DTS | -| 网络类型 | 公网 / 私网 | 公网 / 私网 | 仅私网 | -| 特殊配置 | 无 | `wal_level=logical` | 需 SLR 授权（DTS、PolarDB-X，DMS 方式加 DMS 角色） | +## 使用方式 -三类流处理数据库连接器均支持两种数据来源配置：**创建自定义数据源**（手动配置连接信息）与**从 DMS 导入数据源**（导入 DMS 中已有数据源）。数据库用户须具备读取权限，配置后可点击检测验证连通性。完整字段说明见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的各连接器配置章节。 +1. **创建连接器**：进入 [数据连接](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/connector/list) 页面 → 单击“创建连接器” → 选择类型 → 填写基本信息与连接参数 → （可选）点击“开始检测”或“连接检测”验证连通性 → 确认创建。 -> **注意**：仅通过**从 DMS 导入数据源**方式创建的 MySQL / PostgreSQL / PolarDB-X 2.0 连接器支持执行 SQL 查询；通过**创建自定义数据源**方式添加的连接器不支持直接执行 SQL。 +2. **导入数据（仅平台托管型）**： + - 文件连接器：进入详情页 → 选择类目 → “导入数据” → 本地上传 → 选择解析方式（默认/自定义）→ 配置标签（可选）→ 确认。 + - 表格连接器：进入详情页 → 数据表管理 → 新建或选择数据表 → 上传 Excel 或自定义表头（列名、类型必填，描述建议填写）→ 确认。 -## 创建连接器 +3. **调用数据**： + - 平台托管型：通过知识库检索接口（如 `retrieveFromKnowledgeBase`）或智能体内置工具自动触发； + - 流处理型：MySQL/PostgreSQL/PolarDB-X 需通过 DMS 导入方式创建后，方可使用 `executeSQL` 工具；语雀/OSS 使用对应工具（如 `searchYuQueDoc`、`searchOSSFile`）。 -1. 在[数据连接](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/connector/list)页面点击**创建连接器**。 -2. 选择连接器类型，填写**连接器名称**和**描述**（描述会用于指导应用调用的准确度，建议写明数据内容和用途）。 -3. 按类型填写存储位置或数据源信息，必要时执行连通性检测。 -4. 点击**确认**完成创建。 - -存储位置选择要点： - -- **文件连接器**：平台存储限时免费（最多 200,000 个文件、1 TB）；或使用自有 OSS，需为 Bucket 添加 `bailian-connector-access` 标签（值 `ReadAndWrite`）。 -- **表格连接器**：平台存储提供 1 TB 免费额度，用尽后转按量付费；自有 OSS 同样需上述标签。 -- **OSS 连接器**：从下拉列表选择 Bucket，需添加 `bailian-datahub-access` 标签（值 `read`）。 - -## 导入数据 - -- **导入文件**：进入文件连接器详情页，在**类目**下选择或新建类目后导入。平台暂不支持直接导入 JSON、CSV、YAML，需先转换为 XLSX/XLS。可选择**默认设置**或**自定义设置**解析（电子文档解析、文档智能解析、大模型文档解析、Qwen VL 解析、音视频解析等，具体能力取决于文件类型）。可为文件配置**标签**，API 调用时通过 `tags` 参数筛选以提升检索效率。 -- **导入表格**：进入表格连接器详情页，在**数据表管理**下选择或新建数据表。支持**直接上传 Excel**（自动识别表头）或**自定义表头**。数据表结构（列名、描述、类型）一旦确定不可修改，且上传文件的列数与列名必须与表结构一一对应，否则导入失败。 +完整操作流程与界面指引见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md)。 ## 限制和注意事项 -- 数据库连接器执行 SQL 的限制见上文注意框（仅 DMS 导入方式支持）。 -- **OSS 连接器**：使用需开通[向量检索服务](https://help.aliyun.com/zh/oss/user-guide/vector-retrieval/)，否则无法使用 `searchOSSFile` / `searchOSSFileByFileName` 工具；不支持归档/冷归档/深度冷归档类型的 Bucket；支持内容加密与私有 Bucket；开启 Referer 防盗链时需将 `*.console.aliyun.com` 加入白名单。 -- **文件导入**：文件作为独立副本存储在平台免费空间（当前无容量限制），仅支持查看最近 **90** 天内导入的文件（超期不可查看但不删除），且仅供当前业务空间使用。请求高峰期解析可能耗时数小时甚至偶现超时，需耐心等待或重试。 -- **语雀连接器**：仅支持公网版本语雀，需提供有效的 Tenant access token。 - -以上流程、字段和限制的完整细节，请以原文 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 为准。 +- **权限要求**：RAM 用户需主账号授予 `AliyunBailianFullAccess` 或最小权限策略（含 `bailian:CreateConnector`、`bailian:DescribeConnectors` 等动作），授权方法参见[权限管理](https://help.aliyun.com/zh/model-studio/application-permission-management-overview)。 +- **网络与配置依赖**： + - PostgreSQL 必须将 `wal_level` 设置为 `logical`，且自建实例需配置 `listen_addresses` 允许 `100.64.0.0/16` 访问； + - MySQL/PolarDB-X 自建实例需确保网络可达，并将百炼服务 IP 段加入白名单； + - PolarDB-X 2.0 **不支持公网连接**，且仅限阿里云实例。 +- **文件处理限制**： + - 不支持直接导入 JSON、CSV、YAML；需先转为 XLSX/XLS； + - 导入文件仅保留最近 90 天的查看记录（但数据本身不删除）； + - 电子文档解析不支持插图与图表识别；如需图文理解，必须选用“大模型文档解析”或“Qwen VL解析”。 +- **OSS 特别说明**：若 Bucket 启用 Referer 防盗链，须将 `*.console.aliyun.com` 加入白名单 Referer，否则连接失败。 ## 来源文档 - [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md b/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md index 3781b4b1..24d6d377 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md @@ -1,82 +1,71 @@ # fine tuning -模型微调（Fine-tuning）是阿里云百炼在 Prompt 工程、插件调用等手段仍无法满足效果时提供的深度定制手段。它覆盖文本生成、视觉理解（Qwen-VL）、图像/视频生成（万相）以及语音合成（CosyVoice）等多种模态，通过 SFT、CPT、DPO 等训练方式，把领域知识、任务能力、人类偏好或特定音色/风格直接写入模型参数。 +fine tuning 是阿里云百炼平台提供的核心模型优化能力，允许开发者基于自有数据对预训练大模型进行定制化训练，从而提升其在特定业务场景、领域知识或风格表达上的表现。该能力覆盖文本生成、视觉理解、语音合成、图像生成、视频生成及强化学习等多种模态与范式，支持高效微调（LoRA）、全参微调、持续预训练（CPT）、直接偏好优化（DPO）和强化学习（RL）等多种技术路径。所有 fine tuning 任务当前均仅限华北2（北京）地域使用，且需配置对应地域的 API Key [原文标题](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 -> **注意**：以下所有微调、部署与调用能力均**仅在华北2（北京）地域可用**，且必须使用该地域的 API Key；子账号（RAM 用户）需预先被授予调用、训练和部署权限。 +## 支持的模型/功能 -## 支持的模型与训练方式 +百炼平台支持多模态、多范式的 fine tuning，具体能力按模型类型划分： -按模态划分，不同模型支持的训练方式差异明显： +- **文本生成模型**：支持 Qwen 系列（如 `qwen3-8b`, `qwen3.5-9b`, `qwen3-32b`）、千问VL系列（如 `qwen3-vl-8b-instruct`）等，提供 SFT（监督微调）、CPT（持续预训练）、DPO（直接偏好优化）三种训练方式，其中 SFT 高效训练（`efficient_sft`）为默认推荐方案 [原文标题](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 +- **图像生成模型**：支持万相系列（如 `wan2.7-image-pro`, `wan2.7-image`），采用 SFT-LoRA 高效微调，适用于文生图（t2i）和图生图（i2i）两种模式，可定制特定 IP 形象、艺术风格或画面特效 [原文标题](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +- **视频生成模型**：支持万相图生视频系列（如 `wan2.7-i2v`, `wan2.2-kf2v-flash`），同样采用 SFT-LoRA，支持基于首帧或首尾帧的微调，用于稳定复现特定运动特效（如“金钱雨”、“时尚杂志”） [原文标题](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md)。 +- **语音合成模型**：支持 CosyVoice 系列（仅 `cosyvoice-v3-flash`），通过 SFT 高效微调实现同一发音人的高还原度专属音色定制，产物为独立部署的单音色模型 [原文标题](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +- **强化学习（RL）**：支持 Qwen3.5-9B 等基座模型，通过“生成-评分-优化”循环进行策略自主探索，适用于数学推理、Agent 工具调用等需深度推理的场景，**必须使用模型训练单元（MTU）计费** [原文标题](../../raw/model-user-guide/fine-tuning/rl-training-overview.md)。 -- **文本生成（千问系列）**：支持 CPT、SFT（全参 `sft` / 高效 `efficient_sft`）、DPO（全参 `dpo_full` / 高效 `dpo_lora`）。是否支持某种方式因模型而异，例如 Qwen3-32B、Qwen3-4B/1.7B/0.6B、Qwen2.5 系列支持全部 5 种；而 Qwen3.5-Plus/Flash、Qwen3.6/3.7 等新模型往往仅支持 `sft`。详见 [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 -- **视觉理解（千问 VL）**：Qwen3-VL、Qwen2.5-VL 系列支持 SFT 全参与高效训练，不支持 CPT/DPO。 -- **图像生成（万相）**：`wan2.7-image-pro`、`wan2.7-image`，仅支持 SFT-LoRA 高效微调，见 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 -- **视频生成（万相）**：图生视频-基于首帧 `wan2.7-i2v`/`wan2.5-i2v-preview`/`wan2.2-i2v-flash`，基于首尾帧 `wan2.2-kf2v-flash`，同样仅支持 SFT-LoRA，见 [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md)。 -- **语音合成（CosyVoice）**：`cosyvoice-v3-flash`，仅支持 `efficient_sft`，且**当前只能通过 API 发起，控制台暂不支持**，见 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +> **注意**：文档 4 和文档 5 中关于“SFT 全参训练（sft）”与“SFT 高效训练（efficient_sft）”的支持列表存在不一致。例如，文档 4 显示 `qwen3.5-9b` 同时支持二者，而文档 5 的表格中 `qwen3.5-9b` 行在“SFT全参训练（sft）”列为“支持”，但在“SFT高效训练（efficient_sft）”列为空白。根据文档 1、2、6、8 的实操示例及文档 7 的明确推荐，`efficient_sft` 是图像、视频、语音及多数文本模型的主流且推荐方式，应以实际 API 接口和控制台可选项为准。 -## 三种调优方式（文本生成） +## 关键参数 -推荐按递进顺序组合使用：`CPT（可选）→ SFT → DPO（可选）`。 +不同训练方式的核心超参数差异显著，开发者需根据任务类型选择： -| 方式 | 目标 | 数据量 | 数据形态 | -| --- | --- | --- | --- | -| CPT（持续预训练） | 补领域知识 | 1000 万+ Token | 无标签领域文本 `{"text":"..."}` | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条 | ChatML「问-答」对 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组 | 同指令下「更好/更差」回答对（`chosen`/`rejected`） | +- **通用 SFT 参数（文本/视觉/语音）**： + - `learning_rate`：高效训练推荐 `1e-4` 量级（如 `3e-4`），全参训练推荐 `1e-5` 量级；过高易导致发散，过低收敛缓慢。 + - `n_epochs` / `max_steps`：控制训练轮次或总步数。文本 SFT 默认 `3` 轮；图像微调（文档 1）使用 `max_steps=800`；视频微调（文档 2）使用 `n_epochs=50`；语音微调（文档 8）则解耦为 `lm_max_epoch` 和 `fm_max_epoch`。 + - `batch_size`：影响显存占用与收敛稳定性。文本推荐 `16` 或 `32`；图像微调中 `t2i` 模式建议 `1k` token；视频微调中 `wan2.7-i2v` 推荐 `batch_size=1`；语音微调中 `lm_batch_size=1000`。 + - `lora_rank`：LoRA 低秩矩阵维数，决定微调参数量。图像微调（文档 1）设为 `32`；文本微调（文档 5）默认 `8`；语音微调（文档 8）无此参数，因其采用专用 LM/FM 架构。 -训练模式分**全参训练**与**高效训练（LoRA）**：两者费用相同，官方建议在模型支持全参训练时优先选择全参（效果更好、性价比更高）；LoRA 适合对训练时间/成本敏感或数据集较小的场景。 - -## 关键超参数 - -文本生成调优的常用超参及默认值（以控制台实际显示为准）： - -- `learning_rate`：高效训练建议 `1e-4` 量级，全参/CPT 建议 `1e-5` 量级。 -- `n_epochs`：默认 `3`，范围 `[1, 200]`；数据量 <10000 建议循环 3~5 次，>10000 建议 1~2 次。 -- `batch_size`：一般 16/32。 -- `max_length`：建议设为模型支持的最大值；SFT 会**丢弃**超长数据，DPO 则**截断**后仍训练。 -- `lora_rank` / `lora_alpha` / `lora_dropout`：LoRA 专用，秩越大效果略好但更慢、更易过拟合。 -- 通过 API 创建任务时，`n_epochs`、`batch_size`、`max_length` 因影响计费而**必填**。 - -> **注意**：默认学习率各文档取值不一致。控制台参数面板列出的 `learning_rate` 默认值为 `3e-4`（对应高效训练默认场景），而 API 示例中 SFT 全参使用的是 `1.6e-5`。请以实际训练方式对应的量级为准，切勿照搬。 - -万相图像/视频与 CosyVoice 使用各自独立的超参集，例如万相有 `max_steps`/`generation_type`/`val_img_size`，CosyVoice 分 `lm_*`（影响韵律）与 `fm_*`（影响音色）两组网络的 `*_max_epoch`/`*_step`/`*_num`/`*_batch_size`（8 个子字段全部必填）。 +- **模态特有参数**： + - 图像/视频：`max_pixels`（训练图片/视频最大像素总数）、`val_img_size`（验证图分辨率）、`generation_type`（`t2i` 或 `i2i`）等，直接影响输入尺寸与输出质量。 + - 视觉理解（VL）：`resized_width`/`resized_height` 可在 data.jsonl 中为每张图/视频帧指定缩放目标。 + - 强化学习（RL）：`algorithm`（如 `gspo`）、`kl_loss_coef`（KL 散度系数）、`n_rollouts`（每样本采样次数）等，构成 RL 特有的算法栈。 ## 使用方式 -**控制台（推荐入门）**：在[模型调优](https://bailian.console.aliyun.com/?tab=model#/efm/model_manager)页面创建训练任务 → 选训练方式与模型 → 配置训练集/验证集（可自动切分）→ 配置 Checkpoint 保存 → 开始训练 → 部署 → 评测。零代码场景可参考 [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md)，其中给出了以 Qwen3-8B 为例的完整安全对齐 SFT 流程与超参实验对照（全参 `n_epochs=3`/`lr=1e-5` 或 LoRA `n_epochs=3`/`lr=3e-4` 效果较好）。 - -**API / 命令行**：统一四步流程，详见 [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md)： +fine tuning 流程标准化为四步：准备数据 → 上传文件 → 创建任务 → 部署调用。 -1. 上传数据集到 `POST /api/v1/files`（`purpose=fine-tune`），获取 `file_id`。 -2. `POST /api/v1/fine-tunes` 创建任务，关注返回的 `job_id`、`finetuned_output`、`status`。 -3. 轮询 `GET /api/v1/fine-tunes/` 直到 `status` 变为 `SUCCEEDED`。 -4. `POST /api/v1/deployments` 部署（`plan=lora`），轮询直到 `status` 为 `RUNNING`，再用 `deployed_model` 调用。 +1. **准备数据集**： + - 文本 SFT：使用 ChatML 格式 JSONL 文件，每行含 `messages` 数组（含 `system`/`user`/`assistant` 角色），`assistant` 输出即为监督信号 [原文标题](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md)。 + - 图像/视频：ZIP 压缩包，内含 `data.jsonl`（定义 [prompt](prompt.md)/image/video 路径）及对应媒体文件；图像单张不超过 `10MB`，分辨率建议控制在 `8K` 内。 + - 语音：ZIP 包含 `data.jsonl`（字段 `wav_fn` 和 `text`）及 `train/` 目录下的 `.wav` 文件，采样率 ≥ `16kHz`，单条时长 `2-30` 秒。 + - 强化学习：JSONL 格式，每行含 `messages`（用户问题）和 `rollout_extra`（参考答案），用于 Reward 函数评分。 -> **注意**：通过 API 创建的训练任务**仅支持按 Token 计费**，不支持模型训练单元（预付费/后付费）；如需使用训练单元，必须通过控制台创建。 +2. **上传文件**： + - 通过 `/api/v1/files` 接口上传 ZIP 或 JSONL 文件，`purpose="fine-tune"`，获取 `file_id`。 + - 支持 OSS 挂载（需指定 `region`/`bucket`/`file_path`），但图像/视频 ZIP 不支持，仅支持解压后的原始文件结构。 -数据集除 `file_id` 外还可用 OSS 挂载（`data_source_type=oss_mount`），OSS Bucket 地域支持 `cn-beijing` 与 `ap-southeast-1`，挂载时只需指定 `data.jsonl` 路径。 +3. **创建训练任务**： + - 调用 `/api/v1/fine-tunes`，传入 `model`、`training_datasets`（含 `file_id` 或 OSS 配置）、`training_type`（如 `efficient_sft`）及 `hyper_parameters`。 + - 任务状态初始为 `PENDING`，需轮询 `/api/v1/fine-tunes/{job_id}` 直至 `status="SUCCEEDED"`。 -## 数据格式要点 +4. **部署与调用**： + - 成功后，`finetuned_output` 即为新模型名，需调用 `/api/v1/deployments` 部署为在线服务。 + - 部署状态变为 `RUNNING` 后，即可用 `deployed_model` 名称调用对应 API（如 `/services/aigc/image-generation/generation`）。 -- **SFT**：ChatML `{"messages":[...]}`，支持多轮；不支持 OpenAI 的 `name`/`weight`，所有 assistant 行都会被训练；思考模型（thinking）只训练**最后**一个 assistant 输出且须保留 `` 标签前后的换行。 -- **DPO**：在 `messages` 基础上追加 `chosen`/`rejected`。 -- **CPT**：纯文本 `{"text":"..."}`。 -- **视觉理解**：`content` 用数组，`image`/`video` 声明文件名（不含路径），打包为 ZIP（≤2GB），`data.jsonl` 必须在根目录，文件名全局唯一且仅含 ASCII。 -- **CosyVoice**：`data.jsonl` 每行 `{"wav_fn":"train/xxx.wav","text":"..."}`，`wav_fn` 必须以 `train/` 前缀；`text` 须为纯文本，禁止 SSML/LaTeX/情感标注。 +## 限制和注意事项 -## 限制与注意事项 - -- **成本与耗时高**：文本模型微调需构建大规模数据集，且调优后模型**必须部署才能使用**，部署费用较高；官方明确将模型调优定位为「最后的手段」，建议先充分尝试 Prompt 工程与插件调用。 -- **训练耗时差异大**：万相文生图约数十分钟，视频微调可达数小时；文本 LoRA 通常 15~30 分钟；部署一般需 3~10 分钟。 -- **计费方式各异**：文本/VL 按训练 Token 计费（`训练Token × 循环次数 × 单价`）；CosyVoice 按 `(lm_max_epoch+fm_max_epoch)×25×音频总秒数` 估算 Token，单价 0.2 元/千 Token，另加部署时长费用。 -- **能力边界不可突破**：CosyVoice 调优产物为单音色模型（`voice` 锁定 `default`），无法新增基础模型不支持的语种、也不支持 `instruction` 指令控制。 -- **过拟合/欠拟合判断**：观察 Training/Validation Loss 曲线，欠拟合可增大 `n_epochs`/`lora_rank`，过拟合则反向调整。 -- 万相 LoRA 调用需在提示词中包含**触发词**以激活风格；图像模型部署后当前仅支持异步调用。 +- **地域与权限**：所有 fine tuning 功能**仅限华北2（北京）地域**，且必须使用该地域的 API Key。子账号需被授予模型调用、训练、部署的完整权限 [原文标题](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +- **数据与文件**：ZIP 包最大 `2GB`，单个文件上传上限 `300MB`；图像分辨率需满足宽高比 ≤ `200:1`，最小尺寸 > `10px`；语音训练数据必须为同一发音人。 +- **计费模式**： + - 文本/图像/视频/语音微调：按训练消耗 [Token](../concepts/token.md) 总数计费，公式为 `Token总数 × 单价`。 + - 强化学习：**强制使用模型训练单元（MTU）计费**，不支持 [Token](../concepts/token.md) 计费，需预先购买或开通后付费 MTU [原文标题](../../raw/model-user-guide/fine-tuning/rl-training-overview.md)。 +- **产物特性**：微调产物是独立模型（如 `xxxx-ft-...`），非基础模型的音色 ID 或插件；CosyVoice 微调后 `voice` 参数固定为 `default`，不可切换音色。 +- **效果预期**：fine tuning 旨在提升特定场景表现，**无法扩展基础模型能力边界**（如 CosyVoice 无法通过微调支持新语种，Qwen VL 无法通过微调支持新视频格式）。 ## 来源文档 - [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md) - [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md) +- [强化学习训练概述](../../raw/model-user-guide/fine-tuning/rl-training-overview.md) - [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md) - [在控制台进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md) - [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md) @@ -84,4 +73,3 @@ - [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md b/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md index 56974632..88ef97cc 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md @@ -1,73 +1,50 @@ # get started with models -阿里云百炼是一站式大模型开发与应用平台，集成千问（Qwen）全系列及 DeepSeek、Kimi、GLM 等主流第三方模型，并提供兼容 OpenAI 的 API。本页面帮助开发者快速完成从选择模型、获取 API Key、配置 Base URL 到发起首次调用的全流程，并梳理地域、接入域名与限流等关键约束。 +阿里云百炼提供开箱即用的大模型服务，支持通过兼容 OpenAI 的 API 快速调用千问（Qwen）及第三方模型。开发者无需部署运维，只需配置 API Key 和 Base URL 即可发起首次请求。本文档聚焦模型调用的入门路径，涵盖模型选择、参数配置、调用方式及关键限制。 -## 支持的模型与能力 +## 支持的模型与功能 -百炼提供开箱即用的模型服务，无需自行部署或运维即可调用。文本生成方面，千问旗舰模型按能力与成本分层选择（参见 [选择模型](../../raw/model-user-guide/get-started-with-models/models.md)）： +百炼提供多系列千问模型（如 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.7-flash`）及 DeepSeek、Kimi、GLM 等第三方模型，覆盖文本生成、多模态理解与生成、嵌入向量等能力 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。模型按能力与成本分层：`qwen3.7-max` 适合复杂任务；`qwen3.7-plus` 是效果、速度与成本的均衡推荐；`qwen3.7-flash` 适用于低延迟简单任务。部分模型（如 `qwen3.8-max-preview`）仅限 [Token](../concepts/token.md) Plan 订阅用户使用 [选择模型](../../raw/model-user-guide/get-started-with-models/models.md)。 -- **千问 Max**（如 `qwen3.7-max`）：Qwen 系列效果最好的模型，适合复杂、多步骤任务。 -- **千问 Plus**（如 `qwen3.7-plus`）：效果、速度和成本均衡，多数场景的**推荐选择**。 -- **千问 Flash**（如 `qwen3.6-flash`）：高性价比、低延迟，适合需要快速响应的简单任务。 +> **注意**：文档 1 中称 “qwen3.7-max 推理能力全面超越前代”，而文档 5 的限流表中列出 `qwen3.7-max` RPM/TPM 为 30,000/5,000,000，但同表中 `qwen3.7-max-2026-06-08` 等快照版本限流仅为 600/1,000,000。这表明高限流额度仅适用于稳定版（无日期后缀），而非所有带版本号的变体。实际选型应以[限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)文档中的具体数值为准。 -此外还覆盖视觉理解、图像生成、视频生成、语音识别与合成、嵌入向量等多模态能力，以及长文本、翻译、法律等细分领域模型。平台同时支持模型调优（SFT / CPT / DPO）、模型部署与模型评测，详见 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 +## 关键参数 -## 关键概念与参数 - -调用前需要先确定四个维度（详见 [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md)）： - -- **地域（Region）**：决定接入点和数据存储位置。目前提供华北2（北京，`cn-beijing`）、新加坡（`ap-southeast-1`）、日本（东京，`ap-northeast-1`）、德国（法兰克福，`eu-central-1`）、美国（弗吉尼亚，`us-east-1`）。就近选择可降低延迟。 -- **服务部署范围**：决定推理执行位置。有数据合规需求时选择特定地理边界（如中国内地、欧盟、美国），无合规需求可选全球部署（推理资源池更大）。德国、日本地域通过[业务空间（Workspace）](../concepts/workspace.md)区分部署范围；美国地域可用带 `-us` 后缀的模型名（如 `qwen-plus-us`）限定境内推理。 -- **接入域名**：影响并发上限、超时等服务保障，推荐使用**业务空间专属域名**（`{WorkspaceId}.{region}.maas.aliyuncs.com`，SLA 99.9%、请求超时 3600 秒、支持 HTTP/SSE/WebSocket/WebRTC），另有 Dashscope 域名（现有，超时 600 秒）和试用域名（限流小，不建议生产）。 -- **API Key**：各地域相互独立、不能跨地域混用；Base URL 也必须与同一计费方案的 API Key 配套使用，否则报 401。 - -> **注意**：各地域接入点（Base URL）、API Key 和模型列表均不能跨地域混用；使用北京、新加坡、日本、德国地域时，业务空间专属域名中的 `{WorkspaceId}` 需替换为真实业务空间 ID（可在业务空间管理页查看）。 +- **API Key**：必须通过[阿里云百炼控制台](https://bailian.console.aliyun.com/?tab=model#/api-key)创建，不同地域的 Key 不通用。 +- **Base URL**：必须与 API Key 所属地域和计费方案严格匹配。生产环境**强烈推荐使用业务空间专属域名**（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`），其具备更高吞吐、更低时延与流量隔离；Dashscope 域名（如 `https://dashscope.aliyuncs.com/compatible-mode/v1`）为兼容性保留，试用域名（如 `https://trial.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`）仅限快速验证 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)。 +- **WorkspaceId**：在华北2（北京）、新加坡、日本（东京）、德国（法兰克福）地域调用业务空间专属域名时必需，可在[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)页面获取。 +- **模型 ID**：如 `qwen3.7-plus`，需与所选地域实际支持的模型一致（例如 DeepSeek 仅支持北京地域）。 ## 使用方式 -**1. 账号与凭证准备**（参见 [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md)）：注册并开通百炼 → 在 API Key 页面创建 Key → 获取业务空间 ID。建议将 Key 配置到环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄露。 - -**2. 选择 Base URL**（详见 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)）：百炼提供 OpenAI 兼容（`/compatible-mode/v1`）、Anthropic 兼容（`/apps/anthropic`）、DashScope（`/api/v1`）三类接口。以北京地域业务空间专属域名的 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)为例：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。 - -**3. 发起调用**：兼容 OpenAI 接口规范，迁移现有代码只需调整 API Key、base_url 和模型名称。Python 示例： - -```python -import os -from openai import OpenAI - -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -) -completion = client.chat.completions.create( - model="qwen-plus", - messages=[ - {'role': 'system', 'content': 'You are a helpful assistant.'}, - {'role': 'user', 'content': '你是谁？'} - ] -) -print(completion.choices[0].message.content) -``` - -也可通过 DashScope SDK（`pip install -U dashscope`）、curl，或 Chatbox、Claude Code 等客户端/开发工具调用。OpenAI Python SDK 要求 Python ≥ 3.8。 - -## 限制与注意事项 - -- **限流按主账号维度合并计算**：账号下所有 RAM 子账号、业务空间和 API Key 的调用量合并统计，不同模型限流额度相互独立。超限请求被拒绝，通常一分钟内自动恢复。详见 [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)。 -- **RPM 与 TPM 双重约束**：`Requests rate limit exceeded` 表示触发每分钟请求数（RPM）限流；`Allocated quota exceeded` 表示触发每分钟 Token 数（TPM）限流；`Request rate increased too quickly` 表示请求瞬时激增触发稳定性保护。限流可能按秒级 RPS（RPM/60）、TPS（TPM/60）执行。 -- **规避限流**：优先选用高限流额度模型（稳定版比日期快照版更宽松）、平滑请求速率（匀速/指数退避/队列）、配置备选模型自动切换、拆分任务、无需实时响应时改用 Batch API（不受实时限流约束）。北京与新加坡地域支持在控制台申请临时 TPM 提额（生效 30 天）。 -- **计费独立**：模型推理按 Token 用量计费，知识库（RAG）按规格时长与调用独立计费，两者互不相通。限流只约束速率、不限制累计用量；如需控费可设置费用告警、开启"免费额度用完即停"或订阅 Coding Plan（固定月费）。 -- **各地域功能差异**：批量推理、模型调优、应用开发等能力目前主要在华北2（北京）支持，海外地域功能覆盖较少，选型前请核对目标地域的功能与模型列表。 - -> **注意**：Token Plan 与 Coding Plan 的 Base URL 及专属 API Key 仅限 Claude Code、Codex 等 AI 工具交互式使用，不能用于后端服务；使用非专属 Base URL 调用将按量付费。 +1. **环境准备**：注册阿里云账号并完成实名认证；开通百炼服务；创建 API Key 并配置为环境变量 `DASHSCOPE_API_KEY`（避免硬编码）[首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md)。 +2. **代码调用**：使用 OpenAI Python SDK 或 DashScope SDK。OpenAI SDK 示例： + ```python + from openai import OpenAI + client = OpenAI( + api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" + ) + response = client.chat.completions.create( + model="qwen3.7-plus", + messages=[{"role": "user", "content": "你是谁？"}] + ) + ``` +3. **地域适配**：不同地域 Base URL 不通用。美国（弗吉尼亚）使用 `https://dashscope-us.aliyuncs.com/compatible-mode/v1`；新加坡使用 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 + +## 限制和注意事项 + +- **限流策略**：按主账号维度合并计算所有子账号、业务空间和 API Key 的调用量。触发条件包括每分钟请求数（RPM）或 [Token](../concepts/token.md) 数（TPM）超限，或请求速率突增（`Request rate increased too quickly`）。`qwen3.7-plus` 在北京地域 RPM/TPM 为 30,000/5,000,000，而同模型在新加坡地域为 15,000/5,000,000 [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)。 +- **地域约束**：各地域 API Key、Base URL、模型列表相互独立，不可混用。德国（法兰克福）和日本（东京）地域不支持 DashScope 域名，美国（弗吉尼亚）地域不支持业务空间专属域名 [地域及接入域名](../../raw/model-user-guide/get-started-with-models/regions.md)。 +- **费用控制**：模型调用按量付费，知识库（RAG）功能计费独立。新用户可享北京地域免费额度，建议开启“免费额度用完即停”开关防止意外扣费 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 ## 来源文档 - [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md) -- [选择模型](../../raw/model-user-guide/get-started-with-models/models.md) - [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) -- [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) -- [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) +- [选择模型](../../raw/model-user-guide/get-started-with-models/models.md) - [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) +- [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) +- [地域及接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md b/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md index 8f2e139a..67fefc13 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md @@ -1,89 +1,49 @@ # [knowledge](../api/knowledge.md) base -知识库（Knowledge Base）是阿里云百炼平台基于 RAG（检索增强生成）技术为大模型补充私有数据和最新信息的能力。大模型在生成回答前先从知识库中检索语义相关的内容，从而显著提升在特定领域问题上的准确性。围绕知识库，平台还提供了知识检索、知识问答、API 集成、日志监控与计费等一整套配套能力。 +知识库是阿里云百炼平台提供的 RAG（[检索增强生成](../concepts/rag.md)）核心能力，用于为大语言模型注入私有、领域专属或时效性强的结构化与非结构化数据。它通过语义检索从用户上传的文档、表格、图片或音视频中召回相关内容，并将结果作为上下文输入至大模型，从而提升回答的准确性、专业性与事实一致性。该功能仅在中国站华北2（北京）地域可用。 -> **注意**：知识库功能仅能在中国站 **华北2（北京）** 地域开通和使用，新加坡、德国（法兰克福）等其他地域均不支持。 +## 支持的模型/功能 -## 支持的模型与知识库类型 +知识库支持与多种预置及自定义模型协同工作。**预置模型**包括千问全系（QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research、VL-Max/Plus/Flash/OCR、开源版 Qwen3/Qwen2.5/Qwen2 等）以及第三方模型（DeepSeek-R1、Llama3.1、Yi-Large 等）；**自定义模型**指在百炼平台调优后的千问系列模型（如 Plus/Turbo/VL-Max 等），其兼容性以 [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) 页面实际可选为准 [知识库 (raw/application-user-guide/knowledge-base/rag-knowledge-base.md)](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 -预置模型（千问-QwQ/Long/Max/Plus/Turbo/Coder、千问VL 系列、Qwen 开源版，以及 DeepSeek-R1/V3.1、Llama3.1 等第三方文本生成模型）和部分调优后的自定义模型均可挂载知识库。具体可选模型以[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)页面创建应用时实际可选项为准。 +知识库提供三类核心服务： +- **知识检索**：支持单库或多库（最多 15 个）联合检索，具备 Query 改写、混合检索（向量 + 关键词）、Rerank 排序与多模态（文本/图片/音视频）能力 [知识检索 (raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md)](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md)； +- **知识问答**：在检索基础上集成大模型生成，支持极速模式（单轮）与多轮智能模式（Agentic 规划），并提供文件预解析、拒答、防泄漏、引用溯源等生成控制能力 [知识问答 (raw/application-user-guide/knowledge-base/rag-knowledge-qa.md)](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md)； +- **日志与监控**：所有检索调用自动投递至 SLS 日志服务，支持用量统计、错误分析与性能告警 [知识库日志与监控 (raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md)](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md)。 -创建知识库时按场景选择类型（创建后不可更改）： - -- **文档搜索**：企业内部文档、产品手册等非结构化数据检索。可细分为基础文档问答、图文并茂回复、视觉理解（富文本文档）、极速问答四种使用场景。选择视觉理解后向量模型自动切换为 qwen3 多模态向量（qwen3-vl-embedding），不可更改。 -- **数据查询（表格库）**：结构化 Excel/CSV，单库仅支持 1 篇。 -- **图片问答类**：仅支持 multimodal-embedding-v1 向量模型。 -- **音视频搜索类**：支持语音识别、视频帧提取与剧情解析。 - -不同解析方式（电子文档解析、文档智能解析、大模型文档解析、Qwen VL 解析、音视频解析）在速度与图表理解能力上有明显差异，详见[知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +> **注意**：文档 1 中列出的“千问-开源版（Qwen3、Qwen2.5、Qwen2等）”在文档 4 的 API 指南中未明确提及支持，且文档 4 明确声明“本文档仅适用于文档搜索类知识库”，而文档 1 将其列为通用支持模型。实际集成时请以控制台创建应用时的模型列表为准，避免依赖过时文档描述。 ## 关键参数 -知识库的检索效果主要由以下参数决定，在命中测试、检索服务和问答服务中可反复调优： - -- **相似度阈值（0.01~1.0）**：仅语义相似度高于阈值的切片会被召回。阈值过高会导致相关切片被全部丢弃（例如调至 0.60 可能返回无召回结果）。 -- **初步向量检索 TopK / 初步关键词检索 TopK（1~100，默认各 50）**：控制初步召回的切片数量，直接影响送入 Rerank 模型的 Token 量与成本。 -- **最大召回数量 / 召回片段数（1~20）**：即多路召回的 K 值，最终提供给大模型的切片数。对总结、列举、比较类复杂问题应适当调大。 -- **权重**：多知识库联合召回时用于干预排序，但**仅在同类型知识库之间生效**。 -- **排序模型（Rerank）**：纯文本可选 qwen3-rerank、qwen3-rerank(hybrid)；多模态可选 qwen3-vl-rerank。支持问答模式与相似模式。 -- **Meta 信息抽取与标签过滤**：通过元数据（常量/变量/大模型/正则/关键词方式提取）和标签在向量检索前做结构化筛选，精准定位目标文件。注意元数据只能在创建时配置，创建后无法再开启。 +| 参数类别 | 参数名 | 取值范围 | 说明 | +|----------|--------|----------|------| +| **检索控制** | 相似度阈值 | 0.01–1.0 | 过滤排序后分数低于该值的切片；值过高易漏召，过低引入噪声 [知识库 (raw/application-user-guide/knowledge-base/rag-knowledge-base.md)](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 | +| | 初步向量检索 TopK / 初步关键词检索 TopK | 1–100 | 控制各阶段初步召回切片数；影响 Rerank 模型费用（费用 = 初步召回总切片数 × 平均 [Token](../concepts/token.md) 数 × 单价） [知识库计费说明 (raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)。 | +| | 最大召回数量 | 1–20 | Rerank 后最终返回给大模型的切片数；工作流中对应 `TopK` 配置。 | +| **元数据与过滤** | Meta信息抽取 | — | 创建知识库时一次性配置，支持常量、变量（`file_name`/`cat_name`）、大模型提取、正则、关键词搜索五种方式；启用后可显著提升跨文档精准召回能力 [知识库 (raw/application-user-guide/knowledge-base/rag-knowledge-base.md)](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 | +| | 标签过滤 | — | 单文件最多 32 个标签，支持上传时设置或后期编辑；调试时可在“召回策略”页签启用，实现基于标签的前置筛选。 | ## 使用方式 -**控制台快速构建**：进入[知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base)选择规格后，按"填写基础信息 → 配置数据来源 → 设置索引参数"三步完成创建，随后可关联到智能体应用、工作流应用或外部应用。工作流应用需将知识库节点接在开始节点之后、大模型节点之前，并在大模型提示词中引用 `result` 变量。 - -**知识检索服务**：面向多知识库联合检索（最多 15 个），提供 Query 改写、混合检索（向量+关键词）、Rerank 排序的流水线，支持知识库路由、混排模型模式等全局与单库独立参数配置，详见[知识检索](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md)。 - -**知识问答服务**：在检索基础上由大模型（如 qwen3.6-plus）生成自然语言回答，提供**极速模式**（单轮检索+生成）和**多轮智能模式**（Agentic 多轮规划搜索），并支持文件预解析、拒答、防泄漏、多模态回复、引用来源等生成控制。 - -**API/SDK 集成**：通过[阿里云百炼 SDK](https://api.aliyun.com/api-tools/sdk/bailian?version=2023-12-29)调用，典型创建流程为：申请上传租约（ApplyFileUploadLease）→ 上传文件 → AddFile → CreateIndex → SubmitIndexJob → 轮询 GetIndexJobStatus。子账号需先获取 AliyunBailianDataFullAccess 策略并加入业务空间，详见[知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md)。 - -## 效果优化 - -RAG 效果由建立索引、检索召回、生成答案三个阶段决定。优化前建议用[自动评测](https://help.aliyun.com/zh/model-studio/application-auto-evaluation)建立至少 100 组用例的评估基线，再针对失败用例（打分 < 4）诊断改进： - -- **检索无效（没找到）**：补充知识、优化源文件排版（推荐转 Markdown、移除水印、避免复杂表格）、统一实体表述、启用多轮对话改写。 -- **召回不相关**：使用标签过滤或元数据做结构化搜索。 -- **切片不完整**：采用"智能切分"（基于语义自适应切分），并人工检查修正异常切片。 -- **重排不佳**：调整相似度阈值与召回片段数，在漏召回与噪声之间平衡。 - -## 日志与监控 +知识库可通过三种方式集成： +1. **控制台零代码集成**：在[知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base)页面创建标准版/旗舰版知识库 → 上传文件（支持 PDF/DOCX/TXT/图片等，详见 [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)）→ 在智能体或工作流应用的“文档知识库”节点中添加并配置相似度阈值、权重等参数； +2. **工作流节点集成**：拖入“知识库”节点，配置 `content` 输入为 `query`，选择固定知识库或动态引入（`CodeList` 变量），设置 `TopK`，再连接大模型节点并在提示词中插入 `{result}` 变量； +3. **API 集成**：需子账号获取 `AliyunBailianDataFullAccess` 权限、加入业务空间、配置 AccessKey 与 `WORKSPACE_ID`，使用 SDK 调用 `CreateIndex`、`Retrieve` 等接口；完整示例见 [知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md)。 -所有检索调用都会以日志形式投递到日志服务（SLS），topic 为 `log_dispatch`，包含 `request_id`、`pipeline_id`（知识库 ID）、`workspace_id`、`latency`、`response_status_code`、`response_code`、`request_body`、`response_body` 等索引字段，可用于调用审计、用量统计、慢查询与错误率监控。首次使用需在知识库列表页的**监控配置**中授权 SLS 角色、开通日志服务并创建 LogStore。SLS 存储与流量单独计费，关闭检索日志开关只停止新投递，历史日志仍保留计费，需彻底停止请到 SLS 删除对应 LogStore。 +## 限制和注意事项 -## 限制与配额 - -| 类别 | 上限 | -| --- | --- | -| 知识库数量 | RDS 数据源 100，其它数据源无限制 | -| 存储容量 | 旗舰版 9,999 GB / 标准版 100 GB | -| 单个文档搜索类知识库文件数 | 无硬性上限（数据查询类仅 1 篇） | -| 单次控制台导入文件数 | 50（API 批量建议单次 ≤ 10,000） | -| 单文件标签数 | 32 | -| 文本切片长度 | 6,000 Token | -| 召回文本切片数量 | 20 | -| 检索并发 | 旗舰版 50-10,000 QPS（1-200 RCU）/ 标准版 1 QPS 固定 | - -文件格式限制：pdf/docx/ppt 等最大 150MB 且页数 ≤ 1,000；txt/markdown/html 最大 10MB；图片最大 20MB；音视频最大 512MB。向量模型仅支持 text-embedding-v3/v4（512 维）与 multimodal-embedding-v1（1024 维），维度不可更改。完整清单见[知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)。 - -## 计费注意事项 - -知识库服务自 **2026 年 1 月 4 日**起正式计费，总费用由**规格费用**（运行时长）和**模型调用费用**（向量化 + Rerank）两部分构成，扣费顺序为免费额度 > 资源包 > 按量付费。 - -- 规格费用：标准版 0.03 元/知识库/小时；旗舰版 0.2 元/RCU/小时（1 RCU ≈ 50 QPS）。 -- 模型调用费用独立计费，公式为 `(输入 Token 总数 / 1000) × 模型单价`。**Rerank 费用取决于初步召回的总切片数，而非最终返回数**，因此降低初步 TopK 或关闭排序可显著省钱。 -- 应用挂载多个知识库时，模型 Token 消耗按知识库数量倍增（N 个库则 × N）。 -- 平台提供一次性 720 小时免费额度（仅抵扣标准版规格费用，不含模型调用费）。删除知识库以停止计费，但删除会**永久清除数据且无法恢复**。 - -> **注意**：免费额度有效期存在新老用户差异——老用户统一截至 2026 年 2 月 3 日 23:59，新用户自开通起 30 天内有效，逾期作废。此外《知识库计费说明》以 2026 年计费规则描述，而《知识库》文档仍按旧的即时计费口径（0.03/0.2 元/小时）介绍创建流程，接入时以[知识库计费说明](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)为准。 +- **地域限制**：知识库功能**仅支持华北2（北京）地域**，新加坡、法兰克福等国际地域不可用，此限制在文档 1 和文档 4 中均被明确强调； +- **配额限制**：标准版知识库免费存储上限 100 GB，旗舰版 9,999 GB；单次控制台导入文件数上限 50 个（API 无此限制）；音视频搜索类知识库**不支持新增切片**，仅支持编辑与删除 [知识库配额与限制 (raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)； +- **计费要点**：费用由两部分构成——**规格费用**（标准版 0.03 元/小时，旗舰版按 RCU 计费）与**模型调用费用**（向量化、Rerank、路由、问答生成均按 [Token](../concepts/token.md) 单独计费）；Rerank 费用取决于初步召回总切片数，而非最终返回数；关闭 Rerank 可降本但牺牲精度 [知识库计费说明 (raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)； +- **配置不可变性**：知识库类型（文档搜索/数据查询/图片问答）与 Meta 信息抽取配置**创建后不可修改**，需重建知识库；多轮对话改写功能仅能在创建时开启，后续无法补开 [RAG效果优化 (raw/application-user-guide/knowledge-base/rag-optimization.md)](../../raw/application-user-guide/knowledge-base/rag-optimization.md)。 ## 来源文档 -- [知识库日志与监控](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) - [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md) +- [知识库日志与监控](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) - [RAG效果优化](../../raw/application-user-guide/knowledge-base/rag-optimization.md) -- [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - [知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) +- [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - [知识库计费说明](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - [知识检索](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识问答](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md b/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md index cf82ffab..f5e88fe0 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md @@ -1,124 +1,68 @@ # llm application -阿里云百炼平台提供三种核心应用构建模式：智能体（Agent）、工作流（Workflow）和高代码应用，用于突破大模型在私有知识访问、实时信息获取和复杂任务规划方面的原生局限。开发者可根据开发门槛、控制粒度和业务场景选择合适的应用类型，并通过集成知识库、MCP 工具、插件等能力构建完整的 AI 应用。 +百炼平台的 LLM Application 是面向业务场景的 AI 应用构建范式，旨在突破大语言模型在私有知识接入、实时信息获取、流程控制与复杂任务规划等方面的原生局限。通过智能体（Agent）、工作流（Workflow）和高代码应用三种模式，开发者可按需选择零代码、低代码或专业编码方式，快速构建具备知识增强、工具调用、多步推理与企业级运维能力的生产级 AI 服务。 -## 应用类型与选型 +## 支持的模型/功能 -| 对比维度 | 智能体（Agent） | 工作流（Workflow） | 高代码应用 | -|---------|---------------|-------------------|-----------| -| 开发方式 | 自然语言配置（零代码） | 可视化节点编排（低代码） | Python 编码 | -| 核心特点 | AI 自主决策、动态规划 | 预定义流程精确控制 | 完全由代码控制 | -| 适合人群 | 业务人员、产品经理 | IT 运维、业务分析师 | AI 工程师、开发者 | -| 开发门槛 | 低 | 中 | 高 | +百炼 LLM Application 支持三类核心构建模式，各自适配不同技术栈与业务复杂度： -详细的类型介绍参见 [应用类型介绍](../../raw/application-user-guide/llm-application/application-introduction.md)。 +- **智能体（Agent）应用**：以提示词驱动，支持自主意图理解、多步规划与工具调度。新版智能体（Agent 2.0）将知识库、MCP 服务统一为可规划调用的工具，显著提升过程透明性与任务泛化能力；旧版（Agent 1.0）则采用“先检索后决策”的串行逻辑，适合意图单一、流程固定的场景。详细对比见 [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 -## [智能体应用](../concepts/agent-application.md) +- **工作流（Workflow）应用**：基于可视化节点编排，支持大模型节点、意图分类、变量处理、智能体群组等十余类节点，实现确定性、可复现的多步骤自动化。典型场景包括诈骗识别、智能导购、日程管理等，其执行链路完全由预定义逻辑控制，不依赖模型动态规划 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 -### 新版智能体（Agent 2.0） +- **高代码应用**：面向专业开发者，提供完整的 Python 工程化部署能力，支持 Serverless Function 与 K8s 两种运行时，内置 MCP 工具接入、前端定制（Spark Design）、API 网关与可观测性等企业级能力 [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 -新版智能体将知识库、MCP 等能力统一为工具，由智能体自主规划调用顺序，支持完整的"规划-执行-反思"链路展示。推荐在无旧版依赖时使用新版。 +所有模式均支持主流千问系列模型（如 Qwen-Max、Qwen-Plus、Qwen-VL 系列），并兼容部分 DeepSeek 及开源模型。文件问答能力覆盖文档、图片、音视频，提供全文引用、切片检索（RAG）和自定义处理三种模式，具体支持格式与参数详见 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 -核心能力配置： +> **注意**：文档 2 与文档 3 对智能体能力描述存在关键差异——文档 2 将知识库与插件视为独立能力模块，而文档 3 明确将其统一为“工具”并纳入 ReAct 规划调度体系。实际开发应以 [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md) 的架构为准，旧版已不推荐新项目使用。 -- **模型选择**：推荐具备强工具调用能力的模型（如千问-Max 系列）。支持配置最长回复长度、temperature、enable_thinking（思考模式）等参数。 -- **提示词**：定义角色、行为指令与能力边界，支持自定义变量嵌入。 -- **内置工具**：沙箱环境中的 bash、write、read、edit、glob、grep、download_file 等工具，默认关闭需按需开启。 -- **知识库**：作为工具由智能体自主调用，支持标签过滤限定查询范围。 -- **MCP**：外部工具以 MCP 协议接入，支持动态非固定顺序调用。 -- **记忆**：短期记忆支持 0-30 轮上下文；长期记忆暂未支持。 -- **ReAct 最大轮次**：取值 1-50，限制单次会话中工具调用最大次数。 +## 关键参数 -详细配置方法参见 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 +| 类别 | 参数名 | 说明 | 可配置位置 | +|--------|---------|------|-------------| +| **模型层** | `temperature` | 控制生成随机性，取值 0–2，推荐 0.1–0.7 保证稳定性 | 智能体/工作流节点的模型参数配置器 | +| | `enable_thinking` | 是否开启思考模式（仅限支持模型），影响规划链路可视化 | [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md) 的模型参数配置器 | +| | `ReAct 最大轮次` | 单次会话中工具调用最大次数（1–50），超限则终止调用并生成终局回复 | 新版智能体「运行与结果分析」区域 | +| **文件处理** | `单文件最大解析长度（token）` | 全文引用模式下，单个文件提取 token 上限，超出从末尾截断 | [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md) 的全文引用配置页 | +| | `召回片段数` / `最大拼装长度` | 切片检索模式下，控制 RAG 检索精度与上下文开销 | [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md) 的切片检索配置页 | +| **会话控制** | `短期记忆轮数` | 新版智能体支持 0–30 轮上下文传递，0 表示禁用多轮记忆 | [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md) 的「记忆」配置项 | +| | `historyList` / `imageList` | 工作流预置会话变量，用于跨节点传递对话历史与图片列表 | [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md) 的「开始/结束」节点说明 | -### 旧版智能体（Agent 1.0） +## 使用方式 -旧版智能体通过知识库（RAG）和插件扩展能力，适合意图单一、流程固定的简单任务。知识库检索后再决策是否调用其他工具。 +### 创建与配置 +- **智能体**：控制台 → 应用管理 → 创建应用 → 选择「智能体应用」→ 优先选用 Agent 2.0；配置模型、系统提示词、知识库（作为工具）、MCP 服务及技能。 +- **工作流**：控制台 → 应用管理 → 创建应用 → 选择「工作流应用」→ 拖拽节点（开始/大模型/意图分类/结束等）→ 连线配置 → 启用「自定义缓存」以支持多轮上下文。 +- **高代码应用**：控制台 → 创建应用 → 选择「高代码应用」→ 选择模板或上传 `.whl` 包 → 配置部署方式（Serverless/K8s）与资源规格 → 一键部署。 -> **注意**：新版智能体与旧版智能体基于不同技术架构，不支持直接升级或版本切换。需要迁移时必须重新创建新版应用。 +### 发布与调用 +- 所有应用**必须发布后方可调用**（未发布状态仅支持控制台调试）。发布操作位于应用配置页右上角「发布」按钮，发布前需确认 RAM 权限（如 `ram:CreateServiceLinkedRole`）。 +- API 调用路径统一：应用详情页 → 「发布渠道」页签 → 「API 调用」→ 查看 endpoint 与鉴权方式（Bearer [Token](../concepts/token.md) + API Key）。 +- 文件上传支持三种方式：聊天窗口直接上传（会话级有效）、`file_list`/`image_list` 传公网 URL（需 OSS 等可公开访问）、调用文件上传 API 获取 `session_file_id`（推荐生产环境使用）。 -旧版智能体的自定义插件有 5 秒超时限制。详情参见 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)。 +### 集成扩展 +- 智能体与工作流均可通过「应用组件」能力复用已发布应用（如将智能体嵌入工作流作为子节点）。 +- 高代码应用支持通过「工具」Tab 一站式关联知识库、MCP 服务，并在代码中调用 `agentscope` SDK 实现深度集成。 +- 所有模式均支持发布至钉钉、微信公众号等第三方渠道，或通过 API 网关暴露为标准 RESTful 接口。 -## 工作流应用 +## 限制和注意事项 -工作流通过可视化节点编排将多步骤任务串联为稳定可控的执行链路，适合固定流程自动化场景。 - -主要节点类型： - -- **开始/结束节点**：定义输入输出参数结构，预置 query、historyList、imageList 变量 -- **大模型节点**：配置模型、提示词和用户提示词 -- **意图分类节点**：根据用户输入分发到不同处理分支 -- **智能体群组节点**：将子智能体作为工具组合调用 -- **变量处理节点**：文本输出或变量转换 - -工作流支持会话变量作为全局参数在节点间传递，支持记忆功能（本节点缓存或自定义缓存）。 - -创建和配置详情参见 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 - -## 高代码应用 - -面向专业开发者，支持基于 Python 项目部署 AI 后端服务。 - -关键特性： - -- **部署方式**：Serverless Function（无状态快速拉起）和 K8s（高性能有状态长程任务） -- **MCP 工具接入**：控制台直接关联知识库、工作流、插件等 MCP 服务 -- **前端体验**：支持直接体验、自定义交互卡片、基于 Spark Design 的自定义 WebUI -- **企业级能力**：自动化运维、可观测、日志服务、API 网关 -- **代码提交**：支持控制台模板创建或命令行上传 .whl 代码包 - -生产环境建议开启网关功能，通过自定义域名访问。时延敏感业务建议最小实例数大于等于 1。 - -详细开发和部署流程参见 [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 - -## 文件问答 - -[智能体应用](../concepts/agent-application.md)支持上传文件进行智能问答，提供三种处理模式： - -| 模式 | 适用场景 | 特点 | -|------|---------|------| -| 全文引用 | 文档总结、全文翻译 | 简单直接，受上下文长度限制 | -| 切片检索（RAG） | 长文档问答、知识库检索 | 能处理超长文件，效果依赖检索策略 | -| 自定义处理 | 图片转换、视频分析等需工具介入的任务 | 功能灵活，依赖配置的工具 | - -文件限制：单会话最多 10 个文件，单文件不超过 10MB。超过 10MB 需使用文件上传 API。 - -支持格式：文档（doc/docx/pdf/md/txt 等）、图片（png/jpg/bmp/gif）、视频（mp4/mkv/avi 等）、音频（mp3/wav/flac 等）。 - -详细使用方式参见 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 - -## 发布与调用 - -所有应用类型均需先发布才能通过 API 集成： - -1. 在应用配置页点击"发布"，确认变更后完成发布 -2. 在"发布渠道"页签查看 API 调用方式 -3. [智能体应用](../concepts/agent-application.md)还支持发布到钉钉、微信公众号等第三方平台 - -RAM 账号发布前需确认拥有 `ram:CreateServiceLinkedRole` 权限。 - -## 计费说明 - -- **模型调用**：按模型类型和 [Token](../concepts/token.md) 用量计费 -- **知识库**：按量付费，召回的文本切片会增加输入 [Token](../concepts/token.md) -- **MCP/插件**：部分官方 MCP 按调用计费，第三方 MCP 费用由第三方收取 -- **高代码应用**：部署后函数计算、API 网关、存储均按量计费 -- **文件上传**：上传本身不收费，问答消耗按所选模型标准计费 +- **文件限制**：单次会话最多上传 10 个文件，单文件 ≤10MB；聊天窗口上传的文件仅在当前会话有效，刷新即失效；通过 `session_file_id` 上传的文件有效期为 24 小时。 +- **调用限频**：每个智能体应用默认限流 100 次/分钟，该配额被所有 API 请求共享（含文件问答、普通对话等）。 +- **模型兼容性**：`enable_thinking` 参数仅对 Qwen-Max 等明确标注支持思考模式的模型生效；千问-VL 系列模型在「自定义处理」模式下可直接解析图片，无需开启预解析。 +- **计费要点**： + - 模型调用费用按输入/输出 [Token](../concepts/token.md) 计费，RAG 检索内容、[长期记忆](../concepts/long-term-memory.md)体、文件解析文本均计入输入 [Token](../concepts/token.md)； + - 知识库、MCP 服务、工具调用可能产生独立费用（如第三方 API 费用由服务商收取）； + - 上下文缓存仅支持隐式缓存（自动生效，按 20% 输入单价计费），暂不支持显式缓存配置。 +- **版本隔离**：Agent 1.0 与 Agent 2.0 架构不兼容，无法升级或降级，需重新创建应用 [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 ## 来源文档 - [应用类型介绍](../../raw/application-user-guide/llm-application/application-introduction.md) -- [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md) - [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md) -- [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md) +- [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md) - [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md) +- [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md) - [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md) - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md index 6ad7d7d1..23aa390d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md @@ -1,62 +1,44 @@ # managed agents -Managed Agents 是百炼提供的智能体托管运行时，面向多步工具调用、代码执行、文件处理等长时运行任务。与无状态的[智能体应用](../concepts/agent-application.md)不同，它由平台在服务端托管会话状态、沙箱环境与工具执行，智能体在独立云端容器中自主执行命令、读写文件、安装依赖，事件历史在服务端持久化并支持中断与续接。 +Managed Agents 是百炼平台提供的智能体托管运行时，专为多步工具调用、代码执行、文件处理等长时运行任务设计。平台统一托管会话状态、沙箱环境与工具执行生命周期，智能体在隔离的云端容器中自主执行命令、读写文件、安装依赖，并通过服务端持久化的事件流反馈执行过程。相比无状态的智能体应用，Managed Agents 支持中断续接、跨轮次上下文保持与沙箱级状态隔离。 -## 与[智能体应用](../concepts/agent-application.md)的区别 +## 支持的模型与功能 -| 维度 | [智能体应用](../concepts/agent-application.md) | Managed Agents | -| --- | --- | --- | -| 运行模式 | 无状态调用，应用侧维护上下文 | 服务端维护会话状态，支持中断与续接 | -| 执行环境 | 共享运行时 | 独立沙箱，云端容器 | -| 事件模型 | 响应级[流式输出](../concepts/streaming.md) | 会话级 SSE 事件流，事件历史持久化 | -| 典型场景 | 问答、对话、轻量任务 | 多步工具调用、代码执行、文件处理等长时任务 | +- **模型支持**：当前支持 `qwen3-max`、`qwen3.7-plus` 等 Qwen 系列大模型（详见 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)）；模型需通过 `model.id` 字段显式指定，不支持动态路由或 fallback 模型。 +- **内置工具集**：默认提供 7 个基础工具：`bash`（命令执行）、`read`/`write`/`edit`（文件操作）、`glob`/`grep`（文件搜索）、`download_file`（URL 下载）。所有工具均在沙箱内受限执行，无 root 权限。 +- **扩展能力**： + - 可集成 MCP 服务（通过 `mcp_servers` 字段注册外部工具服务）； + - 可挂载 Skill 封装端到端流程（如数据清洗、PDF 解析）； + - 支持挂载用户上传文件或远程 URL 资源（见 [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)）。 -## 核心概念 +> **注意**：文档 2 中示例使用 `qwen3-max`，但文档 1 的概述表格中仅列出 `qwen3.7-plus` 作为示例模型。实际可用模型以控制台下拉列表或 [API 文档](https://help.aliyun.com/zh/model-studio/agent-create) 为准，建议以 `qwen3-max` 为首选生产模型。 -四个核心对象构成完整的运行链路，详见 [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md)： +## 关键参数 -- **智能体（Agent）**：模型、系统提示词、工具、MCP 服务和 Skill 的组合配置。创建后通过 ID 引用，可在多个会话中复用。 -- **运行环境（Environment）**：会话运行的沙箱配置，由百炼托管的云端容器，独立于智能体管理，可被多个会话复用。 -- **会话（Session）**：智能体在指定环境中的一次运行实例，承载任务执行与输出。 -- **事件（Event）**：应用与智能体之间交换的消息，包括用户消息、工具调用结果和状态变更。 - -## 支持的工具 - -智能体通过以下工具与运行环境交互： - -- **命令执行**：在沙箱中运行 shell 命令（`bash`）。 -- **文件操作**：`read`、`write`、`edit`、`glob`、`grep`，以及从 URL 下载的 `download_file`；也可上传本地文件挂载到沙箱。 -- **MCP 服务**：接入外部工具服务，详见 [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md)。 -- **Skill**：挂载预置的工具组合，封装端到端任务流程。 - -快速开始阶段默认全选 7 个内置工具（`bash`、`read`、`write`、`edit`、`glob`、`grep`、`download_file`），可按需取消勾选。 +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `name` | string | 是 | 智能体名称，仅用于标识，不影响行为 | +| `model.id` | string | 是 | 模型 ID，如 `"qwen3-max"`，不可省略 | +| `system` / `system_prompt` | string | 是 | 系统提示词，定义角色与行为边界，直接影响工具调用倾向 | +| `tools` | array | 否（默认全启用） | 工具配置数组，如 `[{"type": "builtin_toolkit"}]`；禁用某工具需显式排除 | +| `environment_id` | string | 创建会话时必填 | 指向已创建的运行环境 ID，决定沙箱配置与预装依赖 | +| `resources` | array | 否 | 挂载资源列表，格式为 `[{"resource_id": "...", "mount_path": "/mnt/session/uploads/data"}]` | ## 使用方式 -典型流程分为四步（控制台向导或 API 均可完成），参见 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)： - -1. **配置智能体**：指定名称、模型（如 `qwen3-max`）、系统提示词与工具。API 为 `POST /api/v1/agentstudio/agents`。 -2. **配置运行环境**：默认云端托管沙箱，可通过 `config.packages` 预装 apt / pip 依赖并设置网络策略。API 为 `POST /api/v1/agentstudio/environments`。 -3. **发起会话**：绑定智能体 ID 与环境 ID 创建会话实例。API 为 `POST /api/v1/agentstudio/sessions`。 -4. **发送事件并接收响应**：向会话写入用户消息触发处理（`POST /sessions/{id}/events`），通过 SSE 事件流实时接收工具调用过程与输出（`GET /sessions/{id}/events/stream`）。 - -控制台的**预览调试**标签页可直接对话并按事件类型（User、Agent、Tool、Tool_output、Error、Model、System）筛选查看执行过程。 - -### 上下文与资源挂载 - -上下文中的挂载资源独立于会话管理，可被多个会话复用，详见 [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)： - -- **挂载时机**：创建会话时在 `resources` 字段指定，或运行时通过 `POST /sessions/{session_id}/resources` 追加，实时生效且无需重启会话。 -- **路径约定**：挂载资源统一放在 `/mnt/session/uploads` 前缀下，可在系统提示词中直接引用完整路径。 -- **会话隔离**：平台为挂载资源做内部拷贝放入沙箱，会话内的修改不影响原始资源，也不影响挂载同一资源的其他会话；卸载后副本被清理。 +1. **创建智能体**：调用 `/api/v1/agentstudio/agents` 接口，传入 `name`、`model.id`、`system` 和 `tools`（参考 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md) 中的 API 示例）。 +2. **创建运行环境**：调用 `/api/v1/agentstudio/environments`，指定 `config.type="cloud"` 及 `packages`（apt/pip）、`networking` 等沙箱配置（见 [配置 Agent 环境](../../raw/application-user-guide/managed-agents/managed-agents-environment.md)）。 +3. **发起会话**：调用 `/api/v1/agentstudio/sessions`，绑定 `agent` ID 与 `environment_id`，可选传入 `title` 和 `resources`。 +4. **交互与监听**： + - 发送用户消息：`POST /sessions/{session_id}/events`，`input` 为标准 message 数组； + - 实时接收事件：`GET /sessions/{session_id}/events/stream`，使用 SSE 流解析 `message`、`tool_output`、`session_status` 等事件类型（见 [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md)）。 -## 限制与注意事项 +## 限制和注意事项 -- 单个上传文件不超过 **10 MB**。 -- 沙箱内文件路径遵循 `/mnt/session/uploads` 约定，代码中引用文件应使用该完整路径。 -- 会话状态、中断续接与工具审批由会话状态机管理，详见 [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md)。 - -> **注意**：文档中出现的模型名称不一致——控制台向导示例填写 `qwen3.7-plus`，而 API 代码示例使用 `qwen3-max`。请以控制台模型下拉列表中实际可选的模型 ID 为准。 +- **沙箱限制**：单次命令执行超时为 60 秒；内存上限 4 GB；磁盘空间上限 10 GB；网络访问默认受限（需显式配置 `networking.type="unrestricted"` 才允许外网请求）。 +- **文件限制**：上传挂载的单个文件 ≤ 10 MB；沙箱内文件路径必须以 `/mnt/session/uploads/` 开头，硬编码路径在系统提示词中需严格匹配（见 [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)）。 +- **状态持久性**：会话事件历史在服务端持久化，但沙箱文件系统**不跨会话保留**；同一资源被多个会话挂载时，各会话获得独立副本，修改互不影响。 +- **权限隔离**：所有工具执行均在非 root 用户下运行，无法执行 `sudo`、`apt install` 等特权操作；依赖须在环境创建阶段预装，运行时不可动态安装。 ## 来源文档 @@ -64,12 +46,7 @@ Managed Agents 是百炼提供的智能体托管运行时，面向多步工具 - [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md) - [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md) - [配置 Agent 环境](../../raw/application-user-guide/managed-agents/managed-agents-environment.md) -- [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md) - [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md) - - - - - +- [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md index 3ee13beb..553bcdf5 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md @@ -1,166 +1,46 @@ # memory library overview -百炼记忆库（Memory Library）通过长期记忆 API 解决大模型跨会话上下文丢失的问题：自动从对话中提取关键信息并持久化存储，再在后续对话中基于语义检索召回相关记忆注入 Prompt，使智能体能够持续理解用户偏好与历史信息。该能力既可在百炼控制台可视化管理，也提供开放的 HTTP API 接入任意应用，并支持通过 OpenClaw 插件以"自动捕获 / 自动召回"的方式零侵入接入 Agent。详见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)、[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 与 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 +记忆库是百炼平台提供的[长期记忆](../concepts/long-term-memory.md)能力核心组件，用于突破大模型上下文窗口限制，实现跨会话、跨对话的语义化记忆持久化与智能召回。它通过自动从对话中提取关键信息（记忆片段）或结构化属性（用户画像），并基于向量检索技术在后续交互中精准召回相关记忆，从而支撑个性化、连贯的智能体体验。该能力以开放 API 形式提供，可集成至任意应用或 Agent 框架。 -## 核心能力 +## 支持的模型/功能 -记忆库提供两类持久化记忆内容，二者可独立或组合使用： - -- **记忆片段**：从对话中自动提取的关键事件和信息（如"用户每天上午9点需要喝水提醒"），适用于大多数长期记忆场景。支持自动去重、动态更新，也可通过 `custom_content` 直接写入指定内容。 -- **用户画像**：基于自定义画像模板从对话中提取的结构化属性（如年龄、职业、偏好等），适用于需要固定属性持久化存储的场景。属性字段及描述应清晰具体，避免"姓名/名称/名字"等同义字段并存，且不应期望一次对话就提取全部信息。 - -> **注意**：记忆有效期在不同入口存在差异。[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 文档指出"生成的记忆片段与用户画像暂无失效日期"，而 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) 控制台的默认记忆片段规则预置了"默认有效期 180 天"，并支持按规则配置 7/30/180 天或永不过期。以控制台记忆规则配置为准；通过 API 直写且不指定 `project_id` 时使用默认规则。 - -## 接入方式 - -### 方式一：API 直连 - -通过 HTTPS 调用 `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` 系列接口，需在环境变量中配置 `DASHSCOPE_API_KEY`。典型流程为：对话结束调用 `AddMemory` 写入记忆 → 调用 `SearchMemory` 语义检索 → 将结果注入 Prompt。 - -```bash -# 写入记忆（从对话自动提取） -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午9点提醒我喝水"}, - {"role": "assistant", "content": "好的，已记录"} - ], - "user_id": "user_001" - }' - -# 语义检索记忆 -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "我需要做什么？"}], - "top_k": 5 - }' -``` - -Python 用户可安装 `agentscope-runtime`，使用 `AddMemory`、`SearchMemory`、`ListMemory`、`CreateProfileSchema`、`GetUserProfile` 等封装类（均需在 `finally` 中调用 `close()`）。 - -### 方式二：OpenClaw 记忆插件 - -OpenClaw Agent 可通过插件实现零侵入的[跨会话记忆](../concepts/cross-session-memory.md)。插件在 Gateway 内通过 `before_agent_start`（自动召回）和 `agent_end`（自动捕获）两个生命周期钩子与长期记忆 API 交互，所有读写均由百炼服务端完成提炼、向量化和语义检索。 - -```bash -# 安装 -openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw - -# 验证 -openclaw plugins info modelstudio-memory-for-openclaw -openclaw modelstudio-memory stats -openclaw gateway restart -``` - -插件配置写入 `~/.openclaw/openclaw.json`，关键项：`slots.memory` 注册为记忆槽位（会自动禁用内置 `memory-core` 和 `memory-lancedb`）；`apiKey` 填 DashScope [API Key](../concepts/api-key.md)；`userId` 用于隔离不同用户记忆空间。 - -> **注意**：OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置；不支持阿里云百炼 Coding Plan 的 [API Key](../concepts/api-key.md)。 +- **记忆片段（Memory Snippet）**：支持从多轮对话消息中自动提炼关键事件（如“每天上午9点提醒我喝水”），也支持直接写入自定义内容（`custom_content` 字段）。适用于大多数[长期记忆](../concepts/long-term-memory.md)场景 [原文标题](../../raw/application-user-guide/memory-library-overview/memory-library.md)。 +- **用户画像（User Profile）**：基于预定义 Schema 从对话中抽取结构化属性（如年龄、职业、爱好），支持字段级描述引导与初始值设置，适用于需固定属性建模的场景 [原文标题](../../raw/application-user-guide/memory-library-overview/memory-library.md)。 +- **自动捕获与召回**：OpenClaw 等 Agent 框架可通过插件实现 `autoCapture`（对话结束自动写入）和 `autoRecall`（对话开始前自动检索注入）闭环 [原文标题](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 +> **注意**：文档 3 声称“生成的记忆片段与用户画像暂无失效日期”，但文档 1 明确指出默认记忆片段规则有效期为 180 天，且控制台支持配置 7/30/180 天或永不过期。实际行为以控制台配置及 API 中 `expiration_time` 参数为准，文档 3 的表述已过时。 ## 关键参数 -### AddMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `messages` | 与 `custom_content` 二选一 | 对话内容，系统自动从中提取记忆片段 | -| `custom_content` | 与 `messages` 二选一 | 直接指定要存入的记忆内容，不经过对话提炼 | -| `user_id` | 是 | 记忆空间用户标识，同 `user_id` 共享命名空间，不同 `user_id` 完全隔离 | -| `memory_library_id` | 否 | 记忆库 ID，不填使用默认记忆库 | -| `project_id` | 否 | 记忆片段规则 ID，不填使用默认规则 | -| `profile_schema` | 否 | 用户画像规则 ID，传入后同时提取画像 | -| `meta_data` | 否 | 自定义元数据，用于分类管理 | - -### SearchMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 记忆空间用户标识 | -| `messages` | 是 | 查询对话内容 | -| `memory_library_id` | 否 | 限定检索的记忆库 | -| `top_k` | 否 | 返回记忆条数，建议 3–10 | - -### OpenClaw 插件配置项 - -| 配置项 | 类型 | 默认值 | 说明 | -| --- | --- | --- | --- | -| `apiKey` | string | - | 必填，以 `sk-` 开头 | -| `userId` | string | - | 必填，记忆空间用户标识 | -| `autoCapture` | boolean | `true` | 对话后自动提取并存储记忆 | -| `autoRecall` | boolean | `true` | 对话前自动检索并注入记忆 | -| `topK` | number | `5` | 每次召回返回的记忆条数 | -| `minScore` | number | `0` | 最小相似度阈值（0–100） | -| `profileSchema` | string | - | 用户画像 ID | -| `memoryLibraryId` | string | - | 记忆库 ID | -| `projectId` | string | - | 记忆片段规则 ID | - -## 记忆库与记忆规则管理 - -每个账号自带一个无法删除的默认记忆库，预置一条"默认项目"记忆片段规则（默认有效期 180 天，可编辑但不可删除）。可按业务场景创建新记忆库并配置记忆规则，每个记忆库最多 50 条记忆片段规则和 50 条用户画像规则。 - -- **记忆片段规则**：定义从对话中提取关键事件和信息的策略，可选择默认或自定义规则指令，支持自动更新和过期时间（7/30/180 天或永不过期）。 -- **用户画像规则**：定义画像字段名称、描述和初始值。当用户尚未通过对话提供信息时，系统使用初始值作为属性值。 - -控制台记忆检索页支持配置最大召回数量（1–100）、意图判别召回（建议开启）、查询改写（口语化提问时开启）和排序（使用 `gte-rerank-v2` 模型，相似度阈值建议 0.5–0.7）。 - -## OpenClaw 插件工具 - -除自动捕获/召回外，插件向 Agent 注册四个可主动调用的工具： - -- **memory_search**：语义检索记忆库，返回相似度最高的记忆列表，适用于"之前讨论过什么"等回顾性问题。 -- **memory_store**：直接写入记忆，不经过对话提炼，适用于"记住我的服务器 IP 是 192.168.1.x"等显式记忆请求。 -- **memory_list**：分页列出当前 `userId` 下所有记忆条目。 -- **memory_forget**：按记忆 ID 删除指定记忆，通常先 `memory_search` 定位再删除。 +| 参数 | 类型 | 是否必填 | 说明 | +|------|------|----------|------| +| `user_id` | string | 是 | 记忆隔离的唯一标识，不同 `user_id` 数据完全隔离 | +| `memory_library_id` | string | 否 | 指定记忆库 ID；不填则使用默认记忆库 [原文标题](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) | +| `project_id` | string | 否 | 指定记忆片段规则 ID；不填则使用默认规则 | +| `profile_schema` | string | 否 | 用户画像 Schema ID；用于触发结构化属性抽取 | +| `meta_data` | object | 否 | 自定义元数据，支持按业务维度分类管理（如 `{"category": "reminder"}`） | +| `top_k` | number | 否 | 检索返回最大条数，默认 5（OpenClaw 插件）或未指定（API 默认值由服务端决定） | +| `min_score` / `similarity_threshold` | number | 否 | 相似度阈值（0.0–1.0），用于过滤低相关性结果；OpenClaw 插件用 `minScore`（0–100 整数），API 文档用小数制，需注意单位差异 | -CLI 等效：`openclaw modelstudio-memory search|list|stats`。 +## 使用方式 -## 配额与限制 +1. **准备环境**：配置 `DASHSCOPE_API_KEY` 环境变量，获取方式见 [获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 +2. **写入记忆**：调用 `AddMemory` 接口，传入 `messages`（对话历史）或 `custom_content`（直接内容），指定 `user_id` 及可选参数。Python 用户推荐使用 `agentscope-runtime` 封装的 `AddMemory` 工具 [原文标题](../../raw/application-user-guide/memory-library-overview/memory-library.md)。 +3. **检索记忆**：调用 `SearchMemory` 接口，传入自然语言查询（`query`）或 `messages`，系统执行语义检索并返回匹配的记忆节点。 +4. **集成 Agent**：OpenClaw 用户可安装 `@modelstudio/modelstudio-memory-for-openclaw` 插件，通过 `openclaw.json` 配置 `apiKey` 和 `userId` 即可启用全自动捕获与召回 [原文标题](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 +5. **管理与调试**：通过百炼控制台 [记忆库](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/memory/list) 查看、检索、编辑规则；支持在“记忆检索”标签页调试召回效果（含改写、排序、意图判别等开关）。 -长期记忆 API 速率限制（阿里云账号级别）： +## 限制和注意事项 -| API 操作 | 速率上限 | -| --- | --- | -| AddMemory（写入） | 120 次/分钟 | -| SearchMemory（查询） | 300 次/分钟 | -| 所有操作合计 | 3000 次/分钟 | - -性能指标：SearchMemory 端到端延迟 200–500ms；AddMemory 延迟 500–1000ms；自动捕获异步执行，不影响响应速度。该功能与 API 调用限时免费。 - -## 排错要点 - -- **OpenClaw 插件重启后状态为 not loaded**：检查 `openclaw.json` 中 `plugins.entries.modelstudio-memory-for-openclaw.enabled` 是否为 `true`，以及 `plugins.slots.memory` 是否指向该插件，修正后重新 `openclaw gateway restart`。 -- **日志出现 InvalidApiKey**：DashScope [API Key](../concepts/api-key.md) 无效或过期，到百炼控制台确认状态或重新创建；若用环境变量引用，确认 `DASHSCOPE_API_KEY` 已设置且 Gateway 进程可读取。 -- **查看插件日志**：日志按日期存储在系统临时目录，文件名 `openclaw-YYYY-MM-DD.log`，可用 `grep modelstudio-memory` 过滤。 - -> **注意**：[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 为新版本，相比旧版长期记忆 API 在延迟、自动提取、语义检索准确性和用户画像能力上均有改进，建议新接入直接使用新版接口。 +- **配额限制**：阿里云账号级别总调用上限为 3000 QPM；其中 `AddMemory` 不超过 120 QPM，`SearchMemory` 不超过 300 QPM [原文标题](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 +- **延迟特性**：`AddMemory` 端到端延迟约 500–1000ms，`SearchMemory` 约 200–500ms；自动捕获为异步执行，不影响主流程响应速度。 +- **ID 隔离原则**：`user_id` 是记忆空间的唯一隔离键，务必确保同一用户始终使用相同 `user_id`；不同 `user_id` 间数据不可见、不可交叉检索。 +- **Schema 兼容性**：用户画像字段名称应语义唯一（避免同义词混用如“年龄”/“岁数”），且需通过 `CreateProfileSchema` 显式创建后方可使用 `profile_schema` 参数 [原文标题](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)。 +- **默认记忆库限制**：每个账号自带一个不可删除的默认记忆库，但可编辑名称、描述及添加自定义规则；新业务建议创建独立记忆库以实现规则与数据隔离。 ## 来源文档 -- [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) +- [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md index 7b9ed189..20e75760 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md @@ -1,77 +1,40 @@ # model compression -模型压缩是百炼平台提供的量化功能，用于将全精度微调模型转换为低精度版本，在保持模型能力的前提下降低部署所需的 MU 规格，从而减少推理成本。该功能属于模型生产链路中的可选环节，位于模型调优与模型部署之间。需要注意的是，百炼平台的模型压缩特指量化，不涉及结构剪枝或知识蒸馏。 +模型压缩是百炼平台提供的量化能力，用于将全精度微调模型转换为低精度版本，在可控精度损失前提下显著降低推理部署所需的 MU 规格与成本。该功能属于模型生产链路中的可选环节（微调 → 压缩 → 部署），仅作用于百炼平台内完成的微调模型，不支持结构剪枝或知识蒸馏等其他压缩范式。压缩操作不可逆，且压缩后模型不可再微调或二次压缩。 -## 核心概念 +## 支持的模型与功能 -模型压缩通过量化技术降低模型参数精度来实现部署成本优化。完整的模型生产链路为：模型调优 → 模型压缩（可选）→ 模型部署。以 qwen3.5-flash-2026-02-23 微调模型为例，压缩前部署规格为 MU1*2（108 元/小时），压缩后降至 MU8*1（47 元/小时），部署成本节省约 56%。详细说明参见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 +- **支持模型**：仅限通过百炼平台完成微调训练的自定义模型（如 `qwen3.5-flash-2026-02-23`），不支持基础模型、第三方模型或 OSS 直接加载的模型。具体支持列表以控制台实时展示为准，详见 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 +- **核心功能**：仅提供量化（Quantization）一种压缩方式，包括静态量化与需校准数据的量化模板；不支持剪枝、蒸馏等技术，这一点在 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 的“功能概述”和“常见问题”中已明确强调。 +- **地域限制**：当前仅华北2（北京）地域可用。 -> **注意**：压缩操作不可逆。压缩后的模型不支持继续微调，也不支持二次压缩。如需调整，必须从上游全精度微调模型重新压缩。 +## 关键参数 -## 支持的模型 +创建压缩任务时需配置以下必填参数： +- **任务名称**：最长 50 字符，建议包含模型简称、量化方式与版本号（如 `qwen35f-awq-mu8`）； +- **量化产出模型名后缀**：仅支持小写字母和数字，最长 8 位，将拼接到源模型名后（如源模型为 `my-qwen-ft`，后缀填 `awq8`，则产出模型名为 `my-qwen-ft-awq8`）； +- **量化模板**：以卡片形式展示，名称含 MU 编号（如 `MU8`），编号越大表示部署规格越小、成本越低，但潜在精度损失可能越高；**切换源模型会自动清空已选模板**； +- **校准数据（条件必填）**：仅当所选模板要求校准时出现，最多选 5 个已在[数据管理](https://help.aliyun.com/zh/model-studio/manage-data/#9d2f7039bfo1a)中发布成功的数据集，不支持 OSS 挂载数据。 -当前支持压缩的模型以控制台展示为准。已知支持的模型系列及规格如下： +> **注意**：文档中“压缩前部署规格”示例为 `MU1*2`，而“压缩后”为 `MU8*1`，但 MU 编号逻辑易引发误解——实际 `MU8` 表示单卡规格为 MU8，非“8 张 MU1 卡”。该表述已在 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 的表格与计费说明中统一使用，开发者应以控制台模板卡片标注的实际规格为准，避免按数值大小线性推断资源量。 -| 模型系列 | 基础模型 | 压缩前部署规格 | 压缩后部署规格 | -|---------|---------|--------------|--------------| -| Qwen | qwen3.5-flash-2026-02-23 | MU1*2（108 元/小时） | MU8*1（47 元/小时） | +## 使用方式 -仅支持通过百炼平台微调产出的自定义模型，不支持基础模型或第三方模型。 +1. **前提**：确保工作空间中存在状态为“成功”的微调模型（参见 [模型调优](https://help.aliyun.com/zh/model-studio/fine-tuning/#73749f1ee5634)）； +2. 控制台路径：**模型 > 模型训练 > 模型压缩** → 单击**创建压缩任务**； +3. 完成参数配置（特别注意：所有必填项填完后“开始压缩”按钮才可点击）； +4. 创建后不可修改配置，任务状态可在列表页或详情页跟踪（含 `PENDING`/`QUEUING`/`RUNNING`/`SUCCEEDED`/`FAILED` 等 7 种状态）； +5. 成功后，压缩模型将出现在模型中心，可直接用于部署；失败时请先查看详情页错误信息，再结合 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 中的“失败排查”步骤分析日志。 -## 使用限制 +## 限制和注意事项 -- 仅华北2（北京）地域可用。 -- 仅支持百炼平台微调产出的自定义模型。 -- 压缩后的模型不支持继续微调或二次压缩。 -- 压缩任务创建后不可修改配置。 - -## 创建压缩任务 - -在使用模型压缩前，需确保工作空间中已有微调训练完成的自定义模型。操作路径：控制台 → 模型 → 模型训练 → 模型压缩 → 创建压缩任务。 - -创建任务时需配置以下参数（详见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)中的操作步骤）： - -| 参数 | 必填 | 说明 | -|-----|------|-----| -| 任务名称 | 是 | 最长 50 字符，建议包含模型简称、量化方式和版本号 | -| 任务描述 | 否 | 最长 200 字符 | -| 选择源模型 | 是 | 仅展示可压缩的微调模型，切换源模型会清空已选量化模板 | -| 量化产出模型名后缀 | 是 | 仅支持小写字母和数字，最长 8 位 | -| 量化模板 | 是 | 须先选择源模型，MU 编号越大表示部署规格越小、成本越低 | -| 校准数据 | 条件选填 | 仅当量化模板包含校准输入参数时显示，最多选择 5 个数据集 | - -## 量化模板与校准数据选择 - -**量化模板**决定了压缩后模型的部署规格。模板名称中 MU 编号越大，部署规格越小、成本越低，但精度损失可能越大。建议根据业务对成本和精度的权衡来选择。 - -**校准数据**用于提升量化精度，建议选择与目标推理场景语义相近的数据集。例如用于客服问答场景时，应选择包含客服对话样本的数据集。 - -## 任务状态与管理 - -压缩任务有 7 种状态：PENDING（待开始）→ QUEUING（排队中）→ RUNNING（运行中）→ SUCCEEDED（成功）/ FAILED（失败）/ CANCELING（停止中）→ CANCELED（已取消）。 - -任务管理操作： -- **停止任务**：仅 PENDING 和 RUNNING 状态可停止，停止后不可恢复。 -- **删除任务**：仅终态（SUCCEEDED、FAILED、CANCELED）可删除，删除任务记录不影响已产出的压缩模型。 - -任务失败时，可在详情页查看错误信息，或切换到日志页签搜索 ERROR 级别日志进行排查。如仍无法解决，可提交工单并附上任务 ID 和日志文件。更多管理细节参见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 - -## 计费说明 - -- 压缩任务本身限时免费，截止时间以控制台公告为准。 -- 压缩后的模型在部署阶段按 MU 规格计费。 -- 建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选择最优方案后再正式上线。 +- **不可逆性**：压缩后的模型不支持继续微调，也不支持二次压缩；若需调整，必须从原始全精度微调模型重新发起压缩任务； +- **数据依赖**：校准数据必须提前在数据管理中创建并发布，不支持运行时上传或 OSS 路径引用； +- **免费策略**：压缩任务本身限时免费（截止时间以控制台公告为准），但压缩后模型的部署费用始终按 MU 规格实时计费，与免费期无关； +- **地域与模型绑定**：仅华北2（北京）可用，且仅支持百炼平台内微调产出的模型——该限制在 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 的“使用限制”与“常见问题”中多次重申，无例外情形。 ## 来源文档 - [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md) - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md index d021cb14..96e71e07 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md @@ -1,82 +1,56 @@ # model context protocol -模型上下文协议（Model Context Protocol, MCP）是 Anthropic 提出的开源标准协议，用于在大模型与外部工具之间搭建统一的信息传递通道。阿里云百炼基于 MCP 提供全周期服务：开发者无需为每个外部工具编写专用接口，即可让智能体、工作流应用接入海量第三方工具，也能通过外部调用集成到第三方应用或个人项目中。 +模型上下文协议（Model Context Protocol, MCP）是阿里云百炼平台提供的标准化接口协议，用于在大模型与外部工具（如地图、搜索、数据库等）之间建立安全、可扩展的上下文传递通道。它屏蔽了底层通信细节，使开发者无需为每个工具单独开发适配层，即可在智能体或工作流中声明式接入多种能力。MCP 协议本身遵循 [Anthropic 提出的开源标准](https://modelcontextprotocol.io/)，百炼在此基础上提供了云托管、自定义部署和外部集成全链路支持。 -## 服务类型 +## 支持的模型/功能 -百炼将 MCP 服务分为两大类，均需先在 [MCP 广场](https://bailian.console.aliyun.com/?tab=mcp#/mcp-market) 开通或部署后使用： +MCP 服务**不直接绑定特定大模型**，而是作为独立能力模块被智能体（Agent）或工作流（Workflow）调用。当前百炼平台支持以下两类 MCP 服务： -- **官方 MCP 服务**：百炼官方云端部署，开通即用（如 Amap Maps、Sequential Thinking、QuickChart、联网搜索 WebSearch 等）。详见 [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 -- **自定义 MCP 服务**：由开发者自行部署，支持三种方式（详见 [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md)）： - - **使用脚本部署**：面向遵循 MCP 协议的代码包，托管到函数计算 FC，支持 `npx`（Node.js）、`uvx`（Python）、`http`（远程服务）三种安装方式。 - - **从 AI 网关导入**：把已有的 RESTful API 通过 AI 网关升级为 MCP 服务后导入。 - - **从阿里云 OpenAPI 导入**：通过 OpenAPI 开发者门户将官方 OpenAPI 发布为 MCP 服务，用于操作 OSS、ECS 等阿里云产品。 +- **官方 MCP 服务**：由阿里云预部署并维护，开箱即用，包括 Amap Maps（地理信息）、WebSearch（联网搜索）、Firecrawl（网页爬取）、Sequential Thinking（逻辑推理）、QuickChart（图表生成）等。其中 Amap Maps 服务限时免费，[联网搜索MCP服务](https://bailian.console.aliyun.com/cn-beijing?tab=app#/mcp-market/detail/WebSearch)提供 2000 次/月免费额度 [原文标题](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md)。 +- **自定义 MCP 服务**：支持三种部署方式： + - **使用脚本部署**（`npx`/`uvx`）：适用于已发布至 npm 或 PyPI 的开源或自研 MCP Server； + - **从 AI 网关导入**：将现有 RESTful API 封装为 MCP 接口； + - **从阿里云 OpenAPI 导入**：将 OSS、ECS 等云产品 OpenAPI 快速转为 MCP 工具 [原文标题](../../raw/application-user-guide/model-context-protocol/custom-mcp.md)。 -## 使用方式 - -MCP 服务既可在平台内部集成，也可通过外部调用集成到第三方应用。 - -### 平台内部：智能体与工作流 - -- **[智能体应用](../concepts/agent-application.md)**：大模型根据对话内容自动判断是否调用 MCP 服务，单个智能体最多可同时添加 **5 个** MCP 服务。适合路径规划、逻辑推理、多工具组合（如天气查询 + 图表绘制）等场景。 -- **工作流应用**：每个 MCP 节点只能使用一个工具，需手动指定输入参数并将输出传递到下一节点。通常需先用大模型节点把自然语言解析为 MCP 工具所需的输入参数（在 System Prompt 中描述工具的名称、功能、输入输出格式），再接入 MCP 节点。 - -> **注意**：在工作流中仅使用单一工具（如 Amap Maps 的 `maps_weather`）时，工作流只能回答与该工具相关的问题。 - -### 外部调用 - -百炼 MCP 服务支持集成到第三方应用或个人项目，详见 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)： - -- **集成至第三方应用**：支持一键自动配置或手动配置到 Cherry Studio、Cursor 等客户端。手动配置需获取 `DASHSCOPE_API_KEY` 并替换配置文件中的对应变量。 -- **集成至个人项目**：通过 MCP SDK 灵活编码。可结合 OpenAI SDK 调用百炼 MCP 服务，典型端点如 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`，鉴权使用 `Authorization: Bearer `。 +> **注意**：文档 3 明确指出“阿里云百炼 MCP 服务不能在调用千问 API 时直接接入”，即 MCP 仅支持集成于百炼平台内的智能体或工作流应用，**不可用于直连 `dashscope` SDK 的纯 API 调用场景** [原文标题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 -## 关键参数与配置 +## 关键参数 -- **传输协议**：`type` 字段须与端点路径一致，`"sse"` 对应 GET `/sse`，`"streamableHttp"` 对应 POST `/mcp`。配置不匹配会触发 405/404 等错误。 -- **自定义服务配置示例**（脚本部署）： +MCP 服务配置的核心参数均在创建/导入时指定，关键项如下： -```json -{ - "mcpServers": { - "memory": { - "command": "npx", - "args": ["-y", "@modelcontextprotocol/server-memory"] - } - } -} -``` +| 参数 | 说明 | 示例值 | +|--------|------|---------| +| `service name` / `description` | 仅用于控制台识别，不影响模型调用逻辑 | `"长期记忆"`, `"该服务使大模型能够记录个性化信息..."` | +| `install method` | 决定运行环境与启动方式 | `npx`, `uvx`, `http`（对应 `stdio` 或 `sse/streamableHttp` 类型） | +| `deployment mode` | 影响计费与延迟 | `基础模式：按次计费`（冷启动延迟）、`极速模式`（常驻，需额外部署费） | +| `mcpServers` 配置块 | 定义实际服务端点，必须严格匹配协议类型 | `{"memory": {"command": "npx", "args": ["@modelcontextprotocol/server-memory"]}}` | +| `url`（HTTP/SSE 模式） | 必须与 `type` 字段一致：`"sse"` → `GET /sse`；`"streamableHttp"` → `POST /mcp` | `"https://your-server/sse"` | -- **敏感信息加密**：涉及敏感数据的服务在创建时使用 KMS 凭据加密管理。 -- **部署后可修改项**：部署完成后仅支持编辑服务名称和描述；修改部署方式、地域、安装方式或服务配置须先停止部署再重新部署。 +所有自定义服务均需确保其符合 MCP 协议规范，否则会触发 `11200054 - MCP_PROTOCOL_ERROR` 等错误码 [原文标题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 -> **注意**：百炼 MCP 服务已从旧版 SSE 协议升级为新版 **Streamable HTTP** 协议。已开通用户需在 MCP 广场执行"取消开通 → 立即开通"完成协议升级；SDK 调用请使用 `streamablehttp_client` 连接。 - -## 计费 +## 使用方式 -- **云部署 MCP 服务**：限时免部署费用；部分服务涉及第三方 API 调用，费用由第三方收取。联网搜索 MCP 服务免费额度 2000 次，用尽后按 29 元/千次计费，限流 15 QPS（主账号与 RAM 子账号共享）。 -- **自定义部署 MCP 服务**： - - **基础模式**：无部署费用，按调用时长计费（0.000156 元/秒），首次调用有冷启动延迟，适合偶尔调用。 - - **极速模式**：有部署费用（0.000036 元/秒）+ 调用费用（0.000156 元/秒），适合长时间在线、调用频繁的场景。 +### 在平台内集成 +- **智能体应用**：最多可同时启用 5 个 MCP 服务。模型根据提示词自动判断是否调用及调用哪个工具，例如输入“从杭州萧山国际机场到杭州西湖景区”将触发 Amap Maps 的路径规划 [原文标题](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 +- **工作流应用**：每个 MCP 节点需**手动指定具体工具名**（如 `maps_weather`）和输入参数（支持变量引用），适用于确定性编排场景。 -## 限制与注意事项 +### 外部调用 +- **第三方应用集成**：支持一键配置至 Cherry Studio、Cursor 等工具，自动注入 `DASHSCOPE_API_KEY` 和服务地址。 +- **SDK 编程集成**：推荐使用 `mcp` Python SDK + `OpenAI` 兼容客户端，通过 `streamablehttp_client` 连接 `/mcp` 端点，实现工具发现（`list_tools`）与调用（`call_tool`）全流程 [原文标题](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)。 -结合 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)，接入时需注意以下限制： +## 限制和注意事项 -- **不能直连千问 API**：MCP 服务必须集成在智能体或工作流应用中使用，无法在直接调用千问 API 时接入。 -- **无法访问本地资源**：自定义 MCP 服务托管在函数计算 FC，暂不支持访问用户本地数据库、文件、硬件等资源；需要访问本地资源的 MCP Server 建议在本地部署。 -- **访问远程资源需配置网络**：FC 无固定出口公网 IP，访问云数据库等远程资源需配置 FC 的 IP 白名单或打通 VPC 网络。 -- **仓库与版本限制**：私有 npm 仓库暂不支持，需发布到公共仓库或改用 SSE；通过 npx/uvx 部署的服务在源版本更新后不会自动更新，需手动重新部署。 -- **调用增加 Token 消耗**：MCP 返回的内容会作为上下文传入模型，增加输入 Token，并可能间接增加输出 Token。 -- **调用失败排查**：优先确认已开通/升级服务、API Key 有效、额度未用尽；若模型无报错但不调用工具，应在提示词中明确工具名称与能力，必要时更换更强的推理模型（如千问 3 系列）。自定义服务的连接、超时、鉴权、协议等错误可对照 `11200044`~`11200060` 系列错误码逐项排查。 +- **网络与资源限制**：MCP 服务运行于函数计算（FC），**无法访问用户本地资源（如本地数据库、文件）**；若需访问云数据库，必须配置 FC 出口 IP 白名单或 VPC 打通 [原文标题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 +- **协议兼容性**：百炼已全面升级至 **Streamable HTTP 协议**（`POST /mcp`），旧版 SSE（`GET /sse`）需手动取消再重新开通以完成升级 [原文标题](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)。 +- **版本与更新**：通过 `npx`/`uvx` 部署的服务**不会自动更新**，包版本变更后必须手动重新部署 [原文标题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 +- **安全要求**：部署自定义 MCP 服务前，务必核实源代码可信度；涉及敏感参数（如 API Key）必须通过 KMS 凭据加密，禁止明文配置。 ## 来源文档 - [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md) -- [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) - [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md) -- [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md) - - +- [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) +- [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md index a2d4f8ac..a25b0634 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md @@ -1,173 +1,43 @@ # model data overview -百炼平台的数据管理功能用于在[模型调优](../concepts/fine-tuning.md)和[评测](../concepts/evaluation.md)前创建、清洗、增强训练集与[评测](../concepts/evaluation.md)集。它统一管理[业务空间](../concepts/workspace.md)下的大模型相关数据集，分为训练集（用于[模型调优](../concepts/fine-tuning.md)）和[评测](../concepts/evaluation.md)集（用于模型评测）两类，并支持基于数据流的可视化数据处理能力。本文汇总训练集与评测集的格式规范、关键参数以及数据清洗与增强的使用方式。 +百炼平台的模型数据体系为模型调优与评测提供结构化、可管理的数据支撑，涵盖训练集、评测集的构建、处理与回流三大核心能力。所有功能当前仅在华北2（北京）和新加坡地域可用，数据集统一通过[数据管理](https://bailian.console.aliyun.com/#/efm/model_data)控制台进行生命周期管理。本文档面向开发者，聚焦数据格式、参数约束与工程实践要点。 -> **注意**：本文涉及的数据管理与数据处理能力**仅适用于华北2（北京）地域**。此外，阿里云百炼目前暂未提供可用的数据处理 API，所有数据处理操作需在控制台完成。 +## 支持的模型/功能 -## 支持的数据集类型 +- **训练集类型**：支持文本生成（SFT、DPO、CPT）、多模态理解（Qwen-VL 系列）、图生视频（首帧/首尾帧）三类训练场景；其中 SFT 和 DPO 均要求 ChatML 格式，CPT 为纯文本 JSONL；图生视频需 ZIP 压缩包含 `data.jsonl` 及对应图像/视频文件。 +- **评测集类型**：当前仅支持文本生成单轮对话评测集（Excel 或 JSONL 格式），用于模型泛化能力评估。 +- **数据处理能力**：支持对 SFT-文本生成训练集（ChatML 格式）进行清洗（如敏感信息打码、重复去重）与增强（基于千问-Max 的 Few-Shot 生成），详见 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **日志回流能力**：支持将 SLS 推理日志自动转化为结构化训练集（SFT/DPO/CPT）或评测集，适用于文本生成场景，详见 [日志回流](../../raw/model-user-guide/model-data-overview/model-log-backflow.md)。 +> **注意**：文档1中称“数据处理暂不支持[SFT-图片理解训练集]和[DPO-文本生成训练集]”，而文档3明确日志回流支持 DPO 训练集生成——二者无冲突，因日志回流产出的是原始结构化数据，后续仍需经人工校验或清洗才可用于训练；但文档2未提及对 DPO 数据的清洗/增强支持，该限制依然有效。 -数据集分为训练集和评测集两类，详见 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +## 关键参数 -| 类型 | 用途 | 支持的子类型 | -| --- | --- | --- | -| 训练集 | 用于[模型调优](../concepts/fine-tuning.md)，通过在特定任务上进行有监督训练提升模型表现 | 文本生成、[多模态](../concepts/multimodal.md)理解、图生视频（首帧）、图生视频（首尾帧） | -| 评测集 | 用于评估模型在未见过数据上的泛化能力 | 文本生成 | +- **`loss_weight`**：SFT（所有 assistant 行）和 DPO（仅 `chosen` 字段）支持该邀测参数，取值范围 `0.0 ~ 1.0`，用于调节单条样本训练权重；需联系商务经理开通权限。 +- **视觉输入字段**：Qwen-VL 训练集中，`system` 消息的 `content` 必须为数组格式 `[{"text":"..."}]`，不可用字符串；图像/视频路径需与 ZIP 包内实际文件名严格一致，且全局唯一。 +- **图生视频坐标规范**：Qwen2.5-VL 使用缩放后图像的绝对像素坐标；Qwen3-VL 使用 `[0, 999]` 归一化相对坐标。 +- **日志回流上限**：单次任务最多回流 10 万条日志，但可通过多次追加版本突破总量限制，详见 [日志回流](../../raw/model-user-guide/model-data-overview/model-log-backflow.md)。 -## 训练集格式 +## 使用方式 -### SFT 训练集（文本生成） +- **创建数据集**：在[数据管理](https://bailian.console.aliyun.com/#/efm/model_data)页面，选择“新建数据集” → 指定类型（训练集/评测集）→ 上传 ZIP（图生视频）、JSONL（SFT/DPO/CPT）、Excel（评测集）或选择“日志回流”导入。 +- **数据处理**：仅支持 SFT-文本生成训练集（ChatML 格式）。需先在数据流画布中编排“数据清洗”与/或“数据增强”节点，发布后创建任务；处理结果自动生成新版本，原数据集不受影响。 +- **日志回流配置**：需先完成 SLS 审计日志与推理日志的开通及角色授权（见 [日志回流](../../raw/model-user-guide/model-data-overview/model-log-backflow.md)），再通过模型监控页或数据管理页进入表单，严格按顺序填写时间范围 → API Key → 模型 → 训练方式等参数。 +> **注意**：文档2强调“阿里云百炼目前暂未提供可用的API进行数据处理”，所有数据流操作必须通过控制台完成；而日志回流虽无直接 API，但其底层依赖 SLS OpenAPI，高级用户可通过 SLS 查询后手动构造 JSONL 导入。 -采用 ChatML 格式，支持多轮对话和多种角色设置。一行训练数据为一个 JSON 对象，结构如下： +## 限制和注意事项 -```json -{"messages": [ - {"role": "system", "content": "系统输入1"}, - {"role": "user", "content": "用户输入1"}, - {"role": "assistant", "content": "期望的模型输出1"}, - {"role": "user", "content": "用户输入2"}, - {"role": "assistant", "content": "期望的模型输出2"} -]} -``` - -不支持 OpenAI 的 `name`、`weight` 参数，所有的 assistant 输出都会被训练。单条训练数据的所有 assistant 行支持 `loss_weight` 参数（设置范围 `0.0~1.0`，数值越大重要性越高），该参数属于邀测参数，如需使用请联系商务经理。 - -### SFT 思考模型（thinking） - -只能针对**最后**的 assistant 输出进行训练，思考内容必须用 `导读` / `` 标签包裹，且思考标签前后的若干个 `\n` 必须保留。中间的 assistant 输出不应添加思考标签。也可以在训练样本中设置模型不输出 `导读` 标签，但训练完成后不建议再开启思考模式调用。 - -### SFT 视觉理解（千问 VL） - -支持图片、视频文件路径和图片帧列表三种输入。如需传入 `system` 消息，对应 `content` 必须使用数组格式 `[{"text":"..."}]`，不能使用字符串格式。关键参数： - -| 字段 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `image` | str | 是 | 图片文件路径 | -| `resized_width` / `resized_height` | int | 否 | 图片目标缩放尺寸（像素） | -| `video`（路径模式） | str | 是 | 视频文件路径，仅 qwen3.5 及以后 VL 模型支持 | -| `video`（帧列表模式） | `List[str]` | 是 | 图片帧列表，需配合 `sample_fps` | -| `fps` / `sample_fps` | float | 否 | 训练输入频率 / 帧率 | -| `video_start` / `video_end` | float | 否 | 视频截取起止时间（秒） | - -关于物体定位坐标：Qwen2.5-VL 使用相对缩放后图像左上角的绝对像素坐标；Qwen3-VL 使用相对坐标，坐标值会缩放到 `[0, 999]` 范围。 - -### DPO 数据集 - -DPO ChatML 格式将 `messages` 内所有内容作为输入，通过 `chosen` 与 `rejected` 字段训练模型对最后一条用户输入的正负反馈。针对深度思考内容需使用 `导读` 标签包裹。`chosen` 模块支持 `loss_weight` 参数（邀测参数）。 - -### CPT 训练集 - -纯文本格式，一行训练数据结构为 `{"text":"文本内容"}`。 - -### 图生视频训练集 - -分为基于首帧和基于首尾帧两种。训练集必须提供，验证集可选（无需提供视频，训练任务会在评估节点自动调用模型生成预览视频）。 - -- 标注文件固定命名为 `data.jsonl`，最大 20MB，每行为一个 JSON 对象。 -- 字段包括 `prompt`、`first_frame_path`、`last_frame_path`（仅首尾帧）、`video_path`（仅训练集）。 -- 图像最大分辨率 4096*4096，支持 BMP、JPEG、PNG、WEBP；视频支持 MP4、MOV。 - -## 文本生成评测集 - -单轮对话评测数据，使用 Excel 格式，每行包含 Prompt（用户输入）和 Completion（模型期望输出）。参评模型基于评测集中每条 Prompt 进行推理，评分员或自动化评分系统参考 Completion 对推理结果评分。 - -## 数据集构建规模要求 - -不同调优方式对训练集规模有不同最低要求： - -| 调优方式 | 最低规模 | -| --- | --- | -| CPT | 一千万 [Token](../concepts/token.md) 优质预训练数据 | -| SFT | 上千条优质微调数据 | -| DPO | 上百条人类偏好数据 | - -> 如果调优后评测结果不佳，最简单的改进方法是收集更多数据进行训练。 - -## 数据清洗与增强 - -在模型调优前，可使用数据处理功能对训练集进行数据清洗和数据增强，从而获得更高质量的训练集，详见 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 - -| 处理方式 | 适用场景 | -| --- | --- | -| 数据清洗 | 修正训练数据中的规范性、合规性、一致性及重复等问题（如特殊内容移除、敏感信息打码等十种清洗操作） | -| 数据增强 | 增加训练数据的多样性和均衡性，或扩展数据规模 | - -> **注意**：如果训练集数据不适合清洗与增强（如法律文件、医学记录、文学作品、方言汇总、用户评论、技术手册等），建议直接跳过数据处理。此外，数据处理目前仅支持 SFT-文本生成训练集，不支持图片理解训练集和 DPO-文本生成训练集。 - -### 创建数据流任务 - -数据处理通过在控制台搭建自定义数据流完成。典型流程为「先清洗后增强」——确保增强操作在干净、高质量的数据集上进行，保证模型调优数据源的准确性。两步操作： - -1. **创建数据流**：在数据管理页面创建空白数据流，将数据清洗节点（开启所需算子）和数据增强节点拖入画布并依次连接，发布后即可使用。也可直接使用预置的数据流模板。 -2. **创建数据流任务**：选择已发布的数据流，输入任务名称，选择模型数据作为数据来源并指定训练集，任务自动执行。执行期间不支持手动终止。 - -任务处理状态包括：**处理中**（执行中，高峰时段需排队）、**已完成**（成功完成，可查看处理结果）、**处理失败**（建议提交工单咨询）。 - -> 训练集在清洗或增强后会自动生成一个新版本，新版本独立保存，不会覆盖原训练集。建议检查清洗后的训练集，确保数据完整性和真实性未被破坏。 - -### 节点说明 - -数据流由若干节点组成，每个数据流必须包含一个开始节点和一个结束节点： - -| 节点 | 作用 | -| --- | --- | -| 开始/结束 | 开始节点接收待处理训练集（`对话文本`参数无法更改）；结束节点输出处理结果，自动生成新版本 | -| 条件判断 | 设置条件分支，支持且/或配置，多条件自上而下顺序执行 | -| 数据清洗 | 选择清洗算子（如文章相似度去重、敏感词过滤、毒性消除等），按编排顺序执行，输出清洗后训练集及 `dataSetCount` 变量 | -| 数据增强 | 增加数据多样性和规模，分为通用、文本分类、文本抽取、文本创作四种场景 | - -### 数据增强节点关键参数 - -数据增强节点本质上是基于 `千问-Max` 大模型的 Few-Shot 生成器，暂不支持选择其他模型。每次最多生成 2000 条样本。 - -| 参数 | 说明 | -| --- | --- | -| 生成样本数 | 需要生成的数据量。原训练集 N 条 + 生成 M 条 = 增强后 N+M 条 | -| 指令生成依赖样本数 | 从原训练集中选出的种子数量，拼入 Prompt 提供给大模型。若种子+Prompt 总长度超过千问-Max 最大输入 [Token](../concepts/token.md)，系统会自动调整 | -| 过滤相似度阈值 | 控制生成数据的相似度过滤 | -| Prompt 配置 | 定义增强任务输入输出要求，支持 `few_shot_examples` 参数。提供默认模板 | - -Few-Shot 策略示例：训练集 1000 条、指令生成依赖样本数 5、生成样本数 200，则每次从 1000 条中抽样 5 条拼入 `few_shot_examples`，请求千问-Max 生成 1 条，重复 200 次。 - -> 输出数据中 `foreignKey` 为系统后添加的标识字段，增强后的训练集可直接用于模型调优，无需删除该字段。 - -### 数据增强建议 - -- **任务相关性**：确保生成数据与目标任务高度相关，避免引入不相关变体。 -- **多样化策略**：使用同义词替换、随机抽样、翻译变换等多种策略提升数据多样性。 -- **平衡增强**：生成数据应在类别、难度和结构上相对平衡，避免过度接触特定类型数据导致过拟合。 - -## 限制与注意事项 - -- 数据管理与数据处理能力仅适用于华北2（北京）地域。 -- 数据处理暂未提供可用 API，需在控制台完成；仅支持 SFT-文本生成训练集（ChatML 格式），不支持图片理解和 DPO 训练集。 -- 压缩包格式为 ZIP，最大 2GB；VL 训练集图片单张尺寸宽高均不超过 1024px、单张不超过 10MB；图生视频训练集图像/视频最大分辨率 4096*4096。 -- `loss_weight`、思考模式相关参数等部分能力属于邀测参数，如需使用请联系商务经理。 -- 数据流任务执行期间暂不支持手动终止，处理失败建议提交工单咨询。 - -> **注意**：构建有效的 SFT 训练集通常需要 1000+ 样本；数据增强节点暂不支持选择模型，固定使用千问-Max。更多信息参考 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) 与 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **地域限制**：全部功能（数据集管理、清洗增强、日志回流）均仅限华北2（北京）和新加坡 Region，其他地域不可用。 +- **格式强约束**： + - SFT/DPO 训练集必须为 `.jsonl`（每行一个 JSON 对象），ZIP 包内 `data.jsonl` 必须位于根目录； + - 图生视频 ZIP 包内图像/视频文件名不得重复，且 `data.jsonl` 中路径字段（如 `first_frame_path`）仅写文件名，**不可包含子目录路径**； + - 多模态训练集中，`system` 消息 content 若含图像/视频，必须用数组格式声明。 +- **存储与版本**：平台存储模式下，数据处理与日志回流均自动发布新版本；OSS 挂载模式不支持“新增版本”操作，追加数据必须通过“导入数据”页完成。 +- **安全与合规**：数据清洗中的“敏感信息打码”等算子仅作用于文本字段，对图像/视频内容无效；含法律、医疗等高敏领域数据，文档2明确建议跳过自动清洗，应人工审核。 ## 来源文档 - [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) - [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md) - - - - - - - - - - - - - - - - - - - +- [日志回流](../../raw/model-user-guide/model-data-overview/model-log-backflow.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md index 0491c527..a074115c 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md @@ -1,93 +1,50 @@ # model deployment 1 -模型部署让你为平台预置模型或调优后的自定义模型获得独立、资源专享的推理服务，以满足高并发、低延迟等生产需求。本页汇总三种计费方式的选型、PTU 长输入与前缀缓存机制、LoRA 模型导入约束，以及通过控制台或 API 完成部署的完整流程，面向需要落地专属推理服务的开发者。 +百炼平台的 `model deployment 1` 是面向生产级推理服务的模型部署能力，支持将预置模型或用户调优后的模型（如 LoRA）部署为资源独占、性能可预期的专属服务。该能力提供三种核心计费与调度模式：预置吞吐（PTU）、模型单元（MU）和按 [Token](../concepts/token.md) 用量计费，分别适配高并发低延迟、高性能可定制及低成本验证等典型场景。部署后服务通过统一 API 接入，支持 OpenAI、Anthropic 等兼容协议。 -## 三种计费方式与选型 +## 支持的模型/功能 -百炼提供三种互斥的部署计费方式，计费方式在服务创建后无法更改，如需切换必须先下线已部署的模型再重新部署（详见 [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md)）： +- **预置模型**：千问系列（Qwen3/Qwen2.5/Qwen-VL/Qwen-Omni）、DeepSeek（v3/v4）、GLM（5.x/4.7）、Kimi-K2.5、CosyVoice 等主流模型均支持 PTU 和 MU 部署；部分模型（如 `qwen3.7-plus-2026-05-26`、`glm-5.1`）额外支持长输入（最高 256K token）与前缀缓存 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 +- **自定义模型**：仅支持从 OSS 导入的 LoRA 微调模型，需满足 rank ∈ {8,16,32,64}、词汇表与 chat_template 未修改、视觉模型 VIT 冻结等约束 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。全参微调模型暂不支持导入。 +- **关键功能**： + - PTU 模式支持阶梯容量系数（如 glm-5.1 超 32K 输入按 1.33 系数折算）和缓存折扣（命中部分按 20% 折算）； + - MU 模式支持 PD 分离计算（降低首 [Token](../concepts/token.md) 延迟）、推理模式选择（Instruct/Thinking）及最长上下文配置； + - 所有部署均支持自动续费、扩缩容（PTU/MU 可自助调整，[Token](../concepts/token.md) 计费需人工审核）。 -- **预置吞吐（PTU，Provisioned Throughput Unit）**：平台预留资源保障特定 TPM 吞吐能力，额度内不限速。相比按 Token 计费，TPS 通常提升约 1.5～2.0 倍，适合流量可预估的高负载生产环境（智能客服、实时内容审核）。支持预付费（按天）与后付费（按小时），可自助增减吞吐量并设置自动续费。 -- **模型单元（MU）**：按使用时长 × 模型单元数量计费，资源独占，延迟/吞吐等性能指标可自定义。支持部分预置模型与所有调优后模型，可自助增减模型单元数量，支持 PD 分离计算模式（拆分 Prefill 与 Decode 阶段以降低首 Token 延迟、提高吞吐）。 -- **按 Token 使用量**：以每次调用的输入/输出 Token 计量，不使用不计费。仅支持对基础模型完成 SFT 高效训练后的自定义模型，主要用于调优后模型的效果验证；扩缩容需在控制台提交申请等待人工审核。 +> **注意**：文档 1 中称“部分预置模型与所有调优后模型”支持模型单元部署，但文档 3 明确限定“仅支持导入 LoRA 模型”，且文档 4 的 API 示例中 `plan: "lora"` 实际对应 Token 计费模式（非 MU）。此处以文档 3 和文档 4 为准：**模型单元（MU）仅支持预置模型及符合约束的 LoRA 模型；Token 计费（`plan: "lora"`）专用于已调优 LoRA 模型，与 MU 无关**。 -关键计费公式： +## 关键参数 -- 预置吞吐（按时长）：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)` -- 模型单元（按时长）：`费用 = 使用时长（小时）× 模型单元数量 × 模型单元单价`；预付费按月时改为 `包月数 × 模型单元数量 × 月单价` -- 按 Token：`费用 = 输入 Token 数 × 输入单价 + 输出 Token 数 × 输出单价` +| 参数 | 说明 | 适用模式 | 示例值 | +|------|------|----------|--------| +| `plan` | 部署计费模式 | 全部 | `"ptu"`, `"mu"`, `"lora"` | +| `ptu_capacity` | PTU 额度（输入/输出 TPM） | PTU | `{"input_tpm": 10000, "output_tpm": 1000}` | +| `deploy_spec` / `capacity` | MU 规格与副本数 | MU | `"MU1"`, `4` | +| `enable_thinking` | 是否启用思考模式 | MU（部分模型） | `true` | +| `max_context_length` | 最长上下文长度 | MU（部分模型） | `10000` | +| `rpm_limit` / `tpm_limit` | 服务限流阈值 | MU | `500`, `1000` | -## PTU 长输入与前缀缓存 +- PTU 模式下 `rpm_limit`/`tpm_limit` 等参数不可配置，吞吐由预置额度决定； +- Token 计费模式（`plan: "lora"`）中 `capacity` 参数必须填写但实际无效，扩缩容需通过控制台申请 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 -PTU 部署支持长输入请求（部分模型最高 200K token）和前缀缓存，通过阶梯容量系数和缓存折扣管理额度消耗，详见 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)： +## 使用方式 -- **长输入阶梯系数**：超过 32K token 的输入按更高阶梯系数折算 TPM。例如 glm-5.1 在 `[32K, 200K]` 区间输入系数为 1.33、输出为 1.17；deepseek-v4-pro 与 qwen3.7-plus-2026-05-26 无阶梯（1.0）。 -- **前缀缓存折扣**：命中缓存的输入 token 按折扣系数消耗额度（glm-5.1 为 0.2，deepseek-v4-pro 为 0.08，qwen3.7-plus 为 0.2），可显著降低多轮对话和重复前缀场景的额度消耗。 -- **自动转按量计费**：超出 PTU 额度或输入超过模型上限时，请求自动转为按量计费，无需修改调用代码，业务不中断。 +1. **控制台部署**：前往 [模型部署控制台](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/model_deploy/create)，填写服务名称、选择模型与计费方式，确认后提交。状态变为 `RUNNING` 即部署成功。 +2. **API 部署**：使用 DashScope API 发起 HTTP 请求，需配置 `DASHSCOPE_API_KEY` 环境变量。示例： + - PTU：`curl -X POST ... --data '{"name":"my_qwen","model_name":"qwen-flash-2025-07-28","plan":"ptu","ptu_capacity":{...}}'` + - MU：`curl -X POST ... --data '{"name":"my_qwen_plus","model_name":"qwen-plus-2025-12-01","plan":"mu","deploy_spec":"MU1","enable_thinking":true}'` + - Token 计费：`curl -X POST ... --data '{"model_name":"qwen3-8b-ft-xxx","plan":"lora","capacity":1,"name":"qwen3-8b-ft"}'` +3. **调用与监控**：部署成功后，使用 `model_name`（即 `deployed_model` ID）调用推理 API；额度消耗、缓存命中率等指标可通过 [模型监控](https://bailian.console.aliyun.com/?tab=model#/model-telemetry) 查看。 -API 响应关键字段：`service_tier`（值为 `ptu-standard` 表示使用 PTU 额度，`default` 或不返回表示按量计费）、`provisioned_tokens`（折算后实际消耗的额度 token 数）、`cached_tokens`（前缀缓存命中数）。不同 API 格式（OpenAI Chat / Responses、Anthropic、DashScope）下这些字段的 JSON 路径不同，需按对应格式取值。 +## 限制和注意事项 -> **注意**：模型输入上限存在两处口径。[模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) 的价格表中千问系列多为 128K、部分新模型达 256K，而 PTU 文档明确将「千问 128K / DeepSeek 64K」作为触发自动转按量计费的上限。请以控制台实际展示与所选具体模型为准。 - -> **注意**：长输入场景下 PTU 利用率可能超过 100%，这是阶梯系数导致折算消耗高于原始 token 数的正常现象，超出部分自动转按量计费，不影响服务可用性。 - -## 模型导入（LoRA） - -通过**我的模型**页面可将本地训练的 LoRA 模型从 OSS 导入百炼平台，详见 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。当前版本**仅支持 LoRA 模型，不支持全参微调模型**。 - -导入前提与约束： - -- **OSS Bucket**：需为目标 Bucket 添加 `bailian-datahub-access` 标签（标签值 `read`）；不支持归档/冷归档类存储；不支持访问 Bucket 根目录文件，需放入子目录。首次导入需先完成 OSS 服务关联角色授权（子账号还需主账号授予 `ram:CreateServiceLinkedRole` 权限）。 -- **必需文件**：`adapter_model.safetensors`（权重）与 `adapter_config.json`（含 rank、alpha 等配置）。 -- **rank 限制**：必须为 8、16、32、64 之一，且同一模型所有 LoRA 层使用相同 rank。 -- **词汇表与对话模板**：不得修改原始 vocab 或 chat_template，必须与开源基础模型默认配置一致，否则无法导入。 -- **VL 模型**：必须冻结 VIT，若 adapter 中包含 `visual` 开头的权重参数则无法导入。 - -支持导入的基础模型涵盖千问3、千问3-VL、千问2.5、千问2.5-VL 系列的指定版本。导入后模型状态包括创建中、创建成功（可部署）、创建失败、已失效。 - -> **注意**：导入模型若与本地 vLLM/SGLang 推理效果不一致，通常是推理引擎参数默认值差异所致。可将 `temperature`、`top_p`、`repetition_penalty` 设为 1.0、`presence_penalty` 设为 0 以对齐 vLLM 默认行为。 - -## 使用 API/命令行部署 - -除控制台外，可通过 DashScope HTTP API 完成部署，**仅适用于华北2（北京）地域**，需先获取并配置 API Key，详见 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。核心接口为 `POST/GET/DELETE https://dashscope.aliyuncs.com/api/v1/deployments`，通过 `plan` 字段区分计费方式： - -- **PTU**：`plan: "ptu"`，配合 `ptu_capacity.input_tpm` / `output_tpm`。 -- **模型单元**：`plan: "mu"`，配合 `deploy_spec`（如 `MU1`）、`capacity`（副本数）、`enable_thinking`、`max_context_length`、`rpm_limit`、`tpm_limit`。 -- **按 Token（LoRA 自定义模型）**：`plan: "lora"`，`capacity` 必填但设置无效，扩缩容需在控制台申请。 - -典型部署命令（模型单元）： - -```bash -curl "https://dashscope.aliyuncs.com/api/v1/deployments" \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data '{ - "name": "my_qwen_plus", - "model_name": "qwen-plus-2025-12-01", - "plan": "mu", - "deploy_spec": "MU1", - "enable_thinking": true, - "capacity": 4, - "max_context_length": 10000, - "rpm_limit": 500, - "tpm_limit": 1000 -}' -``` - -部署流程：创建部署 → 返回 `deployed_model`（专属服务唯一 ID）→ 轮询 `GET /deployments/{id}` 直到 `status` 为 `RUNNING` → 通过 DashScope SDK 或兼容 API 发起推理 → 不再使用时 `DELETE /deployments/{id}` 下线并停止计费。 - -## 部署配置与列表管理 - -在控制台部署时可配置：服务名称、选择模型、模型单元类型（部署规格）、部署副本数、部署模板（如「单机部署」，仅模型单元模式可用）、推理模式（Instruct 非思考 / Thinking 思考）、最长上下文、服务限流（RPM/TPM）。 - -部署列表页展示服务名称、模型名称、**模型 Code**（API 调用时指定模型的唯一标识）、部署状态（待部署、部署中、运行中、部署失败、下线中、已停止、变配中等）、计费方式、部署详情与限流详情。 - -## 限制与注意事项 - -- **计费不可逆变更**：计费方式创建后不可改；预付费按天/按月无法提前退费，首月内提前退订按日单价 1.2 倍计费。 -- **部署即计费**：模型部署成功后即产生费用，即便尚未发起任何调用；后付费欠费后资源保留并继续计费 24 小时，超时后停止计费并删除底层资源（部署任务保留）。 -- **按 Token 模式约束**：仅支持 LoRA 调优后模型，一个月内不使用将自动释放。 -- **权限**：API 部署报错 `Workspace ... does not have deployment privilege` 或 `Workspace access denied` 时，需检查 API Key 归属业务空间的模型部署授权与账号操作权限。 -- **删除不可恢复**：执行 DELETE 后服务立即下线且不可恢复。 +- **权限要求**：API 部署需确保 API Key 所属业务空间已授权目标模型的部署权限，否则报错 `Workspace xxx does not have deployment privilege for model xxxx` [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 +- **OSS 导入约束**：LoRA 模型导入前须完成 OSS 服务关联角色授权，并为目标 Bucket 添加 `bailian-datahub-access=read` 标签；模型文件须包含 `adapter_model.safetensors`、`adapter_config.json`、`config.json`，且 rank、词汇表、chat_template 必须与基础模型一致 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 +- **计费与生命周期**： + - PTU 预付费订单无法提前终止，到期后延后 2 小时停服，资源保留 14 小时后释放； + - MU 后付费资源“先买到先得”，购买失败全额退款； + - 所有部署服务创建成功即开始计费，即使未发起调用。 +- **溢出行为**：PTU 模式下若超出额度，「自动溢出」策略将切换为按量付费（响应头含 `x-dashscope-ptu-overflow:true`），「仅使用 PTU 容量」则返回 429；单次输入超模型上限（如千问 128K）亦自动转为按量计费 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 ## 来源文档 @@ -97,4 +54,3 @@ curl "https://dashscope.aliyuncs.com/api/v1/deployments" \ - [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md index 674be24b..c114de64 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md @@ -1,97 +1,65 @@ # model evaluation introduction -模型评测是百炼平台提供的模型能力评估功能，支持自定义评测和基线评测两种方式，通过评测维度对模型推理结果进行打分和对比，帮助开发者选择最优模型或验证调优效果。当前仅支持文本生成类模型评测。 +模型评测是百炼平台提供的模型能力量化评估功能，支持通过自定义或基线方式对文本生成类模型进行多维度打分与对比。它面向开发者提供 AI 自动评测、规则评估和人工评估三种评分范式，帮助完成模型选型、调优验证、质量监控等核心任务。当前功能仅支持文本生成类模型，不支持多模态或语音模型。 -## 核心概念 +## 支持的模型/功能 -模型评测涉及两个容易混淆的核心概念： +- **支持的模型类型**：仅限文本生成类模型（包括预置模型与调优后的模型），详见[预置模型列表](https://help.aliyun.com/zh/model-studio/model-deployment-introduction)。多模态、语音等非文本生成模型暂不支持评测 [模型评测产品概览](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 +- **评测方式**： + - **自定义评测**：用户上传评测数据集（EvaluationSet 类型）或推理结果集，自主创建评测维度并关联执行，支持全地域。 + - **基线评测**：使用平台预置的公开标准数据集（如 C-Eval、GSM8K、BBH 等），系统自动评分，**仅北京地域可用** [创建基线评测任务](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 +- **评分器类型**：共五种，分为三大类： + - *大模型评估*（分类型/数值型）：依赖裁判模型（如千问-Max）进行语义级评判，适用于问答质量、内容安全等场景； + - *规则评估*（字符串匹配/文本相似度）：基于算法（ROUGE/BLEU/Cosine/Fuzzy Match 等）或精确匹配逻辑，适用于翻译、摘要、Function Calling 等有确定性标准的场景； + - *人工评估-分类型*：由人工标注 Pass/Fail，适用于创意写作、合规审核等主观性强的场景 [评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 -- **评分器 Prompt**：配置于评测维度，指导裁判模型如何给被评测模型的回答打分。 -- **System Prompt**：配置于评测任务，为被评测模型设定角色定位或行为规范，通常可留空。 - -两者作用对象和费用归属不同，详见[模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 - -## 评测方式 - -### 自定义评测 - -使用自有数据集和自定义评测维度，支持三种评分方式： - -| 评分方式 | 说明 | 适用场景 | -|---------|------|---------| -| 大模型评估（AI 自动评测） | 由裁判模型（推荐千问-Max）对回答评分 | 问答质量、内容安全等语义理解场景 | -| 规则评估（自动化指标） | ROUGE、BLEU、Cosine 等算法直接计算 | 翻译、摘要、Function Calling 等确定性场景 | -| 人工评估（人工标注） | 人工逐条标注 Pass/Fail | 创意性写作、专业领域判断 | - -### 基线评测 - -使用公开标准数据集（C-Eval、MMLU、GSM8K、BBH 等）快速评测模型基础能力，无需自行准备数据集或配置维度。 - -> **注意**：基线评测仅北京地域可用，不支持下载评测结果，不支持推理结果集数据来源。 - -## 评测维度类型 - -评测维度定义模型的评分规则，创建为模板后可被多个评测任务复用。百炼提供五种评分器类型，详细说明参见[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 - -| 维度类型 | 评分方式 | 需要参考答案 | 裁判模型费用 | -|---------|---------|------------|------------| -| 大模型评估-数值型 | 裁判模型打分（整数，如 0-5） | 否 | 有 | -| 大模型评估-分类型 | 裁判模型标签（Pass/Fail） | 否 | 有 | -| 规则评估-文本相似度 | ROUGE/BLEU/Cosine 等算法 | 是 | 无 | -| 规则评估-字符串匹配 | 相等/不相等/包含 | 是 | 无 | -| 人工评估-分类型 | 人工标注 Pass/Fail | 否 | 无 | - -**选型决策路径**：有标准答案且格式固定用字符串匹配；有标准答案但表述多样用文本相似度；无标准答案需语义理解用大模型评估；需主观判断用人工评估。 +> **注意**：文档1称“当前仅支持文本生成类模型评测”，文档2未明确限定模型类型，但其所有示例与参数说明均围绕文本生成展开，且文档1为更高层级的概览文档，应以文档1为准。 ## 关键参数 -### 评测维度参数 - -- **评分范围**（数值型）：裁判模型打分区间，整数，默认 0-5。建议不超过 10，范围过大会降低 LLM 评分一致性。 -- **通过阈值**（数值型/相似度型）：判定 Pass/Fail 的分界线，步长 0.1（数值型）或 0.01（相似度型）。 -- **评分器 Prompt**（大模型评估类型）：至少包含一个变量（`${prompt}`、`${output}`、`${completion}`），长度不超过 50000 字符。 -- **Pass/Fail 标签**（分类型/人工评估）：两组标签不可重复。 - -> **注意**：维度类型创建后不可修改，选错需删除重建。 - -### 评测任务参数 - -- **数据来源**：评测数据集（含 Prompt + Completion，会产生推理费用）或推理结果集（已含 Output，不产生推理费用）。 -- **推理参数**：Temperature、TopP、System Prompt 等，按所选模型动态加载。 -- **排行参与**：开启后须绑定排行榜，结果加入排名对比。 - -## 使用流程 - -端到端流程分四步： - -1. **准备数据集**：在数据管理模块上传评测集类型数据（含 Prompt 和 Completion 列）。 -2. **创建评测维度**：定义评分标准和方式，选择评分器类型并配置参数。 -3. **创建评测任务**：选择被评测模型、关联数据集和维度，提交评测。 -4. **查看结果**：在指标统计 Tab 查看综合得分和通过率，在数据明细 Tab 查看逐条评分。 - -数据量建议：小规模验证 50-100 条，正式评测 200-500 条，全面评估 500 条以上。 - -## 计费说明 - -费用由两部分构成：被评测模型推理费用 + 裁判模型评分费用。 - -- 使用推理结果集可免去推理费用。 -- 规则评估和人工评估无裁判模型费用。 -- 已部署的调优模型评测不额外计费（推理费用包含在部署算力费用中）。 - -**成本优化**：先用 50-100 条小规模验证 → 保存推理结果集复用 → 有确定性标准的场景优先用规则评估。 - -## 限制与注意事项 - -- 当前仅支持文本生成类模型评测。 -- 基线评测仅北京地域可用。 -- 任务提交后不可更换目标模型，需删除重建。 -- 评测维度类型创建后不可修改。 -- 模型评测当前仅支持控制台操作，不提供公开 API/SDK，如需编程化评测可参考 PAI Judge Model API。 -- 1-3% 的分数差异通常为评测噪声，不建议仅凭微小分差做决策。 -- LLM 评分器存在位置偏差和自我偏好偏差，建议定期人工抽查校准。 - -更多操作细节及常见问题排查请参见[模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)和[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 +| 参数类别 | 参数名 | 说明 | 必填性 | 备注 | +|----------|--------|------|--------|------| +| **通用** | 维度名称 | 最长20字符，建议采用“评估方面+评估方式”命名（如`回答准确性-LLM评分`） | 是 | 创建后可修改 | +| | 描述 | 最长100字符，用于补充说明评判目标 | 否 | — | +| **大模型评估专用** | 裁判模型 | 如千问-Max，影响评分质量与费用 | 是（仅大模型评估） | 推荐千问-Max以保障判分一致性 | +| | 评分器模板 | 预置（如综合评测、标准匹配）或自定义 | 是（仅大模型评估） | 切换会覆盖已编辑Prompt | +| | 评分范围 | 数值型维度专属，整数区间（默认0–5） | 是（仅数值型） | 范围过宽（>10）易降低LLM评分一致性 | +| | 通过阈值 | 判定Pass的最低分（数值型）或相似度（规则评估），步长0.1（数值型）/0.01（相似度） | 是（数值型/相似度型） | 与评分范围联动，业务容忍度决定阈值高低 | +| | Pass/Fail标签 | 分类型维度专属，标签互斥且穷尽 | 是（分类型） | 同一标签不可在Pass与Fail中重复出现 | +| **规则评估专用** | 比较操作符 | 相等/不相等/包含（字符串匹配） | 是（字符串匹配） | Function Calling常用“包含” | +| | 评估指标 | ROUGE-1/ROUGE-L/BLEU/Cosine/Fuzzy Match/Accuracy（文本相似度） | 是（文本相似度） | 翻译用BLEU，摘要用ROUGE-L，语义相关用Cosine | + +所有评分器 Prompt 必须至少包含一个变量：`${prompt}`（用户问题）、`${output}`（模型回答）、`${completion}`（参考答案）。变量缺失将阻止提交 [配置评分器Prompt](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 + +## 使用方式 + +1. **准备数据**：在数据管理模块上传 `EvaluationSet` 类型数据集（含 `Prompt` 和 `Completion` 列），或准备已含 `Output` 的推理结果集。 +2. **创建评测维度**：在控制台「模型评测 > 评测维度」创建至少一个维度模板。类型一经创建不可修改，选错需删除重建。 +3. **创建评测任务**： + - *自定义评测*：选择被评测模型、数据来源（评测数据集或推理结果集）、关联维度；可选开启排行参与（需绑定已有排行榜）。 + - *基线评测*：仅北京地域可见，选择模型与预置数据集（如MMLU、GSM8K），无需配置维度。 +4. **查看结果**： + - 「数据明细」Tab：逐条查看 `Prompt`、`Output`、`Completion` 及各维度评分； + - 「指标统计」Tab：查看综合得分（各维度平均分）、通过率（≥通过阈值样本占比）、分数分布图； + - 支持下载结果（待执行状态及基线评测任务不支持）。 + +> **注意**：文档1明确指出“模型评测当前仅支持控制台操作，不提供公开 API/SDK”，而文档2未提及接口能力。若需自动化，应参考 PAI Judge Model API 替代方案 [常见问题](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 + +## 限制和注意事项 + +- **地域限制**：基线评测仅在北京地域可用，其他地域控制台不显示该选项，属正常行为。 +- **模型限制**：仅支持文本生成类模型；非文本生成模型（如多模态、语音）无法参与评测。 +- **维度限制**：维度类型创建后不可修改；已被排行榜绑定的维度删除后，将导致排行榜无法创建新任务。 +- **费用说明**： + - 使用评测数据集时产生**被评测模型推理费用**（按[Token](../concepts/token.md)计费）； + - 大模型评估维度产生**裁判模型评分费用**（按[Token](../concepts/token.md)计费）； + - 规则评估与人工评估无裁判模型费用； + - 使用推理结果集可规避被评测模型推理费用 [计费说明](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 +- **结果解读**：综合得分是各维度平均分，可能掩盖维度间差异；建议结合分数分布图逐维度分析短板。1–3%的分差通常属评测噪声，不宜作为决策依据。 +- **成本优化建议**： + - 小规模验证（50–100条）确认配置正确后再扩大规模； + - 保存并复用推理结果集，避免重复推理； + - 有确定性标准的场景优先选用规则评估（零裁判模型费用）。 ## 来源文档 @@ -99,9 +67,3 @@ - [评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md index 5cedd681..609da754 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md @@ -1,96 +1,78 @@ # model experience -百炼平台按模态和场景组织了一整套可直接调用的模型能力：文本生成、视觉理解、图片/视频/3D 生成、语音合成与识别、语音转语音、全模态、音乐生成，以及向量与重排序。本页汇总各场景的推荐模型、关键参数和选型要点，帮助开发者快速定位到合适的模型；模型的实时上下文窗口、计费等详细参数请以模型广场为准。 - -## 文本生成 - -通用文本场景（聊天机器人、内容生成、摘要、文档处理、办公任务）推荐从 `qwen3.7-plus` 起步——能力与成本均衡，具备 1M 上下文、Function Calling 和内置工具；效果确认后可切到 `qwen3.6-flash` 降本，功能与上下文一致；需要最强推理时用 `qwen3.7-max`。超长文档（多合同审阅、大规模文献）可用 `qwen-long`（10M 上下文）。AI 编程 / Agent 开发推荐 `qwen3.7-plus`（工具调用完整、1M 上下文适合大代码库）。 - -关键能力： - -- **思考模式**：通过 `enable_thinking` 参数开启（Responses API 用 `reasoning.effort` 控制开关与深度），所有 Qwen3 及以上模型均支持，多为混合模式可按请求切换。 -- **Function Calling**：所有通用模型均支持自定义工具调用；**内置工具**（联网搜索、代码解释器、网页抓取）免复杂配置。 -- **结构化输出**：可强制返回有效 JSON，适合信息抽取。 -- **批量推理**：适合大量、低时延要求的请求以降本。 - -详细的模型对位（从 GPT / Claude / Gemini 迁移）与完整模型清单见 [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md)。 - -## 视觉理解与 OCR - -图像 / 视频理解推荐 `qwen3.7-plus`（1M 上下文、最长 2 小时视频、Function Calling + 内置工具），稳定后可降本到 `qwen3.6-flash`。要点见 [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md)： - -- **图像分辨率**：多数模型支持每张最高 1600 万像素，Token 数按 `h x w / (32 x 32) + 2` 计算。 -- **视频支持**：`qwen3.7-plus` / `qwen3.6-plus` / `qwen3.6-flash` / `qwen3.5-plus` / `qwen3.5-flash` 最长 2 小时 / 2GB；`qwen3-vl-plus` / `qwen3-vl-flash` 最长 1 小时。 -- **OCR / 文档提取**：`qwen3.5-ocr` 针对文档、表格、试卷、手写内容优化；通用图片文字提取也可用旗舰模型。 - -## 图片生成与编辑 - -见 [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md)。推荐 `wan2.7-image-pro`，集成文字渲染、品牌色控制、角色一致性多图生成和图片编辑（文生图最高 4096x4096，编辑最高 2048x2048，支持最多 9 张输入图参考）。仅需生图且追求速度/成本时用 `z-image-turbo`（约快 10 倍、价格约 1/5，写实人像与产品照）；需要负向提示词或单次最多 6 张变体时用 `qwen-image-2.0-pro`（生成和编辑同一模型 ID）。 - -## 视频生成与编辑 - -见 [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md)，按子场景选型： - -- **文生视频**：`happyhorse-1.1-t2v`（1080P、单片段最长 15 秒、有声）；需传入自定义音频文件用 `wan2.7-t2v-2026-06-12`。 -- **图生视频**：首帧生视频用 `happyhorse-1.1-i2v`；首尾帧串联长视频用 `wan2.7-i2v-2026-04-25`。 -- **参考生视频**：`happyhorse-1.1-r2v`（保持角色一致性）；需自定义音色或视频参考主体用 `wan2.7-r2v-2026-06-12`。 -- **视频编辑 / 角色动画**：编辑用 `happyhorse-1.0-video-edit`，特效/运镜复刻用 `wan2.7-videoedit`；动作迁移用 `wan2.2-animate-move`，人物替换用 `wan2.2-animate-mix`（均支持 wan-std / wan-pro 两种模式）。 - -## Tripo 3D 模型生成 - -支持文生 3D、单图生 3D、多图生 3D 三种模式，通过 `input` 中互斥的 `prompt` / `image` / `images` 字段区分。`Tripo/Tripo-H3.1` 面向高精度（最高 200 万面，较慢），`Tripo/Tripo-P1.0` 面向快速预览（最高 2 万面，更快）。贴图质量用 `parameters.texture_quality`（`standard` / `detailed`）控制，几何精度用 `parameters.geometry_quality`（仅 H3.1 支持，`standard` 最高 150 万面 / `ultra` 最高 200 万面）。调用为异步任务，需轮询任务状态（`PENDING` → `RUNNING` → `SUCCEEDED` / `FAILED`），建议间隔 15 秒。 - -> **注意**：Tripo 3D 生成仅适用于**华北2（北京）**地域，且必须使用该地域的 API Key。 - -## 语音合成（TTS） - -见 [语音合成](../../raw/model-user-guide/model-experience/tts-model.md)。先确定内置音色还是自定义音色：标准合成推荐 `qwen-audio-3.0-tts-plus` / `MiniMax/speech-2.8-hd`；自定义音色分**声音复刻**（提供音频样本，用 `qwen-audio-3.0-tts-flash` / `MiniMax/speech-2.8-hd`）和**声音设计**（用文字描述音色，用 `cosyvoice-v3.5-plus` / `cosyvoice-v3.5-flash`），音色统一由 `voice-enrollment` 服务注册管理。接入方式上，WebSocket 双向流式延迟最低（实时交互），HTTP 适合有声阅读等；Qwen 系列以 `-realtime` 后缀区分 WebSocket / HTTP。指令控制可用自然语言动态调节语速、情绪和风格。 - -## 语音识别（ASR） - -见 [语音识别](../../raw/model-user-guide/model-experience/asr-model.md)，按维度选型： - -- **实时 vs 非实时**：实时（WebSocket）用 `fun-asr-realtime` 或 `qwen3.5-omni-plus-realtime`；非实时文件转写（HTTP）用 `fun-asr` 或 `qwen3.5-omni-plus`。 -- **专业术语**：Prompt 上下文注入（Qwen3.5-Omni，无需预配置）或热词表（Fun-ASR，适合稳定术语列表）。 -- **说话人分离**：仅 Fun-ASR 非实时模型（`fun-asr`、`fun-asr-mtl`）支持。 -- **情感识别**：Qwen-ASR 与 Qwen3.5-Omni 系列支持，推荐 `qwen3-asr-flash-realtime` / `qwen3-asr-flash-filetrans`。 - -Paraformer 为较早一代模型，新业务建议迁移到 Fun-ASR 或 Qwen-ASR。 - -## 语音转语音（S2S）与全模态 - -构建语音应用可选 **S2S 单模型**（延迟低、端到端感知语调情绪）或 **Pipeline（ASR + LLM + TTS）**（可自定义音色、各阶段独立选优）。S2S 路线推荐：语音助手/客服用 `qwen3.5-omni-plus-realtime`，成本敏感用 `qwen3.5-omni-flash-realtime`，同传/直播翻译用 `qwen3.5-livetranslate-flash-realtime`，视频配音/播客翻译用 `qwen3-livetranslate-flash`。全模态（同时理解文本/音频/图片/视频）三大系列为 Qwen3.5-Omni（旗舰）、Qwen3-Omni-Flash（轻量、支持思考模式）、Qwen3.5-Livetranslate（专业翻译，开箱即用，60 种语言）。 - -> **注意**：Function Calling 与联网搜索能力在不同接入模式下差异较大——例如 `qwen3-omni-flash` 在 HTTP 模式支持 Function Calling 和思考模式，但其 WebSocket（`-realtime`）版本均不支持；联网搜索仅 Qwen3.5-Omni（HTTP / WebSocket）支持，且联网搜索与 Function Calling 不可同时开启；思考模式下不输出语音。选型前务必核对目标模型的具体接入模式。 - -## 音乐生成 - -Fun-Music 是端到端音乐生成模型，通过 `prompt` 描述风格/场景/情绪自动作词谱曲，或通过 `lyrics` 提供自定义歌词，用 `gender` 选男女声（仅 `fun-music-v1`），`is_instrumental=true` 生成纯音乐（此时 `lyrics` / `gender` 被忽略），`format` 指定 mp3 / wav 输出。 - -> **注意**：Fun-Music 处于邀测阶段，需在模型广场申请开通，且服务仅在**华北2（北京）**地域可用。 - -## 向量与重排序 - -见 [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md)。纯文本搜索 / RAG / 聚类推荐 `text-embedding-v4`（维度 64~2048，默认 1024，最大 8192 Token；迁移旧索引可用 `text-embedding-v3`）；跨模态检索用 `qwen3-vl-embedding`（融合向量）或 `tongyi-embedding-vision-plus`（独立向量）。重排序用于 Embedding 检索后对 Top-N 结果精排：纯文本用 `qwen3-rerank`（100+ 语言、最多 500 文档），多模态用 `qwen3-vl-rerank`（文本/图片/视频混排）。 - -## 选型与使用注意事项 - -- 旧版模型（如旧版 Qwen、Paraformer、`qwen-omni-turbo`、`qwen-tts` 等）不再作为首选，新项目建议使用各系列最新版本。 -- 上下文窗口、计费、地域可用性等以模型广场实时信息为准；本页面数据可能滞后。 -- 图片/视频/3D/音乐等生成类模型多为异步任务，需按文档轮询或配置回调。 +`model experience` 是百炼平台面向开发者提供的统一模型能力体验层，涵盖文本、图像、视频、3D、语音、音乐、多模态及向量检索等全栈AI模型服务。所有模型均通过标准化API（HTTP/WebSocket）接入，支持结构化输出、Function Calling、思考模式、[异步任务](../concepts/asynchronous-task.md)等核心能力，并按场景提供推荐选型与参数配置指南。开发者可基于具体需求（如延迟敏感度、精度要求、成本约束、输入模态）快速定位最优模型。 + +## 支持的模型/功能 + +百炼平台提供覆盖全模态的模型矩阵，按能力域组织如下： + +- **文本生成**：以 `qwen3.7-plus` 为旗舰，支持100万上下文、Function Calling、内置工具（联网搜索/代码解释器）、结构化JSON输出及逐步推理（`enable_thinking`）。轻量场景可选用 `qwen3.7-flash` 或 `deepseek-v4-flash` [原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +- **图像生成与编辑**：`wan2.7-image-pro` 集成文生图（最高4096×4096）、多图编辑与角色一致性；`qwen-image-3.0-pro`（邀测中）支持负向提示词与多语言字体渲染；`z-image-turbo` 适用于低成本写实人像生成 [原文标题](../../raw/model-user-guide/model-experience/image-model.md)。 +- **视频生成与编辑**：`happyhorse-1.1-t2v` 支持有声文生视频（1080P，3–15秒）；`wan2.7-i2v-2026-04-25` 支持首尾帧续写；`wan2.2-animate-move` 提供角色动画迁移（pro/std双模式） [原文标题](../../raw/model-user-guide/model-experience/video-generate-edit-model.md)。 +- **视觉理解**：`qwen3.7-plus` 支持图像（最高1600万像素）、视频（最长2小时/2GB）联合分析，具备Function Calling与结构化输出能力；`qwen3.5-ocr` 专用于文档/手写体OCR [原文标题](../../raw/model-user-guide/model-experience/vision-model.md)。 +- **3D生成**：Tripo系列（`Tripo/Tripo-P1.0` / `Tripo/Tripo-H3.1`）支持文生3D、单图生3D、多图生3D，仅限华北2（北京）地域，需异步轮询获取GLB结果 [原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **语音与音频**： + - 合成：`qwen-audio-3.0-tts-plus`（指令控制+声音复刻）、`cosyvoice-v3.5-plus`（声音设计）； + - 识别：`fun-asr`（支持说话人分离）、`qwen3.5-omni-plus`（Prompt上下文注入）； + - 全模态：`qwen3.5-omni-plus-realtime`（实时音视频对话）、`qwen3.5-livetranslate-flash-realtime`（60语种同传）； + - 音乐：`fun-music-v1`（歌词/提示词生成带人声歌曲，MP3/WAV输出） [原文标题](../../raw/model-user-guide/model-experience/fun-music.md)。 +- **向量与重排序**：`text-embedding-v4`（文本嵌入，维度可配64–2048）、`qwen3-vl-embedding`（图文融合向量）、`qwen3-rerank`（纯文本重排序，支持500文档） [原文标题](../../raw/model-user-guide/model-experience/embedding-rerank-model.md)。 + +> **注意**：文档 9 和文档 11 中关于 `qwen3.5-omni-plus-realtime` 的联网搜索支持存在矛盾——文档 9 明确标注其支持联网搜索，而文档 11 的“S2S 单模型的附带能力”章节称“Qwen3.5-Omni 实时（WebSocket）模式……不支持此功能”。实际以文档 9 为准，该模型在 WebSocket 模式下支持联网搜索，但需注意与 Function Calling 不可同时启用。 + +## 关键参数 + +各模型共性参数与关键字段如下： + +| 参数名 | 类型 | 说明 | 示例值 | +|--------|------|------|--------| +| `model` | string | 模型ID，必须精确匹配（含快照版本） | `"qwen3.7-plus"`、`"Tripo/Tripo-P1.0"` | +| `input` | object | 输入数据容器，结构因模态而异 | `{"prompt": "夏日民谣"}`（音乐）、`{"image": "url"}`（3D） | +| `parameters` | object | 模型特有配置 | `{"texture_quality": "detailed"}`（Tripo）、`{"format": "wav"}`（音乐） | +| `reasoning.effort` | string | 控制思考模式深度（仅部分Qwen3模型） | `"low"` / `"medium"` / `"high"` | +| `X-DashScope-Async` | header | [异步任务](../concepts/asynchronous-task.md)必需头（如Tripo、视频生成） | `"enable"` | +| `is_instrumental` | boolean | Fun-Music 纯音乐开关 | `true` | + +- **[异步任务](../concepts/asynchronous-task.md)**：Tripo 3D、视频生成等长耗时任务必须使用 `X-DashScope-Async: enable` 头，并轮询 `/api/v1/tasks/{task_id}` 获取结果（有效期24小时）[原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **[多模态输入](../concepts/multi-modal-input.md)**：视觉理解与全模态模型支持混合输入（如文本+图片+视频），但需注意 `qwen3.7-plus` 最大图片数2048张、视频数64个；`qwen3.5-omni-plus` 视频最大时长3小时 [原文标题](../../raw/model-user-guide/model-experience/vision-model.md)。 +- **语言与方言**：Fun-ASR 支持超100种语言及方言（含吴语、闽南语等），而 Qwen3.5-Livetranslate 仅对60种语言输出语音，“仅文本”语言不生成音频 [原文标题](../../raw/model-user-guide/model-experience/asr-model.md)。 + +## 使用方式 + +- **协议选择**： + - 实时交互（语音助手、直播字幕）→ **WebSocket**（`qwen-audio-3.0-realtime-plus`、`fun-asr-realtime`）； + - 批量处理（文件转写、视频分析）→ **HTTP**（`qwen3.5-omni-plus`、`fun-asr`）； + - 超长任务（3D生成、视频合成）→ **异步HTTP**（Tripo、HappyHorse）[原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **认证**：所有请求需携带 `Authorization: Bearer $DASHSCOPE_API_KEY`，API Key 必须在对应地域（如Tripo仅限华北2）开通并配置 [原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **SDK支持**：Qwen-Audio-TTS/CosyVoice、Fun-ASR 支持 Python/Java/Android/iOS SDK；其他模型建议直接调用REST API。 +- **快速验证**：新项目优先选用 `qwen3.7-plus`（文本）、`wan2.7-image-pro`（图像）、`qwen3.5-omni-plus`（多模态）进行效果验证，再按成本/性能需求降级至 `-flash` 系列。 + +## 限制和注意事项 + +- **地域限制**：Tripo 3D、Fun-Music 仅在华北2（北京）可用；部分Wan视频模型（如 `wan2.6-t2v-us`）专用于美国地域 [原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **资源约束**： + - 视频理解：`qwen3.7-plus` 单次请求最大视频大小2GB、时长2小时； + - 图像分辨率：每张图[Token](../concepts/token.md)消耗 = `h × w / (32 × 32) + 2`，高分辨率显著增加成本； + - 重排序：`qwen3-rerank` 最多处理500个文档，超限需分批。 +- **能力冲突**： + - 联网搜索与 Function Calling 不可同时启用（Qwen3.5-Omni）； + - 思考模式下不支持语音输出（Qwen3-Omni-Flash HTTP 模式除外）； + - `qwen-long`（1000万上下文）不支持 Function Calling 与内置工具 [原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +- **版本兼容性**：旧版模型（如 `qwen2.5-omni-7b`、`paraformer`）已停止更新，新项目应使用 Qwen3.5+ 或 Fun 系列；快照版本（如 `qwen3.7-plus-2026-05-26`）用于生产环境稳定性保障。 +- **成本提示**：`-flash` 后缀模型普遍比 `-plus` 成本低30–50%，但部分能力受限（如 `deepseek-v4-flash` 不支持内置工具）[原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 ## 来源文档 - [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md) -- [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md) -- [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md) - [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md) +- [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md) +- [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md) - [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - [语音合成](../../raw/model-user-guide/model-experience/tts-model.md) -- [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md) -- [语音识别](../../raw/model-user-guide/model-experience/asr-model.md) - [音乐生成](../../raw/model-user-guide/model-experience/fun-music.md) -- [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md) +- [语音识别](../../raw/model-user-guide/model-experience/asr-model.md) - [全模态](../../raw/model-user-guide/model-experience/omni.md) +- [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md) +- [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md index 712b5a20..b7d8e11f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md @@ -1,93 +1,42 @@ # model high speed inference -百炼平台针对高吞吐、高速率的推理场景提供两类能力：**TPM 预留**用于为指定模型锁定专属推理容量，避免业务高峰期受公共限流影响；**快速模式（Fast mode）**则针对输出速度敏感的场景提升 TPS。两者都通过替换或指定 `model` 参数接入，无需大改代码。 - -## TPM 预留：锁定专属容量 - -TPM（Tokens Per Minute）预留为指定模型锁定专属的推理吞吐量，预留容量内的调用不受公共资源限流影响，容量为业务专属、不与其他用户共享。详见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md)。 - -核心特性： - -- **专属模型 code**：创建预留后系统自动生成专属模型 code，需将 API 请求中的 `model` 参数替换为该 code 才能使用预留容量。 -- **超额不中断**：超出预留容量的请求自动降级为按量计费处理，服务不中断，无需修改代码。可在详情页的**超额降级统计**查看降级次数。 -- **计费单位**：按 kTPM 预付费（1 kTPM = 1,000 Tokens/分钟），一次性支付，从购买成功起连续生效。 - -### 方案选型 - -[TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 文档给出了多种容量方案的对比，便于按业务诉求选型： - -| 方案 | 计费单位 | 容量保障 | 适用场景 | 超额处理 | 代码改动 | -| --- | --- | --- | --- | --- | --- | -| 按量付费 | 按 token | 无（共享公共池） | 流量波动大/短期 | 自动服务，受公共限流 | 无需改动 | -| 资源包/节省计划 | 预付费额度 | 承诺用量折扣（非专属） | 费用优化 | 超出转按量 | 无需改动 | -| TPM 预留 | 按 kTPM 预付费 | 专属容量刚性兑付 | 流量可预估、不能接受限流 | 超出自动降级公共池按量，不中断 | 替换 model 参数 | -| PTU 专属部署 | 按 kTPM 预付费 | 专属部署实例 | 高吞吐高性能 | 超出转按量 | 替换 model 参数 | - -### 创建与接入 - -1. 登录百炼控制台创建 TPM 预留，填写预留名称、选择模型、付费周期（按天）、输入/输出 TPM（单位 kTPM）、购买时长（支持 1~30、60、90、120、365 天）等参数。建议先用 **TPM 容量计算器**（根据 RPM、平均输入/输出长度、缓存命中率估算）确认所需额度。 -2. 确认费用后完成支付。 -3. 在详情页**概览** Tab 复制**专属模型 code**。 -4. 将 API 请求的 `model` 参数替换为该 code 即可： - -```python -import dashscope - -response = dashscope.Generation.call( - api_key="your-api-key", - model="your-dedicated-model-code", # 替换为专属模型 code - messages=[{"role": "user", "content": "你好"}], -) -print(response.output.text) -``` - -> **注意**：短时间内请求量快速拉升时，系统需短暂预热以匹配算力，预热期间部分请求可能出现延迟波动，请做好请求排队或重试机制。 - -### 容量换算参数 - -部分模型支持长输入阶梯系数和缓存折扣，容量计算器会自动应用。例如 glm-5.2 输入长度上限 1M、缓存折扣 0.25、无阶梯；glm-5.1 缓存折扣 0.2，且在 [32K, 200K] 区间输入系数 1.33 / 输出 1.17；deepseek-v4-pro 缓存折扣低至 0.08。Qwen3.6-flash-2026-04-16 不支持缓存。 - -### 管理与生命周期 - -- **扩缩容**：在详情页调整输入/输出 TPM，变配期间服务不中断。利用率持续接近 100% 或频繁降级时建议扩容。 -- **续费**：可手动续订，或开启**到期自动续费**（到期前一天 08:00 自动扣款）。 -- **退订**：跳转费用中心完成，退订后专属模型 code 失效、请求回退公共资源，不可恢复。缩容/退订退费按已使用部分 1.5 倍系数结算：`退款 = 降量部分预付费 - (降量部分预付费 × 已用时长/购买时长 × 1.5)`。 - -预留实例状态：服务到期后 2 小时内仍为**运行中**（可调用、可续费）；2~14 小时转为**已停止**（不可调用、仍可续费）；到期 14 小时后**已过期/删除**，不可恢复。其余状态包括待生效、变配中、已取消。 - -## 快速模式（Fast mode）：提升输出速度 - -[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 面向对输出速度敏感的场景（AI 编程助手、Agent 多步推理、实时对话等），当前处于 **preview 阶段**，能力与规格可能随版本调整。 - -关键特性： - -- **高速输出**：TPS 提升至标准 API 的 1.5~2 倍，达 80~100 TPS。 -- **按 token 计费**：计费逻辑与标准 API 一致，按输入/输出 token 计费。 -- **特殊限流**：超出 TPM 额度不会立即限流，请求进入排队队列。 - -### 接入方式 - -将 `model` 参数指定为支持的模型 ID（如 `glm-5.2-fast-preview`）即可开启，无需额外参数。接入域名格式为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`，其中 `{workspace_id}` 可在业务空间管理页面切换到对应地域后查看。 - -```bash -curl -X POST https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ - -H "Authorization: Bearer $API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "glm-5.2-fast-preview", - "messages": [{"role": "user", "content": "你是谁"}], - "stream": false -}' -``` - -glm-5.2 默认返回 `reasoning_content` 思考字段；[流式输出](../concepts/streaming-output.md)时思考内容与回答内容分别通过 `delta.reasoning_content` 与 `delta.content` 推送。 - -## 限制与注意事项 - -- **两者定位不同**：TPM 预留解决"容量保障/不受限流"，快速模式解决"输出速率提升"。快速模式仍为 preview，生产环境需评估稳定性。 -- **接入差异**：TPM 预留通过替换为专属模型 code 接入标准 dashscope 域名；快速模式使用带 `{workspace_id}` 的 maas 域名并指定 fast-preview 模型 ID。 -- **限流行为差异**：TPM 预留超额自动降级按量、不中断；快速模式超额不立即限流而是排队。 -- **计费口径**：具体价格、容量换算与费用以百炼控制台为准，[TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 与快速模式的价格表可能随时间调整，请以控制台实时展示为准。 +百炼平台提供两种面向高吞吐、低延迟场景的推理加速能力：**TPM 预留（[Token](../concepts/token.md) Per Minute Reservation）** 保障专属容量与确定性 SLA，**快速模式（Fast Mode）** 提升单请求输出速度与 TPS。二者适用不同优化目标——前者解决“容量争抢”问题，后者解决“响应延迟”问题，可独立或组合使用（如在 TPM 预留实例上启用 fast-preview 模型）。 + +## 支持的模型与功能 + +- **TPM 预留**：为指定模型锁定专属输入/输出吞吐量（单位：kTPM），支持千问、GLM、DeepSeek、Kimi 等主流模型，覆盖华北2（北京）和新加坡地域。详情见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md)。 +- **快速模式**：当前仅支持 `glm-5.2-fast-preview` 模型（北京/新加坡双地域），提供 1.5~2 倍标准 API 的 TPS（达 80~100 TPS），适用于 AI 编程助手、Agent 多步推理等对输出流速敏感的场景。该能力处于 preview 阶段，规格可能调整，详见 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md)。 +- > **注意**：两文档对 `glm-5.2` 的缓存折扣描述存在差异——[TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 中明确其缓存命中部分按 25% 折算容量；而 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 仅列出“缓存命中”单价为 4 元（北京），未说明容量折算逻辑。实际容量计算请以控制台实时参数为准，或参考 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 中的阶梯系数与缓存折扣规则。 + +## 关键参数 + +| 参数 | TPM 预留 | 快速模式 | +|------|----------|-----------| +| **核心指标** | 输入TPM / 输出TPM（kTPM） | TPS（80~100）、输出延迟降低 | +| **计费单位** | 预付费（按天，kTPM） | 按 token（输入/输出分别计费） | +| **容量保障** | 刚性兑付专属容量，不共享 | 无专属容量，依赖公共资源池，超限请求排队而非直接限流 | +| **溢出策略** | 可选：自动溢出至按量（默认）或返回 429 | 请求排队，不返回 429（但排队时延增加） | +| **缓存支持** | 支持（如 glm-5.2 缓存命中按 25% 折算输入容量） | 支持（明确列出缓存命中单价） | + +## 使用方式 + +- **TPM 预留**：创建后获取专属 `model` code，在 API 请求中替换标准模型 ID 即可生效。需确保实例状态为“运行中”，且调用域名与标准 API 一致（如 `https://dashscope.aliyuncs.com/...`）。详见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 中的“创建 TPM 预留”与“API 接入”章节。 +- **快速模式**：直接使用 `glm-5.2-fast-preview` 作为 `model` 参数，并切换至专属接入域名（格式：`https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`）。无需额外 header 或 query 参数。流式调用时需解析 `delta.reasoning_content` 和 `delta.content` 字段。示例见 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md)。 +- **组合使用**：可在已创建的 TPM 预留实例上，将 `model` 参数设为 `glm-5.2-fast-preview`（前提是该模型 code 已开通快速模式支持），从而同时获得容量保障与高速输出。 + +## 限制和注意事项 + +- **TPM 预留**： + - 创建后需等待实例“运行中”状态方可调用；短时间内请求量激增会触发系统预热，预热期可能出现延迟波动，建议实现客户端重试或排队机制。 + - 缩容/退订会产生违约金（已用部分按 1.5 倍系数结算），且退订后专属 model code 失效。 + - 服务到期后 14 小时内资源不可恢复，务必提前续费。 +- **快速模式**： + - 当前仅 `glm-5.2-fast-preview` 可用，不支持其他模型；preview 阶段能力可能变更，不建议用于生产环境长期依赖。 + - 排队机制不保证端到端延迟上限，高并发下排队时延可能显著上升。 + - 返回结构含 `reasoning_content` 字段，需适配解析逻辑（尤其流式场景）。 +- **通用限制**： + - 两种能力均不改变模型本身的能力边界（如上下文长度、token 限制），具体参数以各模型文档为准。 + - TPM 预留与快速模式的计费相互独立：TPM 预留覆盖其专属容量内调用，超出部分按量计费；快速模式所有调用均按 token 计费，不受 TPM 预留额度影响（除非 model code 显式绑定快速模式）。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md index 6aa8dc00..d6bdb323 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md @@ -1,75 +1,75 @@ # model monitoring -阿里云百炼提供两套互补的用量与监控能力：**模型用量**（控制台聚合视图，用于查看调用量、Token 消耗和费用）与**模型监控**（面向指标、告警、日志的可观测体系）。前者侧重成本与免费额度管理，后者侧重性能、错误、安全指标的采集、告警与对话审计，二者共同覆盖从成本控制到线上运维的完整链路。 +模型监控是百炼平台提供的核心可观测性能力，用于实时跟踪模型调用行为、性能指标、成本消耗及异常事件。它覆盖从基础用量统计到细粒度推理日志的全链路数据采集，并支持告警、自定义分析与第三方集成。该功能面向生产环境运维与成本治理场景，为开发者提供分钟级（高级监控）或小时级（普通监控）的数据洞察。 -## 支持的模型与功能范围 +## 支持的模型/功能 -- **用量查看**：模型列表中的所有模型均支持查看用量，包括基于它们调优后的自定义模型。详见 [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)。 -- **模型监控**： - - **普通监控**支持所有模型（含调优后的自定义模型），延迟通常为小时级。 - - **高级监控**支持北京、新加坡、弗吉尼亚地域下的所有模型，可提供分钟级数据洞察。 - - **告警功能**支持北京、新加坡地域下的所有模型。 -- 监控可查看调用记录、指标监控与告警（Token、延时、调用时长、RPM、TPM、失败率）、以及 Token 消耗统计。详见 [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)。 +- **监控覆盖范围**： + - 普通监控支持[所有在模型列表中可选的模型](../../raw/model-user-guide/model-monitoring/model-telemetry.md)，包括基于其调优的[自定义模型](../../raw/model-user-guide/model-monitoring/model-telemetry.md)； + - 高级监控（含推理日志、TPS、Prometheus API等）仅支持部署在**华北2（北京）、华东2（上海）、新加坡、美国弗吉尼亚**地域的模型； + - 告警功能当前仅支持**北京、新加坡、弗吉尼亚**地域的模型。 -## 用量与费用查看 +- **关键功能模块**： + - **调用记录追踪**：支持按 Request ID 查看单次调用的输入、输出、状态码、用量（[Token](../concepts/token.md) 数）等； + - **多维指标监控**：涵盖安全（内容安全错误次数）、成本（平均单次请求调用量、[Token](../concepts/token.md) 消耗）、性能（RPM、TPM、调用时长、首 [Token](../concepts/token.md) 延时、非首 Token 延时）、错误（失败率、限流错误次数）四大类； + - **历史对话审计**：仅对明确[支持的模型](../../raw/model-user-guide/model-monitoring/model-telemetry.md)（如 `qwen3-plus`、`qwen-flash`、`deepseek-v3.2` 等快照版本）开放，且需提前开通推理日志； + - **主动告警**：支持基于失败率、TPM 突增、首 Token 延时超阈值等条件创建多级告警（INFO / WARNING / ERROR / CRITICAL），通知方式含邮件、短信、钉钉机器人等； + - **Grafana 与自建应用集成**：通过私有 Prometheus 实例暴露标准 HTTP API，支持 `model_call_count`、`model_usage`、`model_tps_per_request` 等 [完整监控指标列表](../../raw/model-user-guide/model-monitoring/model-telemetry.md) 查询。 -数据按[业务空间](https://help.aliyun.com/zh/model-studio/use-workspace)维度统计，不支持按阿里云账号维度统计。 +> **注意**：文档 1 中称“模型列表中的所有模型均支持查看用量”，而文档 2 明确指出**推理日志、TPS、分钟级延迟监控、Grafana 接入等高级能力仅限特定地域且依赖模型快照版本支持**。二者不矛盾，但需区分“基础用量统计”与“高级监控能力”的适用边界——前者全域通用，后者受地域与模型版本双重约束。 -- **数据延迟约 1 小时**；不支持查看 30 天以前的统计数据，更早数据需前往「费用与成本」页面查询。 -- **时间精度**：支持分钟 / 小时 / 天三种精度。时间跨度超过 1 天时分钟精度不可选，超过 7 天时仅支持按天查看。 -- **筛选维度**：仅「大语言模型」页签支持按推理类型（实时推理 / 批量推理）筛选；支持按 API-KEY、模型名称（如 `qwen-plus`）筛选。 -- **费用概览**：可查看当前账期总消费、订阅费用、账单趋势（按月/按天，可按产品分类、API Key ID、模型筛选），并可设置**费用告警**。 +## 关键参数 -不同模型的用量统计口径不同：大语言模型 / 全模态 / 向量模型按 **Token**，图像生成按**张**，视频生成按**秒**，语音模型按**秒、字符或 Token**（视模型而定）。完整口径见 [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)。 +| 参数 | 说明 | 来源约束 | +|------|------|----------| +| `workspace_id` | 业务空间 ID，所有监控数据按此维度隔离与聚合 | 必填（Prometheus 查询、控制台筛选均依赖） | +| `model` | 模型 Code（如 `qwen-plus`），区分大小写 | 必填（日志、告警、API 查询均需精确匹配） | +| `apikey_id` | API Key ID（非密钥本身），用于归因调用来源；值为 `-1` 表示来自控制台调用 | 可选，但推荐用于多 Key 成本分摊 | +| `protocol` / `sub_protocol` | 协议类型（HTTP/SSE/WS）与子协议（DEFAULT/ASYNC），影响延时与吞吐特征 | 高级监控专属标签，用于精细化分析 | +| `start` / `end` / `step` | Prometheus 查询时间范围与步长；`step=60s` 为分钟级精度最低要求 | 仅高级监控 API 支持，普通监控无此接口 | -## 免费额度管理 +## 使用方式 -「免费额度」页面提供使用概览（按模型总数、额度充沛、使用超 50%/80%、无免费额度等维度汇总）及「即将用尽 Top 3」列表。 +1. **基础用量查看**： + 进入[模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)页面，按模型类型、时间范围（≤30 天）、API Key 筛选，查看 Token/张/秒等用量汇总。数据延迟约 **1 小时**（见[模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 -- **免费额度用完即停**：开启后免费额度用尽时服务自动停止（返回 `403 AllocationQuota.FreeTierOnly`），避免产生额度外费用。 -- 支持批量开启/关闭、一键开启/关闭所有模型；账号未绑定有效支付方式时批量操作会失败。 +2. **模型监控详情**： + 进入[模型监控](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)页面 → 点击目标模型操作列的 **监控** 或 **日志**： + - “监控”页签查看调用统计（失败详情可下钻）、性能趋势（RPM/TPM/延时）； + - “日志”页签查看单次调用明细（需已开通推理日志，且模型在[支持列表](../../raw/model-user-guide/model-monitoring/model-telemetry.md)中）。 -> **注意**：「免费额度用完即停」只能在账户仍有未消耗免费额度时开启；一旦开启，需在免费额度完全消耗后才能关闭。控制台免费额度数据为分钟级更新，账单记录按分钟汇总，请以控制台显示数值为准。 +3. **开通高级能力**： + 在模型监控页面右上角点击 **模型监控配置** → 开启： + - **审计日志 + 推理日志**（用于日志回流、内容审计）； + - **性能和用量指标监控**（启用 Prometheus 数据源与 TPS 等指标）。 -## 监控指标与告警 +4. **创建告警**： + 进入[模型告警](https://bailian.console.aliyun.com/?tab=model#/model-alert)页面 → **创建告警规则** → 选择模型、模板（如“TPM 异常突增”）、阈值、通知渠道。 -在模型监控列表中点击目标模型操作列的**监控**，可查询 4 类指标： +5. **接入 Grafana / 自建系统**： + 获取 Prometheus HTTP API 地址后，使用标准 PromQL 查询，例如： + ```http + GET {API}/api/v1/query_range?query=model_usage{workspace_id="llm-xxx",model="qwen-plus"}&start=2025-11-20T00:00:00Z&end=2025-11-20T23:59:59Z&step=60s + ``` -- **安全**：如 `内容安全错误次数`（输入/输出被内容安全服务拦截）。 -- **成本**：如 `平均单次请求调用量`。 -- **性能**：`调用时长`、`首 Token 延时`、RPM、TPM、非首 Token 延时等。 -- **错误**：`失败次数`、`失败率`，其中**限流错误次数**指因 [429 状态码](https://help.aliyun.com/zh/model-studio/error-code)导致的失败。 +## 限制和注意事项 -**创建告警**（仅限新加坡、华北2（北京）地域）：需先开启高级监控（性能和用量指标监控），再在模型告警页面创建规则。 +- **数据延迟**： + - 普通监控（用量汇总、失败率等）延迟 **1–2 小时**； + - 高级监控（推理日志、Prometheus 指标）延迟为**分钟级**，但首次同步需等待日志开通生效。 -- **通知方式**：短信、电子邮件、电话、钉钉群机器人、企业微信机器人、Webhook。 -- **告警等级**：紧急（电话/短信/邮件）、错误（短信/邮件）、警告（短信/邮件）、普通（邮件），不支持自定义。 +- **地域与模型限制**： + - 推理日志、TPS 指标、Grafana 集成、分钟级告警仅支持北京/上海/新加坡/弗吉尼亚地域； + - 并非所有模型都支持请求/响应内容记录，具体以[支持列表](../../raw/model-user-guide/model-monitoring/model-telemetry.md)为准；不支持的模型在日志页签会明确提示“当前模型暂不支持日志”。 -## Token 消耗与历史对话 +- **免费额度联动**： + - “免费额度用完即停”开关仅影响计费行为（返回 403），**不影响监控数据采集**；用量与告警仍持续上报，便于及时发现额度耗尽风险（见[模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 -- **历史 Token 消耗**：最近 30 天可在监控页调用统计的「调用量」区域查看；更早数据前往「费用与成本」页面。 -- **单次调用 Token 消耗 / 历史对话（模型日志）**：需在「模型监控配置」中依次开通审计日志和推理日志，之后在**日志**页签查看每次调用的输入、输出与用量。开通后从调用到记录存在分钟级延迟。 +- **权限约束**： + - 主账号默认可查看全部业务空间数据；子账号仅能访问其所属业务空间，且需被授予 `AliyunBaiLianFullAccess` 或等效权限才能开通日志与告警。 -> **注意**：查看某次调用的 Token 消耗及历史对话（模型日志）功能**目前仅适用于华北2（北京）地域的部分模型**，且仅覆盖特定模型/快照版本（如 qwen3-max、qwen-plus、qwen3-coder 系列、部分开源与三方模型）。详见 [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)。 - -## 接入 Grafana 与自建应用 - -高级监控的指标数据存储在私有 Prometheus 实例中，支持标准 Prometheus HTTP API，可接入 Grafana 或自建应用做可视化分析。 - -1. 确保已开启高级监控，在模型监控配置中查看 Prometheus 实例详情，按网络环境（公网/VPC）复制 HTTP API 地址。 -2. 通过 `GET {HTTP API}/api/v1/query_range?query=<指标名>&start=...&end=...&step=60s` 查询，`Authorization` 需用 `Basic base64Encode(AccessKey:AccessKeySecret)`。 - -常用指标名包括 `model_call_count`（调用次数）、`model_call_duration`（调用时长均值）、`model_usage`（用量总和）等；可用 `{workspace_id="...",model="qwen-plus"}` 形式追加过滤条件（支持 `user_id`、`apikey_id`、`workspace_id`、`model`、`protocol`、`status_code`、`usage_type` 等 LabelKey）。 - -> **注意**：`status_code`、`error_code` 仅 `model_call_count` 支持；`usage_type` 仅 `model_usage` 支持。AccessKey 必须与 Prometheus 实例归属同一阿里云账号。 - -## 生产环境实践建议 - -- **控制输出长度**：合理设置 `max_tokens` 与限制思考长度以控制费用。 -- **按任务选模型**：分类、摘要等简单任务优先用轻量级模型。 -- **监控与告警**：通过模型监控跟踪用量趋势并配置告警。 -- **优化 Prompt**：简洁清晰的 Prompt 可减少输入 Token 消耗。 -- **使用批量推理**：非实时大批量任务用批量推理更具成本优势。 +- **历史数据不可追溯**： + - 推理日志仅记录**开通后**的调用；开通前的历史请求无法补录。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md b/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md index 5a84d01d..f002369f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md @@ -1,158 +1,45 @@ # plug in -百炼插件是一个工具集合，用于扩展大模型的能力边界。一个插件下可包含多个工具（API），每个工具实现特定功能。通过将插件集成到大模型应用中，可弥补大模型在获取最新信息、精确计算、图像处理等方面的不足。百炼支持官方插件、三方插件和自定义插件三类。详见[插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md)。 +插件是百炼平台用于扩展大模型能力的关键机制，通过集成外部工具（如计算器、搜索、代码执行等），弥补大模型在实时信息获取、精确计算、多模态生成等方面的固有局限。插件以工具（Tool）为最小单元，支持官方预置、三方认证及用户自定义三类来源，调用过程由模型自主规划或工作流显式编排。所有插件均需通过服务关联角色授权后方可使用。 -## 插件分类 +## 支持的模型与功能 -- **官方插件**：组件广场预置，无需配置输入输出参数即可直接调用。 -- **三方插件**：涵盖商业服务、图像视频、学习教育等领域，经过效果测试，开通后直接调用，无需额外配置。 -- **自定义插件**：当官方和三方插件无法满足业务需求时，用户可创建或从云市场导入自定义插件，集成到应用中。 +百炼当前支持以下模型调用插件：`qwen-turbo`、`qwen-plus`、`qwen-max`、`qwen-vl-max`、`qwen-vl-plus`。各模型对插件的兼容性存在差异，**实际可用性请以控制台运行结果为准**，而非静态列表 [插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md)。 +插件功能分为三类： +- **官方插件**：开箱即用，无需配置参数，包括 `code_interpreter`（Python代码执行）、`calculator`（复杂数学计算）、`text_to_image`（文生图）、`quark_search`（实时网络搜索）、`generate_qrcode`（URL转二维码）、`github_search`（GitHub项目检索）； +- **三方插件**：覆盖商业服务、图像视频、教育等领域，需开通后使用，详见 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)； +- **自定义插件**：支持用户按 OpenAPI 规范定义并注册，实现业务专属能力集成。 -## 官方插件列表 +> **注意**：文档 1 中称“官方插件无需配置输入输出参数”，而文档 2 在“Python代码解释器”小节明确列出其依赖库清单及网络/文件访问限制，表明**参数约束与运行环境限制实际存在，不可忽略**。开发者应以 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) 中的运行时说明为准。 -| 插件名称 | 工具 ID | 说明 | [计费](../concepts/billing.md)方案 | -| --- | --- | --- | --- | -| Python 代码解释器 | `code_interpreter` | 执行 Python 代码片段，如数学计算、数据分析与可视化、数据处理 | 免费 | -| 计算器 | `calculator` | 进行复杂数学计算，例如计算 `12313x13232` | 免费 | -| 图片生成 | `text_to_image` | 基于文本生成图片，例如"请画一只在笑的小狗" | 限时免费，需申请开通 | -| 夸克搜索 | `quark_search` | 搜索实时信息，查找公开网络知识和信息 | 限时免费，需申请开通 | -| 生成二维码 | `generate_qrcode` | 根据网站链接地址生成二维码 | 免费 | -| GitHub 搜索 | `github_search` | 在 GitHub 中搜索相关项目列表 | 免费 | +## 关键参数 -> **注意**：夸克搜索插件目前支持检索网页标题、关键词和摘要，但不支持直接访问网页详情。GitHub 搜索插件支持检索项目标题、链接和摘要，不支持访问项目详情。Python 代码解释器不支持对外访问网络以及上传本地文件。 - -Python 代码解释器可用依赖包括：matplotlib、pandas、scipy、seaborn、sympy、pillow、pydantic~=1.10.8、requests~=2.31.0、oss2~=2.18.1、pdfminer-six、pypdf、python-pptx、wordcloud 等。详见[官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)。 - -## 支持的模型 - -插件调用对模型有一定要求，目前支持以下模型： - -| 模型 | 模型标识符 | -| --- | --- | -| 通义千问-Turbo | qwen-turbo | -| 通义千问-Plus | qwen-plus | -| 通义千问-Max | qwen-max | -| 通义千问VL-Max | qwen-vl-max | -| 通义千问VL-Plus | qwen-vl-plus | - -> **注意**：各模型对插件的兼容性可能有差异，最新兼容性状态以控制台实际执行结果为准。 - -## 插件调用机制 - -调用插件的本质是调用插件下的工具。百炼支持通过[智能体应用](../concepts/agent-application.md)、[工作流](../concepts/workflow.md)应用以及 Assistant API 调用插件。 - -- **[智能体应用](../concepts/agent-application.md) / Assistant API**：大模型根据用户输入内容、工具名称和工具描述判断是否调用工具。需要调用时，模型选择合适工具，应用内部完成调用后将工具返回结果与用户内容合并再次输入模型，由模型生成最终结果；无需调用时直接生成结果输出。 -- **[工作流](../concepts/workflow.md)应用**：插件作为[工作流](../concepts/workflow.md)的一个节点，按用户编排的方式执行特定任务，而非由模型主动规划和调用。 +插件调用的核心参数为 `tool_id`（如 `calculator`），用于标识具体工具。通过 API 调用时必须准确传递该 ID。 +工具输入参数由插件定义，例如： +- `calculator` 接收 `payload__input__text` 字段（如 `"12313x13232"`）； +- `quark_search` 接收查询文本，返回结构化摘要（标题、关键词、摘要），**不返回原始网页内容**； +- `code_interpreter` 仅支持指定依赖（如 `pandas`, `matplotlib`, `sympy`），且**禁止网络访问与本地文件上传**，详见 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)。 ## 使用方式 -### 首次访问授权 - -主账号或 RAM 用户首次访问插件页面时，需授权服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI`（权限策略 `AliyunServiceRolePolicyForSFMAccessCloudAPI`），用于授权百炼访问云市场商品清单并根据插件配置进行 API 调用。 - -- **主账号**：在插件页面勾选条款，单击"授权并进入"即可。 -- **RAM 用户（子账号）**：会因缺少创建服务关联角色权限报错（错误码 140052）。需先由主账号在 RAM 控制台创建自定义权限策略（Action 为 `ram:CreateServiceLinkedRole`，Condition 中 `ram:ServiceName` 为 `cloundapi-access.sfm.aliyuncs.com`），并授予子账号该策略后，再完成授权。 - -### 调用插件 - -- **方式一（插件页面）**：在插件页面将工具添加至[智能体应用](../concepts/agent-application.md)。官方插件只能与位于相同[业务空间](../concepts/workspace.md)里的[智能体应用](../concepts/agent-application.md)关联。每个[智能体应用](../concepts/agent-application.md)最多支持添加 10 个工具，应用会根据输入选择调用一个或多个工具。 -- **方式二（应用管理页面）**：在指定智能体或工作流应用内添加插件，测试效果并发布应用。 -- **方式三（Assistant API）**：通过 Assistant API 调用工具，需正确传递工具 ID。 - -子[业务空间](../concepts/workspace.md)调用官方插件前，需先在插件详情页为子[业务空间](../concepts/workspace.md)授权；默认[业务空间](../concepts/workspace.md)无需此步骤。 - -### 获取工具 ID - -通过 API 调用工具时需正确传递工具 ID。在插件页面找到目标插件，单击"查看详情"，在"插件工具"下获取工具 ID（例如 `calculator`）。 - -## 自定义插件 - -当官方和三方插件无法满足业务需求时，可创建自定义插件。详见[自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 - -### 工作流程 - -1. **创建/导入插件**：定义插件基础信息，或直接从云市场导入。 -2. **添加工具**（导入插件无需此步）：配置 API 路径、请求参数和返回数据。 -3. **调试与发布**：在线测试 API 连通性，功能正常后发布。只有已发布的工具才能在应用中被调用。 -4. **在应用中使用**：将插件关联到智能体，通过对话测试或 API 集成调用。 - -### 创建插件关键参数 - -- **插件名称**：支持中英文，需具语义。 -- **插件描述**：对插件功能和使用场景的简要说明，帮助大模型判断是否调用，使用自然语言描述。 -- **插件 URL**：插件访问地址。同一域名下不同路径拆分为不同 API（工具路径）。 -- **是否鉴权**：支持服务级鉴权和用户级鉴权。鉴权信息可放在 Header（默认参数名 `Authorization`）或 Query 中；Type 支持 `basic`（[Token](../concepts/token.md) 前不增加内容）、`bearer`（[Token](../concepts/token.md) 前增加 "Bearer"）、`appcode`（[Token](../concepts/token.md) 前增加 "APPCODE"）。 - -### 创建工具关键参数 - -- **工具名称**：支持中英文，有字符数限制（20 字符）。 -- **工具描述**：帮助大模型判断是否调用该工具，尽量给出使用示例。 -- **工具路径**：指向插件 URL 的相对路径，必须以正斜杠（/）开头。 -- **请求方法**：GET 或 POST。 -- **提交方式**：`application/json` 或 `application/x-www-form-urlencoded`。 -- **输入参数传参方式**： - - **大模型识别**：参数值由大模型从用户输入中提取。 - - **业务透传**：参数值从外部主动透传，通过 `biz_params` 和 `user_defined_params` 传递。 -- **高级配置**：提供调用示例（Query 与期望入参 Value），减少漏召回和误召回。 - -> **注意**：Object 类型下的子属性不能为空，需点击对象行末图标新增子属性。GET 请求方法下的输入参数不支持 Object 类型。 - -### 从云市场导入 - -云市场提供丰富 API，可在云市场开通后导入至百炼插件列表。导入的插件为草稿状态，需测试、发布后使用。导入时系统自动填充出入参，但可能存在信息缺失，发布时需根据错误提示修正。 - -### 使用自定义插件 +插件可通过三种方式集成： +1. **智能体应用**：在插件市场页面选择工具 → “添加至智能体” → 绑定同业务空间内的智能体应用（最多 10 个工具）； +2. **工作流应用**：将插件作为独立节点拖入流程，由用户显式编排执行顺序，不依赖模型自动决策； +3. **Assistant API**：在请求体中传入 `tools` 数组（含 `tool_id` 和 `description`），模型根据 `user_message` 自主选择调用；详细用法见 [插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md)。 -控制台内可将插件发布为 MCP 服务，再在[智能体应用](../concepts/agent-application.md)编排页面的 MCP 区块添加该服务；也可直接在应用管理页面的[智能体应用](../concepts/agent-application.md)编排中添加 MCP 服务。无鉴权插件可直接对话测试；用户级/服务级鉴权需在对话前配置鉴权 [Token](../concepts/token.md)；业务透传参数需配置变量值。从云市场导入的插件无需在对话页输入鉴权 [Token](../concepts/token.md)。 - -通过 API 调用时，若应用关联的插件存在业务透传参数或开启了用户级鉴权，需通过 `biz_params` 传递鉴权信息或透传参数。 +> **注意**：官方插件仅支持与**同一子业务空间**内的智能体关联；跨空间调用需先完成插件授权操作，具体步骤见 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)。 ## 限制和注意事项 -- 官方插件只能与位于相同[业务空间](../concepts/workspace.md)的[智能体应用](../concepts/agent-application.md)关联。 -- 每个[智能体应用](../concepts/agent-application.md)最多添加 10 个工具。 -- 只有已发布且启用状态、调试状态为成功的工具才能用于调用。 -- 删除插件会删除其下所有工具，调用该插件的应用会失效，操作不可撤回。 -- 编辑插件信息后立即生效；修改 URL、Header、鉴权信息可能影响工具调用，需重新测试并发布工具。 -- 工具信息修改后需重新测试并发布才能生效。 - -### 常见错误码 - -| 错误码 | 错误信息 | 说明 | -| --- | --- | --- | -| 130040 | xx 缺少参数描述信息 | 参数描述缺失，补充后重新发布 | -| 130022 | 保存工具信息异常/请检查示例参数是否正确 | 原因一：Object 类型参数子属性为空，需新增子属性；原因二：GET 请求下存在 Object 类型输入参数，需选择其他类型 | - -## 常见问题 - -**夸克搜索和联网搜索（enable_search）有什么区别？** - -- **夸克搜索插件**：模型直接调用插件执行搜索，将搜索结果以文本形式返回，可直接用于生成最终输出。 -- **联网搜索（enable_search）**：同样基于夸克搜索，但模型仅利用互联网信息丰富生成内容，不会完全依赖或返回互联网搜索结果。 +- **权限前提**：首次使用插件前，主账号或 RAM 子账号必须授权 `AliyunServiceRoleForSFMAccessCloudAPI` 服务关联角色，否则无法访问插件市场； +- **RAM 子账号特殊处理**：子账号需额外授予 `ram:CreateServiceLinkedRole` 权限（含 `Condition` 限定 `cloundapi-access.sfm.aliyuncs.com`），否则授权失败（错误码 140052）； +- **组合调用**：支持单次请求中调用多个插件（如 `quark_search` + `text_to_image` + `generate_qrcode`），但需确保各工具语义可协同； +- **计费说明**：`code_interpreter`、`calculator`、`generate_qrcode`、`github_search` 免费；`text_to_image` 与 `quark_search` 为限时免费，需单独申请开通； +- **能力边界**：`quark_search` 仅返回摘要，不支持跳转原文；`github_search` 仅返回项目元信息（标题、链接、摘要），不支持克隆或读取代码库详情。 ## 来源文档 -- [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) - [插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md) -- [自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md) - - - - - - - - - - - - - - - - - - - +- [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md b/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md index 1a567cf3..a4fafbc6 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md @@ -1,96 +1,49 @@ # prompt -阿里云百炼提供了一套完整的 Prompt 工程工具链，帮助开发者高效管理和优化提示词。核心能力包括 Prompt 模板（预置与自定义）、Prompt 自动优化、Prompt 样例库以及基于输入输出样例的 Prompt 反馈优化。这些功能覆盖了从模板创建、结构优化到少样本学习引导的完整流程，适用于文本生成、图片生成、智能客服等多种场景。 +Prompt 是百炼平台中驱动大语言模型行为的核心输入指令。通过结构化设计、模板化管理、自动优化及样例引导等能力，开发者可高效构建稳定、可控、可复用的提示工程体系，显著降低模型调用的不确定性与维护成本。所有 Prompt 相关功能均需在华北2（北京）地域使用。 -## Prompt 模板 +## 支持的模型/功能 -Prompt 模板将提示词的固定结构与动态变量分离，实现可复用的统一管理。模板分为**预置模板**和**自定义模板**两类，详见 [Prompt模板概述](../../raw/application-user-guide/prompt/prompt-template.md)。 +百炼平台提供多层次 Prompt 支持能力，覆盖从基础指令到复杂任务编排的全链路： -### 预置模板 +- **Prompt 模板**：支持预置模板（开箱即用，适用于通用场景如文案生成、摘要抽取）和自定义模板（支持文本生成与图片生成两类），后者可基于 [ICIO/CRISPE/RASCEF 等工程框架](../../raw/application-user-guide/prompt/prompt-custom-template.md) 结构化构建，确保指令清晰、角色明确、输出可控 [原文标题](../../raw/application-user-guide/prompt/prompt-template.md)。 +- **Prompt 自动优化**：基于大模型对原始 Prompt 进行结构重组、角色注入、指令增强与安全边界补充，不计费且数据不用于训练 [原文标题](../../raw/application-user-guide/prompt/optimize-prompt.md)。 +- **Prompt 反馈优化**：基于用户提供的输入-输出样例（建议 5–10 条）和评测数据集（建议 ≥20 条），在推理模型（推荐千问-max）上多轮评估迭代，生成更贴合业务实际的 Prompt [原文标题](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md)。 +- **Prompt 样例库（已停用）**：该功能已于近期下线，官方明确要求迁移至 RAG 表格库；当前文档仅作历史参考，**不可新建或启用** [原文标题](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 -由百炼平台提供，涵盖营销文案、摘要抽取、文案润色、商品评论等通用场景，已经过优化，效果稳定，无需额外开发即可通过控制台或 API 调用。预置模板不支持修改，但可通过"复制模板"创建自定义副本后编辑。 +> **注意**：文档 4 中描述的 Prompt 样例库功能已正式废弃，其全部能力由 RAG 表格库承接。若在控制台仍可见相关入口，属界面缓存残留，实际调用将失败。请务必按迁移指南完成数据迁移。 -### 自定义模板 +## 关键参数 -支持两种创建方式： +| 参数 | 说明 | 来源/约束 | +|------|------|-----------| +| `workspaceId` | 业务空间唯一标识，调用所有 Prompt API 的必需参数 | 必须通过 [获取 APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id) 获取 | +| `promptTemplateId` | 模板唯一 ID，用于 `GetPromptTemplate` 等接口 | 预置模板 ID 在控制台卡片中直接可见；自定义模板 ID 创建后生成 | +| `variables` | 模板变量列表（如 `["topic", "platform"]`），由 `GetPromptTemplate` 接口返回 | 填充时需严格匹配变量名，否则渲染为空字符串 | +| `has_thoughts: true` | 启用样例库检索调试（仅限已停用的样例库功能） | 已失效，RAG 表格库使用独立参数 `rag_config` | -- **控制台创建**：在"提示词"页面直接创建，或从预置模板复制后修改。支持"自定义创建"和"基于 Prompt 工程创建"两种输入模式。 -- **API 创建**：通过 `CreatePromptTemplate` 接口创建，需要提供 `workspaceId`（[业务空间](../concepts/workspace.md) ID）。 +## 使用方式 -自定义模板支持文本生成和图片生成两种类型。文本生成模板可选择 ICIO、CRISPE、RASCEF 等 Prompt 工程框架进行结构化设计；图片生成模板支持分别定义正向和负向提示词。具体创建流程参见 [自定义Prompt模板](../../raw/application-user-guide/prompt/prompt-custom-template.md)。 +### 控制台操作 +- **模板创建**：进入「应用开发 > 组件管理 > 提示词」，点击「创建提示词」，选择「文本生成」或「图片生成」类型；文本生成支持「自定义创建」或「基于Prompt工程创建」两种模式 [原文标题](../../raw/application-user-guide/prompt/prompt-custom-template.md)。 +- **模板调用**：在智能体应用配置页，点击「使用prompt」→「创建应用」，模板内容自动填充至系统提示词框，变量以 `${var}` 形式呈现，最大长度 6144 字符。 +- **自动优化**：在「提示词」页面右上角进入「自动优化」，粘贴原始 Prompt，点击「优化」后可复制或「保存为模板」。 -### 模板使用方式 +### API/SDK 调用 +- **获取模板**：调用 `GetPromptTemplate` 接口（[API 文档](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getprompttemplate)），传入 `workspaceId` 和 `promptTemplateId`，响应中包含 `content` 与 `variables` 字段。 +- **反馈优化**：调用 `CreatePromptFeedbackOptimizationTask`（需 SDK V2.0+），上传样例与评测数据 Excel 文件（单次 ≤100 条，文件 ≤20MB），任务完成后可下载优化结果。 +- **模板创建**：使用 `CreatePromptTemplate` 接口，`content` 字段需为字符串（文本生成）或含 `positive_prompt`/`negative_prompt` 的 JSON（图片生成）。 -**控制台**：在模板卡片上点击"创建应用"，模板内容自动填充到[智能体应用](../concepts/agent-application.md)的提示词编辑框中。提示词最大支持 6144 个字符。 +## 限制和注意事项 -**API/SDK**：通过 `GetPromptTemplate` 接口拉取模板内容（需 `workspaceId` 和 `promptTemplateId`），将业务数据填入模板变量后生成最终 Prompt，再发送给目标模型。返回内容包含 `variables`（变量列表）、`content`（模板内容）等字段。 +- **地域限制**：所有 Prompt 功能**仅支持华北2（北京）地域**，跨地域调用将返回错误。 +- **模板长度**：单个模板 `content` 最大 6144 字符（控制台编辑框右下角实时计数），超长将截断。 +- **图片生成模板**：正向/负向 Prompt 均受长度限制，且负向 Prompt 不支持变量插值。 +- **变量安全**：模板变量值若含恶意指令（如 `{{system}}` 或 `