[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-mlflow--mlflow":3,"tool-mlflow--mlflow":61},[4,18,26,36,44,53],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":10,"last_commit_at":24,"category_tags":25,"status":17},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,"2026-04-05T11:01:52",[14,15,13],{"id":27,"name":28,"github_repo":29,"description_zh":30,"stars":31,"difficulty_score":32,"last_commit_at":33,"category_tags":34,"status":17},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",157379,2,"2026-04-15T23:32:42",[14,13,35],"语言模型",{"id":37,"name":38,"github_repo":39,"description_zh":40,"stars":41,"difficulty_score":32,"last_commit_at":42,"category_tags":43,"status":17},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",108322,"2026-04-10T11:39:34",[14,15,13],{"id":45,"name":46,"github_repo":47,"description_zh":48,"stars":49,"difficulty_score":32,"last_commit_at":50,"category_tags":51,"status":17},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[52,13,15,14],"插件",{"id":54,"name":55,"github_repo":56,"description_zh":57,"stars":58,"difficulty_score":32,"last_commit_at":59,"category_tags":60,"status":17},4721,"markitdown","microsoft\u002Fmarkitdown","MarkItDown 是一款由微软 AutoGen 团队打造的轻量级 Python 工具，专为将各类文件高效转换为 Markdown 格式而设计。它支持 PDF、Word、Excel、PPT、图片（含 OCR）、音频（含语音转录）、HTML 乃至 YouTube 链接等多种格式的解析，能够精准提取文档中的标题、列表、表格和链接等关键结构信息。\n\n在人工智能应用日益普及的今天，大语言模型（LLM）虽擅长处理文本，却难以直接读取复杂的二进制办公文档。MarkItDown 恰好解决了这一痛点，它将非结构化或半结构化的文件转化为模型“原生理解”且 Token 效率极高的 Markdown 格式，成为连接本地文件与 AI 分析 pipeline 的理想桥梁。此外，它还提供了 MCP（模型上下文协议）服务器，可无缝集成到 Claude Desktop 等 LLM 应用中。\n\n这款工具特别适合开发者、数据科学家及 AI 研究人员使用，尤其是那些需要构建文档检索增强生成（RAG）系统、进行批量文本分析或希望让 AI 助手直接“阅读”本地文件的用户。虽然生成的内容也具备一定可读性，但其核心优势在于为机器",93400,"2026-04-06T19:52:38",[52,14],{"id":62,"github_repo":63,"name":64,"description_en":65,"description_zh":66,"ai_summary_zh":66,"readme_en":67,"readme_zh":68,"quickstart_zh":69,"use_case_zh":70,"hero_image_url":71,"owner_login":64,"owner_name":72,"owner_avatar_url":73,"owner_bio":74,"owner_company":75,"owner_location":75,"owner_email":75,"owner_twitter":72,"owner_website":76,"owner_url":77,"languages":78,"stars":116,"forks":117,"last_commit_at":118,"license":119,"difficulty_score":32,"env_os":120,"env_gpu":120,"env_ram":120,"env_deps":121,"category_tags":132,"github_topics":134,"view_count":32,"oss_zip_url":75,"oss_zip_packed_at":75,"status":17,"created_at":150,"updated_at":151,"faqs":152,"releases":182},7923,"mlflow\u002Fmlflow","mlflow","The open source AI engineering platform for agents, LLMs, and ML models. MLflow enables teams of all sizes to debug, evaluate, monitor, and optimize production-quality AI applications while controlling costs and managing access to models and data.","MLflow 是一个开源的 AI 工程平台，专为智能体（Agents）、大语言模型（LLMs）及传统机器学习模型的全生命周期管理而设计。它致力于解决 AI 应用从开发到生产落地过程中的核心痛点，帮助团队高效地调试代码、评估模型效果、监控运行状态并优化提示词，同时在确保生产级质量的前提下，有效控制成本并管理模型与数据的访问权限。\n\n无论是初创团队还是大型企业，MLflow 都是开发者、数据科学家和 AI 工程师的理想助手。它提供了生产级的可观测性追踪、自动化评估体系、提示词注册与管理以及统一的 AI 网关等独特功能。通过原生集成 OpenTelemetry 标准，MLflow 能够轻松打通不同技术栈，支持 Python、JavaScript 等多种语言。用户只需简单配置即可开启全链路追踪，在可视化界面中清晰洞察模型行为，从而自信地将高质量的 AI 应用部署至生产环境。作为全球下载量领先的开源项目，MLflow 正成为构建可靠 AI 系统的基础设施。","\u003Ch1 align=\"center\" style=\"border-bottom: none\">\n    \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002F\">\n        \u003Cimg alt=\"MLflow logo\" src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fassets\u002Flogo.svg\" width=\"200\" \u002F>\n    \u003C\u002Fa>\n\u003C\u002Fh1>\n\u003Ch2 align=\"center\" style=\"border-bottom: none\">The Open Source AI Engineering Platform for Agents, LLMs & Models\u003C\u002Fh2>\n\nMLflow is the largest open source **AI engineering platform for agents, LLMs, and ML models**. MLflow enables teams of all sizes to [debug](https:\u002F\u002Fmlflow.org\u002Fllm-tracing),\n[evaluate](https:\u002F\u002Fmlflow.org\u002Fllm-evaluation), [monitor](https:\u002F\u002Fmlflow.org\u002Fai-monitoring), and [optimize](https:\u002F\u002Fmlflow.org\u002Fprompt-optimization) production-quality AI applications while\ncontrolling costs and managing access to models and data. With over **60 million monthly downloads**,\nthousands of organizations rely on MLflow each day to ship AI to production with confidence.\n\nMLflow's comprehensive feature set for agents and LLM applications includes production-grade [observability](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing), [evaluation](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor),\n[prompt management](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fprompt-registry), [prompt optimization](https:\u002F\u002Fmlflow.org\u002Fprompt-optimization) and an [AI Gateway](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway) for managing costs and model access.\nLearn more at [MLflow for LLMs and Agents](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai).\n\n\u003Cdiv align=\"center\">\n\n[![Python SDK](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fmlflow)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fmlflow\u002F)\n[![PyPI Downloads](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fmlflow)](https:\u002F\u002Fpepy.tech\u002Fprojects\u002Fmlflow)\n[![License](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fmlflow\u002Fmlflow)](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fblob\u002Fmaster\u002FLICENSE.txt)\n\u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fintent\u002Ffollow?screen_name=mlflow\" target=\"_blank\">\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fmlflow?logo=X&color=%20%23f5f5f5\"\n      alt=\"follow on X(Twitter)\">\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fmlflow-org\u002F\" target=\"_blank\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_006e5e7b2fbb.png\"\n      alt=\"follow on LinkedIn\">\u003C\u002Fa>\n[![Ask DeepWiki](https:\u002F\u002Fdeepwiki.com\u002Fbadge.svg)](https:\u002F\u002Fdeepwiki.com\u002Fmlflow\u002Fmlflow)\n\n\u003C\u002Fdiv>\n\n\u003Cdiv align=\"center\">\n   \u003Cdiv>\n      \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002F\">\u003Cstrong>Website\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\">\u003Cstrong>Docs\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002Fnew\u002Fchoose\">\u003Cstrong>Feature Request\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fblog\">\u003Cstrong>News\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002F@mlflowoss\">\u003Cstrong>YouTube\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Flu.ma\u002Fmlflow?k=c\">\u003Cstrong>Events\u003C\u002Fstrong>\u003C\u002Fa>\n   \u003C\u002Fdiv>\n\u003C\u002Fdiv>\n\n\u003Cbr>\n\n## Get Started in 3 Simple Steps\n\nFrom zero to full-stack LLMOps in minutes. No complex setup or major code changes required. [Get Started →](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fquickstart\u002F)\n\n**1. Start MLflow Server**\n\n```bash\nuvx mlflow server\n```\n\n**2. Enable Logging**\n\n```python\nimport mlflow\n\nmlflow.set_tracking_uri(\"http:\u002F\u002Flocalhost:5000\")\nmlflow.openai.autolog()\n```\n\n**3. Run Your Code**\n\n```python\nfrom openai import OpenAI\n\nclient = OpenAI()\nclient.responses.create(\n    model=\"gpt-5.4-mini\",\n    input=\"Hello!\",\n)\n```\n\nExplore traces and metrics in the MLflow UI at `http:\u002F\u002Flocalhost:5000`.\n\n## LLMs & Agents\n\nMLflow provides everything you need to build, debug, evaluate, and deploy production-quality LLM applications and AI agents. Supports Python, TypeScript\u002FJavaScript, Java and any other programming language. MLflow also natively integrates with [OpenTelemetry](https:\u002F\u002Fopentelemetry.io\u002F) and MCP.\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd width=\"50%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_d37032c5caac.png\" alt=\"Observability\" width=100%>\n    \u003Cdiv align=\"center\">\n        \u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002F\">\u003Cstrong>Observability\u003C\u002Fstrong>\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n        \u003Cdiv>Capture complete traces of your LLM applications and agents for deep behavioral insights. Built on OpenTelemetry, supporting any LLM provider and agent framework. Monitor production quality, costs, and safety.\u003C\u002Fdiv>\u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fquickstart\u002F\">Getting Started →\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n    \u003C\u002Fdiv>\n    \u003C\u002Ftd>\n    \u003Ctd width=\"50%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_fff72771fc61.png\" alt=\"Evaluation\" width=100%>\n    \u003Cdiv align=\"center\">\n        \u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor\u002F\">\u003Cstrong>Evaluation\u003C\u002Fstrong>\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n        \u003Cdiv>Run systematic evaluations, track quality metrics over time, and catch regressions before they reach production. Choose from 50+ built-in metrics and LLM judges, or define your own.\u003C\u002Fdiv>\u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor\u002F\">Getting Started →\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n    \u003C\u002Fdiv>\n    \u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd width=\"50%\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_04ee5418037c.png\" alt=\"Prompts & Optimization\" width=100%>\n    \u003Cdiv align=\"center\">\n        \u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fprompt-registry\u002F\">\u003Cstrong>Prompts & Optimization\u003C\u002Fstrong>\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n        \u003Cdiv>Version, test, and deploy prompts with full lineage tracking. \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fprompt-optimization\">Automatically optimize prompts\u003C\u002Fa> with state-of-the-art algorithms to improve performance.\u003C\u002Fdiv>\u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fprompt-registry\u002Fcreate-and-edit-prompts\u002F\">Getting Started →\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n    \u003C\u002Fdiv>\n    \u003C\u002Ftd>\n    \u003Ctd width=\"50%\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_40190709a784.png\" alt=\"AI Gateway\" width=100%>\n    \u003Cdiv align=\"center\">\n        \u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway\u002F\">\u003Cstrong>AI Gateway\u003C\u002Fstrong>\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n        \u003Cdiv>Unified API gateway for all LLM providers. Route requests, manage rate limits, handle fallbacks, and control costs through an OpenAI-compatible interface with built-in credential management, guardrails and traffic splitting for A\u002FB testing.\u003C\u002Fdiv>\u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway\u002Fquickstart\u002F\">Getting Started →\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n    \u003C\u002Fdiv>\n    \u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## Model Training\n\nFor machine learning and deep learning model development, MLflow provides a full suite of tools to manage the ML lifecycle:\n\n- [**Experiment Tracking**](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Ftracking\u002F) — Track models, parameters, metrics, and evaluation results across experiments\n- [**Model Evaluation**](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Fevaluation\u002F) — Automated evaluation tools integrated with experiment tracking\n- [**Model Registry**](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Fmodel-registry\u002F) — Collaboratively manage the full lifecycle of ML models\n- [**Deployment**](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Fdeployment\u002F) — Deploy models to batch and real-time scoring on Docker, Kubernetes, Azure ML, AWS SageMaker, and more\n\nLearn more at [MLflow for Model Training](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml).\n\n## Integrations\n\nMLflow supports all agent frameworks, LLM providers, tools, and programming languages. We offer one-line automatic tracing for more than 60 frameworks. See the [full integrations list](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002F).\n\n### OpenTelemetry\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fapp-instrumentation\u002Fopentelemetry\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_ed02af64662b.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>OpenTelemetry\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Agent Frameworks (Python)\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flangchain\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_dabd8835aa5e.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LangChain\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flanggraph\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_cfe49592a4d8.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LangGraph\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopenai-agent\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_d775a43821c2.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>OpenAI Agent\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdspy\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_057f40e7b9ae.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>DSPy\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fpydantic_ai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6ebf96caa33c.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>PydanticAI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fgoogle-adk\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_038fa4fec239.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Google ADK\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fmicrosoft-agent-framework\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_8b59f6eeda5d.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Microsoft Agent\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fcrewai\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fcrewai-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>CrewAI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fllama_index\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fllamaindex-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LlamaIndex\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fautogen\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_34aced3a04ad.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>AutoGen\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fstrands\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_b14b5c74df61.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Strands\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flivekit\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_0ffadadb74dd.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LiveKit Agents\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fagno\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_976ec5a25492.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Agno\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fbedrock-agentcore\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_684780a81d3f.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Bedrock AgentCore\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fsmolagents\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_9ac1a9dfdd2f.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Smolagents\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fsemantic_kernel\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_c675d0715225.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Semantic Kernel\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdeepagent\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fdeepagent-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>DeepAgent\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fag2\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_792cfe304e33.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>AG2\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fhaystack\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_4056d06f2be5.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Haystack\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fkoog\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_551939ffa02d.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Koog\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Ftxtai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_384fa8391416.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>txtai\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fpipecat\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_00b49ba57957.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Pipecat\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fwatsonx-orchestrate\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_86fdb39cf225.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Watsonx\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Agent Frameworks (TypeScript)\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flangchain\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_dabd8835aa5e.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LangChain\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flanggraph\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_cfe49592a4d8.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LangGraph\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fvercelai\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fvercel-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Vercel AI SDK\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fmastra\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_a74088a77c27.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Mastra\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fvoltagent\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_9bc3e38c69b4.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>VoltAgent\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Agent Frameworks (Java)\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fspring-ai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_670fdc662f51.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Spring AI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fquarkus-langchain4j\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Flangchain4j.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Quarkus LangChain4j\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Model Providers\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopenai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_d775a43821c2.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>OpenAI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fanthropic\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_792d403805c2.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Anthropic\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdatabricks\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_11901941dee0.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Databricks\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fgemini\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fgoogle-gemini-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Gemini\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fbedrock\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_684780a81d3f.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Amazon Bedrock\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flitellm\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_e784ce600fd9.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LiteLLM\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fmistral\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fmistral-ai-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Mistral\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fxai-grok\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_416d77b18b1a.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>xAI \u002F Grok\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Follama\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_f3f3884b5345.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Ollama\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fgroq\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fgroq-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Groq\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdeepseek\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_f253add60d10.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>DeepSeek\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fqwen\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_f7635df985f6.jpg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Qwen\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fmoonshot\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_3d4cba369644.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Moonshot AI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fcohere\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_788761fdca66.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Cohere\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fbyteplus\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_a385494c550b.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>BytePlus\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fnovitaai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_bf1e5e11b3f3.jpg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Novita AI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Ffireworksai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_1e905961fc83.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>FireworksAI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Ftogetherai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_4143adc4663b.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Together AI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Gateways\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdatabricks-ai-gateway\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_11901941dee0.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Databricks\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flitellm-proxy\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_e784ce600fd9.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LiteLLM Proxy\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fvercel-ai-gateway\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fvercel-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Vercel AI Gateway\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopenrouter\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_7de0f71108f8.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>OpenRouter\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fportkey\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_63495ffeae87.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Portkey\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fhelicone\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_4b8871308961.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Helicone\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fkong\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_5a04c04f10cc.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Kong AI Gateway\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fpydantic-ai-gateway\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6ebf96caa33c.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>PydanticAI Gateway\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Ftruefoundry\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_ec510063050f.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>TrueFoundry\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### Tools & No-Code\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Finstructor\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Finstructor-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Instructor\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fclaude_code\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_3c923e45545e.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Claude Code\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopencode\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_898440acd2a5.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Opencode\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flangfuse\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_9fdc14d49f74.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Langfuse\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Farize\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_86ad4c621410.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Arize \u002F Phoenix\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fgoose\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_b95b3d9d1f98.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Goose\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flangflow\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Flangflow.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Langflow\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## Hosting MLflow\n\nMLflow can be used in a variety of environments, including your local environment, on-premises clusters, cloud platforms, and managed services. Being an open-source platform, MLflow is **vendor-neutral** — whether you're building AI agents, LLM applications, or ML models, you have access to MLflow's core capabilities.\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Fdocs.databricks.com\u002Faws\u002Fen\u002Fmlflow3\u002Fgenai\u002F\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_11901941dee0.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Databricks\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Faws.amazon.com\u002Fsagemaker-ai\u002Fexperiments\u002F\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_b1d25cfd629c.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Amazon SageMaker\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fmachine-learning\u002Fconcept-mlflow?view=azureml-api-2\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_53e22fb63b40.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Azure ML\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Fnebius.com\u002Fservices\u002Fmanaged-mlflow\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_cb559bad5685.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Nebius\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Ftracking\u002F\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6ff05ad98fd6.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Self-Hosted\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## 💭 Support\n\n- For help or questions about MLflow usage (e.g. \"how do I do X?\") visit the [documentation](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest).\n- In the documentation, you can ask the question to our AI-powered chat bot. Click on the **\"Ask AI\"** button at the right bottom.\n- Join the [virtual events](https:\u002F\u002Flu.ma\u002Fmlflow?k=c) like office hours and meetups.\n- To report a bug, file a documentation issue, or submit a feature request, please [open a GitHub issue](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002Fnew\u002Fchoose).\n- For release announcements and other discussions, please subscribe to our mailing list (mlflow-users@googlegroups.com)\n  or join us on [Slack](https:\u002F\u002Fmlflow.org\u002Fslack).\n\n## 🤝 Contributing\n\nWe happily welcome contributions to MLflow!\n\n- Submit [bug reports](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002Fnew?template=bug_report_template.yaml) and [feature requests](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002Fnew?template=feature_request_template.yaml)\n- Contribute for [good-first-issues](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22) and [help-wanted](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues?q=is%3Aissue+is%3Aopen+label%3A%22help+wanted%22)\n- Writing about MLflow and sharing your experience\n\nPlease see our [contribution guide](CONTRIBUTING.md) to learn more about contributing to MLflow.\n\n## ⭐️ Star History\n\n\u003Ca href=\"https:\u002F\u002Fstar-history.com\u002F#mlflow\u002Fmlflow&Date\">\n \u003Cpicture>\n   \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6e548023fa02.png&theme=dark\" \u002F>\n   \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6e548023fa02.png\" \u002F>\n   \u003Cimg alt=\"Star History Chart\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6e548023fa02.png\" \u002F>\n \u003C\u002Fpicture>\n\u003C\u002Fa>\n\n## ✏️ Citation\n\nIf you use MLflow in your research, please cite it using the \"Cite this repository\" button at the top of the [GitHub repository page](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow), which will provide you with citation formats including APA and BibTeX.\n\n## 👥 Core Members\n\nMLflow is currently maintained by the following core members with significant contributions from hundreds of exceptionally talented community members.\n\n- [Ben Wilson](https:\u002F\u002Fgithub.com\u002FBenWilson2)\n- [Corey Zumar](https:\u002F\u002Fgithub.com\u002Fdbczumar)\n- [Daniel Lok](https:\u002F\u002Fgithub.com\u002Fdaniellok-db)\n- [Gabriel Fu](https:\u002F\u002Fgithub.com\u002Fgabrielfu)\n- [Harutaka Kawamura](https:\u002F\u002Fgithub.com\u002Fharupy)\n- [Joel Robin P](https:\u002F\u002Fgithub.com\u002Fjoelrobin18)\n- [Matt Prahl](https:\u002F\u002Fgithub.com\u002Fmprahl)\n- [Pat Sukprasert](https:\u002F\u002Fgithub.com\u002FPattaraS)\n- [Serena Ruan](https:\u002F\u002Fgithub.com\u002Fserena-ruan)\n- [Tomu Hirata](https:\u002F\u002Fgithub.com\u002FTomeHirata)\n- [Weichen Xu](https:\u002F\u002Fgithub.com\u002FWeichenXu123)\n- [Yuki Watanabe](https:\u002F\u002Fgithub.com\u002FB-Step62)\n","\u003Ch1 align=\"center\" style=\"border-bottom: none\">\n    \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002F\">\n        \u003Cimg alt=\"MLflow logo\" src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fassets\u002Flogo.svg\" width=\"200\" \u002F>\n    \u003C\u002Fa>\n\u003C\u002Fh1>\n\u003Ch2 align=\"center\" style=\"border-bottom: none\">面向智能体、大语言模型及各类模型的开源AI工程平台\u003C\u002Fh2>\n\nMLflow是目前最大的面向**智能体、大语言模型和机器学习模型**的开源AI工程平台。它帮助各种规模的团队在控制成本、管理模型与数据访问权限的同时，对生产级AI应用进行[调试](https:\u002F\u002Fmlflow.org\u002Fllm-tracing)、[评估](https:\u002F\u002Fmlflow.org\u002Fllm-evaluation)、[监控](https:\u002F\u002Fmlflow.org\u002Fai-monitoring)以及[优化](https:\u002F\u002Fmlflow.org\u002Fprompt-optimization)。凭借每月超过**6000万次下载**，成千上万的企业每天都在使用MLflow，自信地将AI部署到生产环境中。\n\nMLflow为智能体和大语言模型应用提供的全面功能包括：生产级别的[可观测性](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing)、[评估](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor)、[提示词管理](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fprompt-registry)、[提示词优化](https:\u002F\u002Fmlflow.org\u002Fprompt-optimization)，以及用于管理成本和模型访问权限的[AI网关](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway)。更多信息请访问[MLflow for LLMs and Agents](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai)。\n\n\u003Cdiv align=\"center\">\n\n[![Python SDK](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fmlflow)](https:\u002F\u002Fpypi.org\u002Fproject\u002Fmlflow\u002F)\n[![PyPI Downloads](https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fdm\u002Fmlflow)](https:\u002F\u002Fpepy.tech\u002Fprojects\u002Fmlflow)\n[![License](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fmlflow\u002Fmlflow)](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fblob\u002Fmaster\u002FLICENSE.txt)\n\u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fintent\u002Ffollow?screen_name=mlflow\" target=\"_blank\">\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Ftwitter\u002Ffollow\u002Fmlflow?logo=X&color=%20%23f5f5f5\"\n      alt=\"关注X(Twitter)\">\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fwww.linkedin.com\u002Fcompany\u002Fmlflow-org\u002F\" target=\"_blank\">\n\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_006e5e7b2fbb.png\"\n      alt=\"关注LinkedIn\">\u003C\u002Fa>\n[![Ask DeepWiki](https:\u002F\u002Fdeepwiki.com\u002Fbadge.svg)](https:\u002F\u002Fdeepwiki.com\u002Fmlflow\u002Fmlflow)\n\n\u003C\u002Fdiv>\n\n\u003Cdiv align=\"center\">\n   \u003Cdiv>\n      \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002F\">\u003Cstrong>官网\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\">\u003Cstrong>文档\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002Fnew\u002Fchoose\">\u003Cstrong>功能需求\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fblog\">\u003Cstrong>新闻\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Fwww.youtube.com\u002F@mlflowoss\">\u003Cstrong>YouTube\u003C\u002Fstrong>\u003C\u002Fa> ·\n      \u003Ca href=\"https:\u002F\u002Flu.ma\u002Fmlflow?k=c\">\u003Cstrong>活动\u003C\u002Fstrong>\u003C\u002Fa>\n   \u003C\u002Fdiv>\n\u003C\u002Fdiv>\n\n\u003Cbr>\n\n## 三步快速入门\n\n几分钟内即可从零开始搭建全栈LLMOps环境。无需复杂配置或大幅修改代码。[立即开始 →](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fquickstart\u002F)\n\n**1. 启动MLflow服务器**\n\n```bash\nuvx mlflow server\n```\n\n**2. 开启日志记录**\n\n```python\nimport mlflow\n\nmlflow.set_tracking_uri(\"http:\u002F\u002Flocalhost:5000\")\nmlflow.openai.autolog()\n```\n\n**3. 运行你的代码**\n\n```python\nfrom openai import OpenAI\n\nclient = OpenAI()\nclient.responses.create(\n    model=\"gpt-5.4-mini\",\n    input=\"Hello!\",\n)\n```\n\n你可以在`http:\u002F\u002Flocalhost:5000`的MLflow UI中查看追踪记录和指标。\n\n## 大语言模型与智能体\n\nMLflow提供了构建、调试、评估和部署生产级大语言模型应用及AI智能体所需的一切工具。支持Python、TypeScript\u002FJavaScript、Java以及其他任何编程语言。MLflow还原生集成OpenTelemetry和MCP。\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd width=\"50%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_d37032c5caac.png\" alt=\"可观测性\" width=100%>\n    \u003Cdiv align=\"center\">\n        \u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002F\">\u003Cstrong>可观测性\u003C\u002Fstrong>\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n        \u003Cdiv>捕获大语言模型应用和智能体的完整追踪信息，以获得深入的行为洞察。基于OpenTelemetry构建，支持任何大语言模型提供商和智能体框架。监控生产质量、成本和安全性。\u003C\u002Fdiv>\u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fquickstart\u002F\">快速入门 →\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n    \u003C\u002Fdiv>\n    \u003C\u002Ftd>\n    \u003Ctd width=\"50%\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_fff72771fc61.png\" alt=\"评估\" width=100%>\n    \u003Cdiv align=\"center\">\n        \u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor\u002F\">\u003Cstrong>评估\u003C\u002Fstrong>\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n        \u003Cdiv>执行系统化的评估，长期跟踪质量指标，并在问题影响生产之前及时发现回归。可选择50多种内置指标和大语言模型评判者，也可自定义评估标准。\u003C\u002Fdiv>\u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor\u002F\">快速入门 →\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n    \u003C\u002Fdiv>\n    \u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd width=\"50%\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_04ee5418037c.png\" alt=\"提示词与优化\" width=100%>\n    \u003Cdiv align=\"center\">\n        \u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fprompt-registry\u002F\">\u003Cstrong>提示词与优化\u003C\u002Fstrong>\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n        \u003Cdiv>对提示词进行版本管理、测试和部署，实现完整的溯源追踪。还可通过最先进的算法\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fprompt-optimization\">自动优化提示词\u003C\u002Fa>,以提升性能。\u003C\u002Fdiv>\u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fprompt-registry\u002Fcreate-and-edit-prompts\u002F\">快速入门 →\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n    \u003C\u002Fdiv>\n    \u003C\u002Ftd>\n    \u003Ctd width=\"50%\">\n      \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_40190709a784.png\" alt=\"AI网关\" width=100%>\n    \u003Cdiv align=\"center\">\n        \u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway\u002F\">\u003Cstrong>AI网关\u003C\u002Fstrong>\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n        \u003Cdiv>面向所有大语言模型提供商的统一API网关。通过兼容OpenAI的接口路由请求、管理速率限制、处理回退逻辑，并借助内置的凭据管理、安全约束和流量拆分功能来控制成本和进行A\u002FB测试。\u003C\u002Fdiv>\u003Cbr>\n        \u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway\u002Fquickstart\u002F\">快速入门 →\u003C\u002Fa>\n        \u003Cbr>\u003Cbr>\n    \u003C\u002Fdiv>\n    \u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## 模型训练\n\n对于机器学习和深度学习模型的开发，MLflow 提供了一整套工具来管理机器学习生命周期：\n\n- [**实验跟踪**](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Ftracking\u002F) — 跟踪跨不同实验的模型、参数、指标和评估结果\n- [**模型评估**](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Fevaluation\u002F) — 与实验跟踪集成的自动化评估工具\n- [**模型注册表**](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Fmodel-registry\u002F) — 协作式地管理机器学习模型的全生命周期\n- [**部署**](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Fdeployment\u002F) — 将模型部署到批处理和实时打分环境中，支持 Docker、Kubernetes、Azure ML、AWS SageMaker 等平台\n\n更多信息请访问 [MLflow 模型训练指南](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml)。\n\n## 集成\n\nMLflow 支持所有代理框架、大语言模型提供商、工具和编程语言。我们为超过 60 种框架提供一行代码的自动追踪功能。完整的集成列表请参见 [集成列表](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002F)。\n\n### OpenTelemetry\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fapp-instrumentation\u002Fopentelemetry\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_ed02af64662b.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>OpenTelemetry\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### 代理框架（Python）\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flangchain\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_dabd8835aa5e.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LangChain\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flanggraph\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_cfe49592a4d8.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LangGraph\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopenai-agent\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_d775a43821c2.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>OpenAI Agent\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdspy\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_057f40e7b9ae.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>DSPy\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fpydantic_ai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6ebf96caa33c.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>PydanticAI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fgoogle-adk\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_038fa4fec239.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Google ADK\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fmicrosoft-agent-framework\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_8b59f6eeda5d.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Microsoft Agent\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fcrewai\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fcrewai-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>CrewAI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fllama_index\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fllamaindex-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LlamaIndex\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fautogen\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_34aced3a04ad.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>AutoGen\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fstrands\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_b14b5c74df61.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Strands\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flivekit\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_0ffadadb74dd.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LiveKit Agents\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fagno\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_976ec5a25492.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Agno\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fbedrock-agentcore\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_684780a81d3f.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Bedrock AgentCore\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fsmolagents\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_9ac1a9dfdd2f.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Smolagents\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fsemantic_kernel\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_c675d0715225.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Semantic Kernel\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdeepagent\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fdeepagent-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>DeepAgent\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fag2\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_792cfe304e33.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>AG2\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fhaystack\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_4056d06f2be5.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Haystack\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fkoog\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_551939ffa02d.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Koog\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Ftxtai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_384fa8391416.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>txtai\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fpipecat\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_00b49ba57957.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Pipecat\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fwatsonx-orchestrate\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_86fdb39cf225.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Watsonx\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### 代理框架（TypeScript）\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flangchain\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_dabd8835aa5e.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LangChain\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flanggraph\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_cfe49592a4d8.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LangGraph\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fvercelai\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fvercel-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Vercel AI SDK\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fmastra\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_a74088a77c27.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Mastra\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fvoltagent\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_9bc3e38c69b4.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>VoltAgent\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### 代理框架（Java）\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fspring-ai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_670fdc662f51.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Spring AI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fquarkus-langchain4j\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Flangchain4j.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Quarkus LangChain4j\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### 模型提供商\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopenai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_d775a43821c2.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>OpenAI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fanthropic\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_792d403805c2.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Anthropic\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdatabricks\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_11901941dee0.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Databricks\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fgemini\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fgoogle-gemini-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Gemini\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fbedrock\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_684780a81d3f.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Amazon Bedrock\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flitellm\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_e784ce600fd9.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LiteLLM\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fmistral\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fmistral-ai-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Mistral\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fxai-grok\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_416d77b18b1a.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>xAI \u002F Grok\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Follama\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_f3f3884b5345.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Ollama\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fgroq\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fgroq-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Groq\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdeepseek\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_f253add60d10.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>DeepSeek\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fqwen\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_f7635df985f6.jpg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Qwen\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fmoonshot\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_3d4cba369644.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Moonshot AI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fcohere\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_788761fdca66.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Cohere\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fbyteplus\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_a385494c550b.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>BytePlus\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fnovitaai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_bf1e5e11b3f3.jpg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Novita AI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Ffireworksai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_1e905961fc83.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>FireworksAI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Ftogetherai\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_4143adc4663b.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Together AI\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### 网关\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fdatabricks-ai-gateway\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_11901941dee0.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Databricks\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flitellm-proxy\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_e784ce600fd9.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>LiteLLM Proxy\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fvercel-ai-gateway\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Fvercel-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Vercel AI Gateway\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopenrouter\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_7de0f71108f8.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>OpenRouter\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fportkey\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_63495ffeae87.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Portkey\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fhelicone\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_4b8871308961.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Helicone\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fkong\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_5a04c04f10cc.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Kong AI Gateway\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fpydantic-ai-gateway\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6ebf96caa33c.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>PydanticAI Gateway\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Ftruefoundry\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_ec510063050f.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>TrueFoundry\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n### 工具与无代码平台\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Finstructor\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Finstructor-logo.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Instructor\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fclaude_code\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_3c923e45545e.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Claude Code\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopencode\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_898440acd2a5.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Opencode\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flangfuse\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_9fdc14d49f74.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Langfuse\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Farize\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_86ad4c621410.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Arize \u002F Phoenix\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fgoose\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_b95b3d9d1f98.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Goose\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"110\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Flangflow\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fmlflow\u002Fmlflow\u002Frefs\u002Fheads\u002Fmaster\u002Fdocs\u002Fstatic\u002Fimages\u002Flogos\u002Flangflow.svg\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Langflow\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## 托管 MLflow\n\nMLflow 可以在多种环境中使用，包括您的本地环境、本地集群、云平台以及托管服务。作为一款开源平台，MLflow 具有**供应商中立性**——无论您是在构建 AI 代理、大模型应用，还是机器学习模型，都可以访问 MLflow 的核心功能。\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Fdocs.databricks.com\u002Faws\u002Fen\u002Fmlflow3\u002Fgenai\u002F\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_11901941dee0.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Databricks\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Faws.amazon.com\u002Fsagemaker-ai\u002Fexperiments\u002F\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_b1d25cfd629c.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Amazon SageMaker\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fmachine-learning\u002Fconcept-mlflow?view=azureml-api-2\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_53e22fb63b40.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Azure ML\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Fnebius.com\u002Fservices\u002Fmanaged-mlflow\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_cb559bad5685.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>Nebius\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n    \u003Ctd align=\"center\" width=\"130\">\u003Ca href=\"https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Ftracking\u002F\">\u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6ff05ad98fd6.png\" height=\"40\">\u003Cbr>\u003Csub>\u003Cb>自托管\u003C\u002Fb>\u003C\u002Fsub>\u003C\u002Fa>\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n## 💭 支持\n\n- 如需帮助或有关 MLflow 使用的问题（例如“如何实现X功能？”），请访问 [文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest)。\n- 在文档中，您可以通过我们的AI驱动聊天机器人提问。请点击右下角的 **“Ask AI”** 按钮。\n- 参加我们的[线上活动](https:\u002F\u002Flu.ma\u002Fmlflow?k=c)，如答疑时间与聚会。\n- 如需报告Bug、提交文档问题或功能请求，请在 [GitHub上新建议题](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002Fnew\u002Fchoose)。\n- 有关版本发布和其他讨论，请订阅我们的邮件列表（mlflow-users@googlegroups.com）或加入我们的 [Slack](https:\u002F\u002Fmlflow.org\u002Fslack)。\n\n## 🤝 贡献\n\n我们非常欢迎对MLflow的贡献！\n\n- 提交 [Bug报告](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002Fnew?template=bug_report_template.yaml) 和 [功能请求](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002Fnew?template=feature_request_template.yaml)\n- 参与 [初学者友好型议题](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22) 和 [寻求帮助的议题](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues?q=is%3Aissue+is%3Aopen+label%3A%22help+wanted%22)\n- 撰写关于MLflow的文章并分享您的使用经验\n\n请参阅我们的 [贡献指南](CONTRIBUTING.md)，了解更多关于如何为MLflow做出贡献的信息。\n\n## ⭐️ 星标历史\n\n\u003Ca href=\"https:\u002F\u002Fstar-history.com\u002F#mlflow\u002Fmlflow&Date\">\n \u003Cpicture>\n   \u003Csource media=\"(prefers-color-scheme: dark)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6e548023fa02.png&theme=dark\" \u002F>\n   \u003Csource media=\"(prefers-color-scheme: light)\" srcset=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6e548023fa02.png\" \u002F>\n   \u003Cimg alt=\"星标历史图表\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_readme_6e548023fa02.png\" \u002F>\n \u003C\u002Fpicture>\n\u003C\u002Fa>\n\n## ✏️ 引用\n\n如果您在研究中使用了MLflow，请使用 [GitHub仓库页面](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow)顶部的“引用此仓库”按钮进行引用，该按钮将为您提供包括APA和BibTeX在内的多种引用格式。\n\n## 👥 核心成员\n\n目前，MLflow由以下核心成员维护，并得到了数百位才华横溢的社区成员的重要贡献。\n\n- [Ben Wilson](https:\u002F\u002Fgithub.com\u002FBenWilson2)\n- [Corey Zumar](https:\u002F\u002Fgithub.com\u002Fdbczumar)\n- [Daniel Lok](https:\u002F\u002Fgithub.com\u002Fdaniellok-db)\n- [Gabriel Fu](https:\u002F\u002Fgithub.com\u002Fgabrielfu)\n- [Harutaka Kawamura](https:\u002F\u002Fgithub.com\u002Fharupy)\n- [Joel Robin P](https:\u002F\u002Fgithub.com\u002Fjoelrobin18)\n- [Matt Prahl](https:\u002F\u002Fgithub.com\u002Fmprahl)\n- [Pat Sukprasert](https:\u002F\u002Fgithub.com\u002FPattaraS)\n- [Serena Ruan](https:\u002F\u002Fgithub.com\u002Fserena-ruan)\n- [Tomu Hirata](https:\u002F\u002Fgithub.com\u002FTomeHirata)\n- [Weichen Xu](https:\u002F\u002Fgithub.com\u002FWeichenXu123)\n- [Yuki Watanabe](https:\u002F\u002Fgithub.com\u002FB-Step62)","# MLflow 快速上手指南\n\nMLflow 是一个开源的 AI 工程平台，专为智能体（Agents）、大语言模型（LLMs）及传统机器学习模型设计。它帮助团队调试、评估、监控和优化生产级 AI 应用，同时有效控制成本和管理模型访问权限。\n\n## 环境准备\n\n在开始之前，请确保您的开发环境满足以下要求：\n\n*   **操作系统**：Linux、macOS 或 Windows。\n*   **Python 版本**：推荐 Python 3.9 或更高版本。\n*   **前置依赖**：\n    *   已安装 `pip` 或 `uv`（推荐使用 `uv` 以获得更快的安装和运行体验）。\n    *   若需测试 LLM 功能，请准备好 OpenAI API Key 或其他大模型服务商的凭证。\n\n## 安装步骤\n\n您可以选择使用标准的 `pip` 安装，或使用更现代的 `uv` 工具直接运行。\n\n### 方式一：使用 pip 安装（推荐国内用户配置镜像源）\n\n为了加速下载，建议使用国内镜像源（如清华大学源）：\n\n```bash\npip install mlflow -i https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple\n```\n\n### 方式二：使用 uv 直接运行（无需预先安装）\n\n如果您已安装 `uv`，可以直接通过以下命令启动服务，无需全局安装：\n\n```bash\nuvx mlflow server\n```\n\n## 基本使用\n\n只需三个简单步骤，即可从零基础搭建全套 LLMOps 流程。\n\n### 1. 启动 MLflow 服务器\n\n在终端中运行以下命令启动本地追踪服务器：\n\n```bash\nuvx mlflow server\n# 或者如果您已通过 pip 安装：\n# mlflow server\n```\n服务器默认将在 `http:\u002F\u002Flocalhost:5000` 运行。\n\n### 2. 启用自动日志记录\n\n在您的 Python 代码中导入 MLflow，设置追踪 URI 并开启 OpenAI 自动日志功能：\n\n```python\nimport mlflow\n\nmlflow.set_tracking_uri(\"http:\u002F\u002Flocalhost:5000\")\nmlflow.openai.autolog()\n```\n\n### 3. 运行您的 AI 代码\n\n正常编写和运行您的 LLM 调用代码。以下以 OpenAI 为例：\n\n```python\nfrom openai import OpenAI\n\nclient = OpenAI()\nclient.responses.create(\n    model=\"gpt-5.4-mini\",\n    input=\"Hello!\",\n)\n```\n\n### 查看结果\n\n代码运行完成后，打开浏览器访问 **`http:\u002F\u002Flocalhost:5000`**。\n您可以在 MLflow UI 中查看完整的调用链路（Traces）、性能指标以及模型交互详情，无需修改任何业务逻辑代码。","某电商公司的算法团队正在开发一款基于大语言模型的智能客服助手，需要快速迭代提示词并监控线上表现。\n\n### 没有 mlflow 时\n- **调试如“盲人摸象”**：当用户反馈回答错误时，开发人员无法还原完整的请求链路，难以定位是提示词问题、模型波动还是代码逻辑错误。\n- **评估全靠“人工抽检”**：缺乏自动化评估框架，每次更新提示词后只能靠人工逐条检查回复质量，效率低下且标准不一。\n- **成本与权限“一笔糊涂账”**：不同部门随意调用高价模型导致账单激增，且无法精细控制谁可以访问哪些敏感数据或模型。\n- **实验记录“散落各处”**：尝试过的提示词版本和参数配置散落在笔记或聊天记录中，无法复现最佳效果，团队协作困难。\n\n### 使用 mlflow 后\n- **全链路追踪“一目了然”**：通过 MLflow 的追踪功能，开发人员可瞬间查看每次交互的完整输入输出及中间步骤，快速锁定故障根源。\n- **自动化评估“量化效果”**：利用内置的评估工具，团队能自动对新提示词进行批量测试，用准确率、延迟等指标客观对比版本优劣。\n- **统一网关“管控成本”**：借助 AI Gateway 集中管理模型访问权限，设置配额限制，既保障了数据安全，又有效控制了 API 调用成本。\n- **提示词注册“版本清晰”**：所有实验过的提示词和参数都被自动记录在提示词注册表中，支持一键回滚和复用，极大提升了协作效率。\n\nMLflow 将原本混乱的黑盒开发过程转变为可观测、可评估、可管控的工程化流程，让团队能自信地将 AI 应用推向生产环境。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fmlflow_mlflow_d37032c5.png","MLflow","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fmlflow_e4b8f76f.png","The open source AI engineering platform for agents, LLMs, and ML models.",null,"https:\u002F\u002Fwww.mlflow.org","https:\u002F\u002Fgithub.com\u002Fmlflow",[79,83,87,91,95,98,102,105,109,112],{"name":80,"color":81,"percentage":82},"Python","#3572A5",61.2,{"name":84,"color":85,"percentage":86},"TypeScript","#3178c6",28.6,{"name":88,"color":89,"percentage":90},"JavaScript","#f1e05a",8.5,{"name":92,"color":93,"percentage":94},"Java","#b07219",0.6,{"name":96,"color":97,"percentage":94},"R","#198CE7",{"name":99,"color":100,"percentage":101},"CSS","#663399",0.2,{"name":103,"color":104,"percentage":101},"HTML","#e34c26",{"name":106,"color":107,"percentage":108},"Shell","#89e051",0.1,{"name":110,"color":111,"percentage":108},"Scala","#c22d40",{"name":113,"color":114,"percentage":115},"Mako","#7e858d",0,25383,5577,"2026-04-15T16:15:50","Apache-2.0","未说明",{"notes":122,"python":120,"dependencies":123},"MLflow 是一个支持多语言（Python, TypeScript\u002FJavaScript, Java 等）的 AI 工程平台。文中示例使用 'uvx' 命令启动服务器，暗示推荐使用 uv 包管理器。它原生集成 OpenTelemetry，并支持超过 60 个 Agent 框架和 LLM 提供商的自动追踪。具体版本依赖需参考官方文档或 PyPI 页面，README 中未列出具体版本号。",[124,125,126,127,128,129,130,131],"openai","opentelemetry","langchain","langgraph","dspy","llama_index","autogen","crewai",[133,15,13,14,35],"其他",[135,136,137,64,138,139,140,141,142,126,143,144,145,146,124,147,148,149],"machine-learning","ai","ml","apache-spark","model-management","agentops","agents","evaluation","llm-evaluation","llmops","observability","open-source","prompt-engineering","ai-governance","mlops","2026-03-27T02:49:30.150509","2026-04-16T08:14:13.088183",[153,158,163,168,173,177],{"id":154,"question_zh":155,"answer_zh":156,"source_url":157},35466,"如何在 MLflow 的指标图表中启用对数刻度（Log Scale）以便更好地观察损失变化？","您可以在图表设置中启用 Y 轴的对数刻度。具体操作是进入图表的设置选项，找到 Y 轴相关配置并开启对数模式。这有助于在损失值跨度较大时更清晰地判断模型是否仍在改进。","https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002F6348",{"id":159,"question_zh":160,"answer_zh":161,"source_url":162},35467,"当本地运行脚本上传 artifacts 到 S3 时报错'NoCredentialsError'，而远程服务器正常，如何解决？","这是因为本地运行时尝试使用本地凭证而非跟踪服务器的凭证。解决方案是确保在运行 MLflow Server 的 Pod 或环境中提供了有效的 AWS 凭证。如果问题依旧，请检查本地环境是否正确配置了代理或凭证传递机制，或者考虑在本地显式配置 AWS 凭证。","https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002F6181",{"id":164,"question_zh":165,"answer_zh":166,"source_url":167},35468,"在 UI 中选择 S3 artifact 时出现'IsADirectoryError'内部服务器错误怎么办？","该错误通常发生在 MLflow 服务器尝试将目录当作文件打开时。常见于使用 MinIO 作为 S3 后端且路径处理不当的情况。建议检查 MLflow 版本是否已知存在此 bug（如 1.9.0），并升级到最新版本。同时确认 S3 存储桶中的对象路径结构正确，避免空目录被误认为文件。","https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002F3154",{"id":169,"question_zh":170,"answer_zh":171,"source_url":172},35469,"修改追踪 URI 后本地 MLflow UI 无法加载显示'Something went wrong'，但程序访问正常，如何修复？","此问题通常由追踪 URI 配置冲突引起。建议在修改追踪 URI 前先停止本地 MLflow 服务器。若已发生，可尝试以下步骤：1) 停止当前服务器；2) 清除浏览器缓存；3) 确认环境变量 MLFLOW_TRACKING_URI 指向正确的本地地址（如 http:\u002F\u002Flocalhost:5000）；4) 重新启动服务器并使用原始命令：mlflow server --backend-store-uri sqlite:\u002F\u002F\u002Fmlflow.db --default-artifact-root .\u002Fartifacts --host 0.0.0.0。","https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002F9843",{"id":174,"question_zh":175,"answer_zh":176,"source_url":157},35470,"如何让 MLflow 实验跟踪界面的指标表格更整洁，减少小数位数？","目前 MLflow UI 尚未提供直接设置小数位数的功能，但社区已提出该需求。临时解决方案是在记录指标前在代码中对数值进行四舍五入处理，例如使用 round(value, 2) 后再调用 mlflow.log_metric()。维护者已将列拖拽和格式化功能列入待办事项。",{"id":178,"question_zh":179,"answer_zh":180,"source_url":181},35471,"是否支持通过跟踪服务代理上传 artifacts，以避免客户端直接访问 S3？","这是社区提出的功能请求，旨在让客户端无需配置 S3 权限即可上传 artifacts。目前 MLflow 原生不支持该模式，但可通过自定义实现或在 CI\u002FCD 流程中引入中间服务来模拟。维护者已将其列为路线图项目，未来可能增加服务端代理上传功能。","https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002F629",[183,188,193,198,203,208,213,218,223,228,233,238,243,248,253,258,263,268,273,278],{"id":184,"version":185,"summary_zh":186,"released_at":187},280580,"ts\u002Fv0.2.0-rc.1","`@mlflow\u002Fvercel` TypeScript 包 0.2.0 版本的候选发布版：https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fpull\u002F22105","2026-04-13T09:42:04",{"id":189,"version":190,"summary_zh":191,"released_at":192},280581,"v3.11.1","MLflow 3.11.1 包含多项重大功能和改进。\n\n**主要新特性**：\n\n- 🔍 **自动问题识别**：借助 AI 自动识别代理中的质量问题！使用跟踪表中的“检测问题”按钮，分析选定的跟踪记录，并按正确性、安全性、性能等类别揭示潜在问题。这些问题会直接链接到相应的跟踪记录，便于快速调查和调试。[文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor\u002Fai-insights\u002Fdetect-issues) (#21431, #21204, #21165, #21163, #21161, @smoorjani, @serena-ruan)\n- 💰 **网关预算告警与限制**：通过可配置的预算策略控制 AI 网关的支出！按时间窗口（每日、每周或每月）设置支出上限，在达到限额前接收告警，并通过自动阻止请求来防止成本失控。全新的预算管理界面允许您跟踪支出、配置 Webhook 通知，并监控所有网关端点的违规情况。[文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway\u002Fbudget-alerts-limits) (#21116, #21534, #21569, #21473, #21108, @TomeHirata, @copilot-swe-agent)\n- 📊 **跟踪图视图**：使用交互式图视图可视化复杂的跟踪层次结构！轻松导航多级跟踪结构，一目了然地理解父子关系，通过跟踪拓扑的可视化表示更高效地调试复杂系统。[文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fobserve-with-traces\u002Fui) (#20607, @joelrobin18)\n- 🌐 **原生支持 OpenTelemetry GenAI 约定**：MLflow 现在原生支持 OpenTelemetry GenAI 语义约定用于跟踪导出！当启用 `MLFLOW_ENABLE_OTEL_GENAI_SEMCONV` 并通过 OTLP 导出跟踪时，MLflow 会自动将其转换为符合 OTel GenAI 语义约定的格式，从而实现与兼容 OTel 的可观测性平台的无缝集成，同时保留 GenAI 特有的元数据。[文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fopentelemetry\u002Fgenai-semconv) (#21494, #21495, @B-Step62)\n- 🔧 **OpenCode 跟踪集成**：通过与 OpenCode CLI 集成，实现更智能的调试！直接从开发工作流中跟踪和分析代码执行流程，从而更容易识别性能瓶颈，并将问题追溯到具体的代码路径。[文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fopencode) (#20133, @joelrobin18)\n- ⚡ **原生支持 UV 模型依赖管理**：自动依赖推断现支持 UV！MLflow 会自动检测 UV 项目，并在记录模型时从锁定文件中捕获精确且已锁定的依赖项，以确保环境的可重复性。[文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Fmodel\u002Fdependencies) (#20344, #20935, @debu-sinha)\n- 🔒 **无 Pickle 模型序列化**：采用无 Pickle 的模型格式提升安全性！MLflow 现在支持使用 `torch.export` 和 `skops` 格式进行更安全的模型序列化，具有改迥\n","2026-04-08T05:29:25",{"id":194,"version":195,"summary_zh":196,"released_at":197},280582,"model-catalog\u002Flatest","每个提供商的模型目录文件。由持续集成系统每周更新。","2026-04-06T05:49:14",{"id":199,"version":200,"summary_zh":201,"released_at":202},280583,"v3.11.0rc1","从评估和 AI 网关功能中移除了第三方依赖，用内置实现替代了外部提供商的路由机制。","2026-04-01T09:14:13",{"id":204,"version":205,"summary_zh":206,"released_at":207},280584,"v3.11.0rc0","我们很高兴地宣布 MLflow 3.11.0rc0 的发布，其中包含多项重要更新：\n\n**重大新特性**：\n\n- 🔍 **自动问题识别**：借助 AI 自动识别代理中的质量问题！使用跟踪表中的“检测问题”按钮，分析选定的跟踪记录，并按正确性、安全性、性能等类别揭示潜在问题。这些问题会直接链接到相应的跟踪记录，便于快速调查和调试。（#21431、#21204、#21165、#21163、#21161、@smoorjani、@serena-ruan）\n- 💰 **网关预算告警与限制**：通过可配置的预算策略控制您的 AI 网关支出！您可以按时间窗口（每日、每周或每月）设置支出上限，在达到限额前接收告警，并通过自动阻止请求来防止成本失控。全新的预算管理界面使您能够跟踪支出、配置通知 Webhook，并监控所有网关端点的违规情况。（#21116、#21534、#21569、#21473、#21108、@TomeHirata、@copilot-swe-agent）\n- 📊 **跟踪图视图**：通过交互式图视图可视化复杂的跟踪层次结构！轻松导航多层级的跟踪结构，一目了然地理解父子关系，借助跟踪拓扑的可视化表示更高效地调试复杂系统。（#20607、@joelrobin18）\n- 🌐 **原生支持 OpenTelemetry GenAI 约定**：MLflow 现在原生支持 OpenTelemetry GenAI 语义约定用于跟踪导出！当启用 `MLFLOW_ENABLE_OTEL_GENAI_SEMCONV` 并通过 OTLP 导出跟踪时，MLflow 会自动将其转换为符合 OTel GenAI 语义约定的格式，从而实现与兼容 OTel 的可观测性平台的无缝集成，同时保留 GenAI 特有的元数据。（#21494、#21495、@B-Step62）\n- 🔧 **Opencode 跟踪集成**：通过与 Opencode CLI 集成，实现更智能的调试！直接从开发工作流中跟踪和分析代码执行流程，更容易发现性能瓶颈，并将问题追溯到具体的代码路径。（#20133、@joelrobin18）\n- ⚡ **UV 包管理器支持**：自动依赖项推断现已支持 UV！MLflow 会自动检测 UV 项目，并在记录模型时从锁定文件中捕获精确且已锁定的依赖项，从而确保环境的可重复性。（#20344、#20935、@debu-sinha）\n- 🔒 **无 Pickle 模型序列化**：采用无 Pickle 模型格式提升安全性！MLflow 现在支持使用 torch.export 和 skops 格式进行更安全的模型序列化，并在 `MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False` 时提供更严格的控制。全面的文档将指导您如何将现有模型迁移到无 Pickle 格式，以用于生产部署。（#21404、#21188、#20774、@WeichenXu123）\n\n**破坏性变更**：\n\n- ⚠️ **TypeScript SDK 包重命名**：MLflow TypeScript SDK 的包已更名为使用 npm 组织作用域。如果您正在使用 TypeScript SDK，请更新您的 `package.json` 中的依赖项和导入语句。","2026-03-16T14:02:42",{"id":209,"version":210,"summary_zh":211,"released_at":212},280585,"v3.10.1","MLflow 3.10.1 是一个补丁版本，包含一些小的功能增强、错误修复和文档更新。\n\n功能：\n\n- [UI] 在网关使用示例模态框中添加“试用”页面 (#21077, @PattaraS)\n- [UI] 在实验列表页面中过滤网关实验 (#21130, @copilot-swe-agent)\n\n错误修复：\n\n- [UI] 修复启用工作区时网关使用选项卡中的“查看完整仪表板”链接 (#21191, @copilot-swe-agent)\n- [UI] 将 AI 网关默认密码安全提示的关闭状态持久化到 localStorage 中 (#21292, @copilot-swe-agent)\n- [评估] 将指令评判器中未使用的参数日志消息的级别从 WARNING 降为 DEBUG (#21294, @copilot-swe-agent)\n- [UI] 切换到概览选项卡时清除“全部”时间选择器 (#21371, @daniellok-db)\n- [提示 \u002F UI] 修复提示选项卡中的追踪视图无法滚动的问题 (#21282, @TomeHirata)\n- [UI] 修复评判器构建器的指令文本区域 (#21299, @daniellok-db)\n- [UI] 修复图表中将“额外运行”聚合为“未分配”组的分组模式 (#21155, @copilot-swe-agent)\n- [UI] 修复启用工作区时下载工件的问题 (#21074, @timsolovev)\n- [追踪] 修复导出追踪时评估表的 trace_id 上的 NOT NULL 约束问题 (#21348, @dbczumar)\n- [跟踪] 修复当 `default_permission=NO_PERMISSIONS` 时，通过查询参数获取工件列表返回 403 Forbidden 错误的问题 (#21220, @copilot-swe-agent)\n- [UI] [ML-63097] 修复损坏的 LLM 评判器文档链接 (#21347, @smoorjani)\n- [追踪] 修复 Run Judge 因 litellm.InternalServerError: Invalid response object 而失败的问题 (#21262, @PattaraS)\n- [追踪 \u002F UI] 更新操作菜单：调整缩进以避免混淆 (#21266, @PattaraS)\n- [模型注册表] 修复 MlflowClient.copy_model_version 在跨工作区复制 UC 模型时的行为 (#21212, @WeichenXu123)\n- [UI] 修复经过清理后的空实验描述在描述框中的渲染问题 (#21223, @copilot-swe-agent)\n- [工件] 修复通过 `HttpArtifactRepository` 下载单个工件的问题 (#12955, @Koenkk)\n- [追踪] 修复 find_last_user_message_index 跳过技能内容注入的问题 (#21119, @alkispoly-db)\n- [追踪] 修复当跨度输出存储为字符串时提取检索上下文的问题 (#21213, @smoorjani)\n- [UI] 修复图表工具提示中的可见性切换按钮无法正常工作的问题 (#21071, @daniellok-db)\n- [UI] 将网关实验过滤移至服务器端查询，以解决页面大小不一致的问题 (#21138, @copilot-swe-agent)\n- [网关] 将带有 fallback_config 但没有 FALLBACK 模型的网关端点发出的误报警告降级为调试日志 (#21123, @copilot-swe-agent)\n- [追踪] 修复 MCP fn_wrapper 在处理具有 UNSET 默认值的可选参数时传递 None 的问题 (#21051, @yangbaechu)\n- [跟踪] 为 `logged_model` 表的 `experiment_id` 外键添加 CASCADE 约束 (#20185, @harupy)\n- [追踪] 修复 MCP fn_wrapper 对 Click UNSET 默认值的处理问题 (#20953) (#20962, @yangbaechu)\n\n文档更新：\n\n- [文档] 更新 SSO OIDC 插件文档：添加 Google Identity Platform、AWS Cognito 和 Azure Entra ID 的配置说明","2026-03-05T14:47:14",{"id":214,"version":215,"summary_zh":216,"released_at":217},280586,"v3.10.0","我们很高兴地宣布 MLflow 3.10.0 版本发布，其中包含多项重要更新：\n\n**重大新特性**：\n\n🏢 **MLflow Tracking Server 中的组织支持**：MLflow 现在支持多工作区环境。用户可以在单个跟踪服务器中以更粗粒度的单元对实验、模型和提示进行组织，并实现逻辑隔离。（#20702、#20657，@mprahl、@Gkrumbach07、@B-Step62）\n\n💬 **多轮评估与对话模拟**：MLflow 现在支持多轮评估，包括使用会话级评分器评估现有对话，以及通过模拟对话来测试代理的新版本，而无需重复生成对话。您可以使用 MLflow 3.8.0 中引入的会话级评分器和全新的会话 UI 来评估对话代理的质量，并启用自动评分功能，在追踪数据被摄入时持续监控质量。（#20243、#20377、#20289，@smoorjani）\n\n💰 **追踪成本跟踪**：现在您可以清晰了解 LLM 的使用支出！MLflow 会自动从 LLM 跟踪跨度中提取模型信息并计算成本，并通过新的 UI 直接在追踪视图中呈现模型和成本数据。（#20327、#20330，@serena-ruan）\n\n🎯 **导航栏重新设计**：我们重新设计了导航界面，旨在提供更加流畅的用户体验。顶级导航栏中的新工作流类型选择器可让您快速在 GenAI 和传统机器学习场景之间切换，同时精简的侧边栏减少了视觉上的杂乱感。（#20158、#20160、#20161、#20699，@ispoljari、@daniellok-db）\n\n🎮 **MLflow 演示实验**：刚接触 MLflow GenAI？只需点击一下，即可启动预填充的演示项目，实时探索追踪、评估和提示管理等功能。无需任何配置或编写代码。（#19994、#19995、#20046、#20047、#20048、#20162，@BenWilson2）\n\n📊 **网关使用情况跟踪**：通过详细的使用分析监控您的 AI 网关端点。新增的使用页面展示了请求模式和指标，并支持将网关调用与您的实验关联起来，从而实现端到端的可观测性。（#20357、#20358、#20642，@TomeHirata）\n\n⚡ **UI 内部追踪评估**：用户现在可以直接从追踪和会话 UI 中运行自定义或预构建的 LLM 评判器。这使得无需切换到 Python SDK 即可快速评估单个追踪条目及其上下文。（#20360，@hubertzub-db、@danielseong1）\n\n其他特性：\n\n- [UI] 为工作流切换组件添加滑动动画（#20831，@daniellok-db）\n- [追踪] 在追踪 UI 中显示缓存的 token（#20957，@TomeHirata）\n- [评估] 将“选择追踪”按钮移至“运行评判器”旁边（#20992，@PattaraS）\n- [网关] 为网关端点提供分布式追踪功能（#20864，@TomeHirata）\n- [网关] 在网关使用页面中添加用户选择器（#20944，@TomeHirata）\n- [文档] 为 GenAI 演示添加文档说明（#20240，@BenWilson2）\n- [UI] 将“入门指南”移至实验列表上方，并使其可折叠（#20691，@B-Step62）\n- [模型注册表\u002F追踪] 添加 mlflow `migrate-filestore` 命令","2026-02-20T16:05:05",{"id":219,"version":220,"summary_zh":221,"released_at":222},280587,"v3.10.0rc0","我们很高兴地宣布 MLflow 3.10.0rc0 的发布，其中包含多项重要更新：\n\n**重大新特性**：\n\n- 🏢 **MLflow Tracking Server 中的组织支持**：MLflow 现在支持多工作区环境！您可以通过全新的登录页面，在不同工作区之间无缝切换，从而更好地组织您的实验和资源。（#20702、#20657，@mprahl、@Gkrumbach07、@B-Step62）\n- 💬 **多轮对话模拟**：基于 3.9 版本中引入的对话模拟器，我们现在将其完全公开，并使其易于子类化。您可以创建自定义的模拟场景，比较具有特定目标或角色设定的会话，并将对话提炼为可重用的目标\u002F角色对，以进行全面的代理测试。（#20243、#20377、#20289，@smoorjani）\n- 💰 **追踪成本跟踪**：现在您可以清晰地了解 LLM 的使用支出！MLflow 会自动从 LLM 的跨度中提取模型信息并计算成本，并通过新的 UI 直接在追踪视图中呈现模型和成本数据。（#20327、#20330，@serena-ruan）\n- 🎯 **GenAI 与传统机器学习的顶级拆分**：我们重新设计了导航界面，提供更流畅的使用体验。顶部导航栏中的全新工作流类型选择器，可让您快速在 GenAI 和传统机器学习场景之间切换，同时精简的侧边栏也减少了视觉上的杂乱感。（#20158、#20160、#20161、#20699，@ispoljari、@daniellok-db）\n- 🎮 **MLflow 演示实验**：现在您可以比以往更快地开始使用 MLflow！新的 `mlflow demo` CLI 命令会生成一个完整的演示环境，包含示例追踪、提示和评估数据，让您无需任何配置即可动手探索 MLflow 的各项功能。（#19994、#19995、#20046、#20047、#20048、#20162，@BenWilson2）\n- 📊 **网关使用情况跟踪**：通过详细的使用分析监控您的 AI 网关端点。新增的使用情况页面展示了请求模式和指标，并支持追踪数据的导入，将网关调用与您的实验关联起来，实现端到端的可观ability。（#20357、#20358、#20642，@TomeHirata）\n\n敬请期待完整版的发布，届时将带来更多功能和问题修复。\n\n要试用此候选版本，请运行以下命令：\n\n`pip install mlflow==3.10.0rc0`","2026-02-12T05:01:00",{"id":224,"version":225,"summary_zh":226,"released_at":227},280588,"v3.9.0","我们很高兴地宣布 MLflow 3.9.0 版本发布，其中包含多项重要更新：\n\n**重大新特性**：\n\n- 🔮 **MLflow 助手**：弄清楚调试应用和智能体的下一步操作可能颇具挑战。为此，我们隆重推出 MLflow 助手——一款内嵌于产品的聊天机器人，可帮助您识别、诊断并修复问题。该助手由 Claude Code 提供支持，并能将 MLflow UI 中的上下文直接传递给 Claude。只需点击 MLflow UI 右下角的浮动“助手”按钮，即可开始使用！\n- 📈 **Trace 概览仪表板**：现在您可以通过 GenAI 实验中的全新“概览”选项卡，一目了然地了解智能体的性能表现。预置了丰富的统计指标，包括性能指标（如延迟、请求数）、质量指标（基于评估结果）以及工具调用摘要等。如果您希望添加更多图表，请随时在 MLflow 仓库中提交问题！\n- ✨ **AI 网关**：我们正在全面升级 AI 网关功能！AI 网关为您的 API 请求提供统一接口，使您可以将查询路由到您选择的 LLM 服务提供商。在 MLflow 3.9.0 中，网关服务器现已直接集成到跟踪服务器中，无需再启动新的进程。此外，还提供了直通端点、流量拆分和回退模型等功能，未来还将带来更多新特性！如需了解更多详细信息，请参阅 [文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway\u002F)。\n- 🔎 **使用 LLM 判官进行在线监控**：无需编写任何代码，即可配置 LLM 判官自动对您的 Trace 进行运行和评估！您可以直接使用我们提供的[预定义判官](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor\u002Fscorers\u002Fllm-judge\u002Fpredefined\u002F)，也可以自定义提示和说明来创建专属指标。请前往 GenAI 实验 UI 中的新“判官”选项卡开始使用。\n- 🤖 **判官构建器 UI**：您现在可以直接通过 UI 定义并迭代自定义的 LLM 判官提示！在新的“判官”选项卡中，您可以为 LLM 判官创建专属提示，并将其应用于您的 Trace 进行测试，以查看输出结果。当您对结果满意时，既可以用于上述的在线监控，也可以通过 Python SDK 应用于评估场景。\n- 🔗 **分布式追踪**：Trace 上下文如今可以在不同服务和进程之间传播，从而实现端到端的请求生命周期追踪。相关 API 已定义在 `mlflow.tracing.distributed` 模块中（更多文档即将发布）。\n- 📚 **MemAlign——全新的判官优化算法**：我们很高兴推出 `MemAlignOptimizer` 算法，它能够随着时间推移不断提升判官的智能化水平。该算法会从过往反馈中学习通用准则，同时在运行时动态检索相关示例，从而为您提供更精准的评估结果。\n\n功能：\n\n- [网关] 添加 LiteLLM 提供商以支持 ma","2026-01-29T08:49:56",{"id":229,"version":230,"summary_zh":231,"released_at":232},280589,"v3.9.0rc0","我们很高兴地宣布 MLflow 3.9.0rc0 预发布版本，其中包含多项值得关注的更新：\n\n**重大新特性**：\n\n- 🔮 **MLflow 助手**：弄清楚调试应用和智能体的下一步操作可能颇具挑战性。为此，我们隆重推出 MLflow 助手——一款内置于产品的聊天机器人，可帮助您识别、诊断并修复问题。该助手由 Claude Code 提供支持，并能将 MLflow UI 中的上下文直接传递给 Claude。只需点击 MLflow UI 右下角的浮动“助手”按钮，即可开始使用！\n- 📈 **Trace 概览仪表板**：现在，您可以通过 GenAI 实验中的全新“概览”选项卡，一目了然地了解智能体的性能表现。预置了大量开箱即用的统计指标，包括性能指标（如延迟、请求数）、质量指标（基于评估结果）以及工具调用摘要。如果您希望添加更多图表，请随时在 MLflow 仓库中提交问题！\n- ✨ **AI 网关**：我们正在全面升级 AI 网关功能！AI 网关为您的 API 请求提供统一接口，使您可以将查询路由到您选择的 LLM 提供商。在 MLflow 3.9.0rc0 中，网关服务器现已直接集成到跟踪服务器中，无需再启动新的进程。此外，还提供了直通端点、流量拆分和后备模型等附加功能，未来还将带来更多新特性！如需了解更多详细信息，请参阅 [文档](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fgovernance\u002Fai-gateway\u002F)。\n- 🔎 **LLM 判断器在线监控**：无需编写任何代码，即可配置 LLM 判断器自动对您的 Trace 进行运行！您可以直接使用我们提供的[预定义判断器](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Feval-monitor\u002Fscorers\u002Fllm-judge\u002Fpredefined\u002F)，也可以自定义提示和说明来创建专属指标。请前往 GenAI 实验 UI 中的新“判断器”选项卡开始使用。\n- 🤖 **判断器构建器 UI**：您现在可以直接在 UI 中定义并迭代自定义 LLM 判断器提示！在全新的“判断器”选项卡中，您可以为 LLM 判断器创建专属提示，并在自己的 Trace 上进行测试运行，以查看输出结果。当您对结果满意时，既可将其用于在线监控（如上所述），也可通过 Python SDK 在评估中使用。\n- 🔗 **分布式追踪**：Trace 上下文现可在不同服务和进程之间传播，从而实现端到端的请求生命周期追踪。相关 API 已在 `mlflow.tracing.distributed` 模块中定义（更多文档即将发布）。\n- 📚 **MemAlign——全新判断器优化算法**：我们很高兴推出 `MemAlignOptimizer`，这是一种能让您的判断器随着时间推移变得更加智能的新算法。它会从过往反馈中学习通用准则，同时在运行时动态检索相关示例，从而为您提供更精准的评估结果。\n\n敬请期待正式发布，届时将…","2026-01-16T04:48:57",{"id":234,"version":235,"summary_zh":236,"released_at":237},280590,"v3.8.1","MLflow 3.8.1 includes several bug fixes and documentation updates.\r\n\r\nBug fixes:\r\n\r\n- [Tracking] Skip registering sqlalchemy store when sqlalchemy lib is not installed (#19563, @WeichenXu123)\r\n- [Models \u002F Scoring] fix(security): prevent command injection via malicious model artifacts (#19583, @ColeMurray)\r\n- [Prompts] Fix prompt registration with model_config on Databricks (#19617, @TomeHirata)\r\n- [UI] Fix UI blank page on plain HTTP by replacing crypto.randomUUID with uuid library (#19644, @copilot-swe-agent)\r\n\r\nSmall bug fixes and documentation updates:\r\n\r\n#19539, #19451, #19409, @smoorjani; #19493, @alkispoly-db","2025-12-27T02:56:22",{"id":239,"version":240,"summary_zh":241,"released_at":242},280591,"v3.8.0","MLflow 3.8.0 includes several major features and improvements\r\n\r\n### Major Features\r\n\r\n- ⚙️ **Prompt Model Configuration**: Prompts can now include model configuration, allowing you to associate specific model settings with prompt templates for more reproducible LLM workflows. (#18963, #19174, #19279, @chenmoneygithub)\r\n- ⏳ **In-Progress Trace Display**: The Traces UI now supports displaying spans from in-progress traces with auto-polling, enabling real-time debugging and monitoring of long-running LLM applications. (#19265, @B-Step62)\r\n- ⚖️ **DeepEval and RAGAS Judges Integration**: New `get_judge` API enables using DeepEval and RAGAS evaluation metrics as MLflow scorers, providing access to 20+ evaluation metrics including answer relevancy, faithfulness, and hallucination detection. (#18988, @smoorjani, #19345, @SomtochiUmeh)\r\n- 🛡️ **Conversational Safety Scorer**: New built-in scorer for evaluating safety of multi-turn conversations, analyzing entire conversation histories for hate speech, harassment, violence, and other safety concerns. (#19106, @joelrobin18)\r\n- ⚡ **Conversational Tool Call Efficiency Scorer**: New built-in scorer for evaluating tool call efficiency in multi-turn agent interactions, detecting redundant calls, missing batching opportunities, and poor tool selections. (#19245, @joelrobin18)\r\n\r\n### Important Notice\r\n\r\n- **Collection of UI Telemetry**. From MLflow 3.8.0 onwards, MLflow will collect anonymized data about UI interactions, similar to the telemetry we collect for the Python SDK. If you manage your own server, UI telemetry is automatically disabled by setting the existing environment variables: `MLFLOW_DISABLE_TELEMETRY=true` or `DO_NOT_TRACK=true`. If you do not manage your own server (e.g. you use a managed service or are not the admin), you can still opt out personally via the new \"Settings\" tab in the MLflow UI. For more information, please read the documentation on [usage tracking](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fcommunity\u002Fusage-tracking\u002F).\r\n\r\n### Features:\r\n\r\n- [Tracking] Add default passphrase support (#19360, @BenWilson2)\r\n- [Tracing] Pydantic AI Stream support (#19118, @joelrobin18)\r\n- [Docs] Deprecate Unity Catalog function integration in AI Gateway (#19457, @harupy)\r\n- [Tracking] Add `--max-results` option to mlflow experiments search (#19359, @alkispoly-db)\r\n- [Tracking] Enhance encryption security (#19253, @BenWilson2)\r\n- [Tracking] Fix and simplify Gateway store interfaces (#19346, @BenWilson2)\r\n- [Evaluation] Add inference_params support for LLM Judges (#19152, @debu-sinha)\r\n- [Tracing] Support batch span export to UC Table (#19324, @B-Step62)\r\n- [Tracking] Add endpoint tags (#19308, @BenWilson2)\r\n- [Docs \u002F Evaluation] Add MLFLOW_GENAI_EVAL_MAX_SCORER_WORKERS to limit concurrent scorer execution (#19248, @debu-sinha)\r\n- [Evaluation \u002F Tracking] Enable search_datasets in Databricks managed MLflow (#19254, @alkispoly-db)\r\n- [Prompts] render text prompt previews in markdown (#19200, @ispoljari)\r\n- [UI] Add linked prompts filter for trace search tab (#19192, @TomeHirata)\r\n- [Evaluation] Automatically wrap async functions when passed to predict_fn (#19249, @smoorjani)\r\n- [Evaluation] [3\u002F6][builtin judges] Conversational Role Adherence (#19247, @joelrobin18)\r\n- [Tracking] [Endpoints] [1\u002Fx] Add backend DB tables for Endpoints (#19002, @BenWilson2)\r\n- [Tracking] [Endpoints] [3\u002Fx] Entities base definitions (#19004, @BenWilson2)\r\n- [Tracking] [Endpoints] [4\u002Fx] Abstract store interface (#19005, @BenWilson2)\r\n- [Tracking] [Endpoints] [5\u002Fx] SQL Store backend for Endpoints (#19006, @BenWilson2)\r\n- [Tracking] [Endpoints] [6\u002Fx] Protos and entities interfaces (#19007, @BenWilson2)\r\n- [Tracking] [Endpoints] [7\u002Fx] Add rest store implementation (#19008, @BenWilson2)\r\n- [Tracking] [Endpoints] [8\u002Fx] Add credential cache (#19014, @BenWilson2)\r\n- [Tracking] [Endpoints] [9\u002Fx] Add provider, model, and configuration handling (#19009, @BenWilson2)\r\n- [Evaluation \u002F UI] Add show\u002Fhide visibility control for Evaluation runs chart view (#18797) (#18852, @pradpalnis)\r\n- [Tracking] Add mlflow experiments get command (#19097, @alkispoly-db)\r\n- [Server-infra] [ Gateway 1\u002F10 ] Simplify secrets and masked secrets with map types (#19440, @BenWilson2)\r\n\r\n### Bug fixes:\r\n\r\n- [Tracing \u002F UI] Branch 3.8 patch: Fix GraphQL SearchRuns filter using invalid attribute key in trace comparison (#19526, @WeichenXu123)\r\n- [Scoring \u002F Tracking] Fix artifact download performance regression (#19520, @copilot-swe-agent)\r\n- [Tracking] Fix SQLAlchemy alias conflict in `_search_runs` for dataset filters (#19498, @fredericosantos)\r\n- [Tracking] Add auth support for GraphQL routes (#19278, @BenWilson2)\r\n- [] Fix SQL injection vulnerability in UC function execution (#19381, @harupy)\r\n- [UI] Fix MultiIndex column search crash in dataset schema table (#19461, @copilot-swe-agent)\r\n- [Tracking] Make datasource failures fail gracefully (#19469, @BenWilson2)\r\n- [Tracing \u002F Tracking] Fix litellm autolog for versions >= 1.78 (#19459, @har","2025-12-22T02:37:46",{"id":244,"version":245,"summary_zh":246,"released_at":247},280592,"v3.8.0rc0","MLflow 3.8.0rc0 includes several major features and improvements. More features to come in the final 3.8.0 release!\r\n\r\nTo try out this release candidate:\r\n\r\n```bash\r\npip install mlflow==3.8.0rc0\r\n```\r\n\r\n### Major Features\r\n\r\n- ⚙️ **Prompt Model Configuration**: Prompts can now include model configuration, allowing you to associate specific model settings with prompt templates for more reproducible LLM workflows. (#18963, #19174, #19279, @chenmoneygithub)\r\n- ⏳ **In-Progress Trace Display**: The Traces UI now supports displaying spans from in-progress traces with auto-polling, enabling real-time debugging and monitoring of long-running LLM applications. (#19265, @B-Step62)\r\n- ⚖️ **DeepEval Judges Integration**: New `get_judge` API enables using DeepEval's evaluation metrics as MLflow scorers, providing access to 20+ evaluation metrics including answer relevancy, faithfulness, and hallucination detection. (#18988, @smoorjani)\r\n- 🛡️ **Conversational Safety Scorer**: New built-in scorer for evaluating safety of multi-turn conversations, analyzing entire conversation histories for hate speech, harassment, violence, and other safety concerns. (#19106, @joelrobin18)\r\n- ⚡ **Conversational Tool Call Efficiency Scorer**: New built-in scorer for evaluating tool call efficiency in multi-turn agent interactions, detecting redundant calls, missing batching opportunities, and poor tool selections. (#19245, @joelrobin18)","2025-12-15T08:20:34",{"id":249,"version":250,"summary_zh":251,"released_at":252},280593,"v2.22.4","Version 2.22.4 is a patch release to backport several important fixes to MLflow 2.\r\n\r\n- Fix mlflow.spark.load_model to handle Unity Catalog Volumes paths correctly (https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fpull\u002F18672)\r\n- Introduce MLFLOW_CREATE_MODEL_VERSION_SOURCE_REGEX to validate source parameter of \u002Fmodel-versions\u002Fcreate request (https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fpull\u002F16081)\r\n- Fix spark udf on Databricks multi driver clusters (https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fpull\u002F18410)","2025-12-05T10:43:16",{"id":254,"version":255,"summary_zh":256,"released_at":257},280594,"v3.7.0","MLflow 3.7.0 includes several major features and improvements for GenAI Observability, Evaluation, and Prompt Management.\r\n\r\n### Major Features\r\n\r\n- 📝 **Experiment Prompts UI**: New prompts functionality in the experiment UI allows you to manage and search prompts directly within experiments, with support for filter strings and prompt version search in traces. (#19156, #18919, #18906, @TomeHirata)\r\n- 💬 **Multi-turn Evaluation Support**: Enhanced `mlflow.genai.evaluate` now supports multi-turn conversations, enabling comprehensive assessment of conversational AI applications with DataFrame and list inputs. (#18971, @AveshCSingh)\r\n- ⚖️ **Trace Comparison**: New side-by-side comparison view in the Traces UI allows you to analyze and debug LLM application behavior across different runs, making it easier to identify regressions and improvements. (#17138, @joelrobin18)\r\n- 🌐 **Gemini TypeScript SDK**: Auto-tracing support for Google's Gemini in TypeScript, expanding MLflow's observability capabilities for JavaScript\u002FTypeScript AI applications. (#18207, @joelrobin18)\r\n- 🎯 **Structured Outputs in Judges**: The `make_judge` API now supports structured outputs, enabling more precise and programmatically consumable evaluation results. (#18529, @TomeHirata)\r\n- 🔗 **VoltAgent Tracing**: Added auto-tracing support for VoltAgent, extending MLflow's observability to this AI agent framework. (#19041, @joelrobin18)\r\n\r\n### Breaking Changes\r\n\r\n- [Tracking] SQLite is now the default backend for the MLflow Tracking server. (#18497, @harupy)\r\n- [Models] Remove deprecated `diviner` flavor (#18808, @copilot-swe-agent)\r\n- [Models] Remove deprecated `promptflow` flavor (#18805, @copilot-swe-agent)\r\n\r\n### Features\r\n\r\n- [Tracking] Create parent directories for SQLite database files (#19205, @harupy)\r\n- [Prompts] Link Prompts and Experiments when prompts are loaded\u002Fregistered (#18883, @TomeHirata)\r\n- [Tracking] Include environment variable fallback for SGC run resumption (#19143, @artjen)\r\n- [Tracking] Add support for SGC run resumption from Databricks Jobs (#19015, @artjen)\r\n- [Evaluation] Add `--builtin\u002F-b` flag to `mlflow scorers list` command (#19095, @alkispoly-db)\r\n- [Tracing] Pydantic AI Chat UI support (#18777, @joelrobin18)\r\n- [Tracking] Add auth support for scorers (#18699, @BenWilson2)\r\n- [Evaluation] Remove experimental flags from scorers (#18122, @BenWilson2)\r\n- [Evaluation] Add description field to all built-in scorers (#18547, @alkispoly-db)\r\n\r\n### Bug Fixes\r\n\r\n- [Tracing] Handle traces with third-party generic root span (#19217, @B-Step62)\r\n- [Tracing] Fix OTLP endpoint path handling per OpenTelemetry spec (#19154, @harupy)\r\n- [Tracing] Add gzip\u002Fdeflate Content-Encoding support to OTLP traces endpoint (#19024, @Miaoxiang-philips)\r\n- [Tracing] Add missing `_delete_trace_tag_v3` API (#18813, @Tian-Sky-Lan)\r\n- [Tracing] Fix bug in chat sessions view where new sessions created after UI launch are not visible due to incorrect timestamp filtering (#18928, @dbczumar)\r\n- [Tracing] Fix OTLP proto conversion for empty list\u002Fdict (#18958, @B-Step62)\r\n- [Tracing] Agno V2 fixes (#18345, @joelrobin18)\r\n- [Tracing] Fix `\u002Fv1\u002Ftraces` endpoint to return protobuf instead of JSON (#18929, @copilot-swe-agent)\r\n- [Tracing] Pin `click!=8.3.0` in MCP extra to fix MCP server failure (#18748, @copilot-swe-agent)\r\n- [Tracing] Fix MCP server `uv` installation command for external users (#18745, @copilot-swe-agent)\r\n- [Evaluation] Fix trace-based scorer evaluation by using agentic judge adapter (#19123, @alkispoly-db)\r\n- [Evaluation] Fix managed scorer registration failure (#19146, @xsh310)\r\n- [Evaluation] Fix `InstructionsJudge` using scorer description as assessment value (#19121, @alkispoly-db)\r\n- [Evaluation] Add validation to correctness judge expectation fields (#19026, @smoorjani)\r\n- [Evaluation] Fix model URI underscore handling (#18849, @RohanRouth)\r\n- [Evaluation] Fix `evaluate_traces` MCP tool error: use `result_df` instead of `tables` (#18825, @alkispoly-db)\r\n- [Evaluation] Fix Bedrock Anthropic adapter by adding required `anthropic_version` field (#17744, @harupy)\r\n- [Evaluation] Fix migration for pre-existing auth tables (#18793, @BenWilson2)\r\n- [Tracking] Fix tracking URI propagation (#18023, @shaperilio)\r\n- [Tracking] Fix `SqlLoggedModelMetric` association with `experiment_id` (#18382, @mcompen)\r\n- [Tracking] Add Flask routes to auth validators (#18486, @BenWilson2)\r\n- [Tracking] Add missing proto handler for Experiment association handling for datasets (#18769, @BenWilson2)\r\n- [UI] Show full dataset record content and add search bar in evaluation datasets UI (#19000, @dbczumar)\r\n- [UI] Request TraceInfo and Trace Assessments from a relative API path (#19032, @kbolashev)\r\n- [UI] Define `LoggedModelOutput.to_dictionary()` so `LoggedModelOutput` and runs containing them can be JSON serialized (#19017, @nicklamiller)\r\n- [UI] Fix router issue in TracesUI page (#19044, @joelrobin18)\r\n- [Build] Fix `mlflow gc` to remove model artifacts (#1728","2025-12-05T17:30:23",{"id":259,"version":260,"summary_zh":261,"released_at":262},280595,"v3.7.0rc0","MLflow 3.7.0rc0 includes several major features and improvements!\r\n\r\n### Major Features\r\n\r\n- ⚖️ **Trace Comparison**: New UI feature allowing side-by-side comparison of traces to analyze and debug LLM application behavior across different runs. (#17138, @joelrobin18, @daniellok-db)\r\n- 💬 **Multi-turn conversation support for Evaluation**: Enhanced evaluation support for multi-turn conversations in `mlflow.genai.evaluate`, enabling comprehensive assessment of conversational AI applications. (#18971, #19039, @AveshCSingh)\r\n- 🔎 **Full Text Trace Search from UI**: Search across all trace content directly from the UI, making it easier to find specific traces by searching through inputs, outputs, and span details. (#18683, @dbczumar)\r\n- 🌐 **Gemini TypeScript SDK**: Auto-tracing support for Gemini in TypeScript, expanding MLflow's observability capabilities for JavaScript\u002FTypeScript AI applications. (#18207, @joelrobin18)\r\n\r\n### Breaking Changes\r\n\r\n- **SQLite as Default Backend**: MLflow now uses SQLite as the default backend instead of file-based storage, unless existing mlruns data is detected. This improves performance and reliability for tracking experiments. (#18497, @harupy)\r\n- **Removed Deprecated Flavors**: The `diviner` and `promptflow` flavors have been removed from MLflow. Please migrate to supported alternatives. (#18808, #18805, @copilot-swe-agent)\r\n\r\n### Important Notice\r\n\r\n- **Installation ID for Telemetry**: MLflow now generates a unique installation ID (a randomly generated UUID) for telemetry purposes to better understand usage patterns. This ID is fully anonymous and persists across sessions. Telemetry can be disabled anytime by setting `MLFLOW_DISABLE_TELEMETRY=true` or `DO_NOT_TRACK=true`. See the [usage tracking documentation](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fcommunity\u002Fusage-tracking\u002F) for details. (#18881, @B-Step62)\r\n\r\nStay tuned for the full release, which will be packed with more features and bugfixes.\r\n\r\nTo try out this release candidate, please run:\r\n\r\n`pip install mlflow==3.7.0rc0`","2025-11-27T10:47:03",{"id":264,"version":265,"summary_zh":266,"released_at":267},280596,"v3.6.0","MLflow 3.6.0 includes several major features and improvements for AI Observability, Experiment UI, Agent Evaluation and Deployment.\r\n\r\n- 🔗 **Full OpenTelemetry Support in OSS Server**: MLflow now offers comprehensive OpenTelemetry integration, allowing you to ingest OpenTelemetry traces into MLflow and use both SDK seamlessly together. (#18540, #18532, #18357, @B-Step62, @serena-ruan)\r\n- 💬 **Session-level View in Trace UI**: New chat sessions tab provides a dedicated view for organizing and analyzing related traces at the session level, making it easier to track conversational workflows. (#18594, @daniellok-db)\r\n- 🧭 **New experiment tab bar**: The experiment tab navigation bar has been moved from the top of the page to the left side. As MLflow continues to grow, this layout provides more room to add new tabs while keeping everything easy to find. (#18594, @daniellok-db)\r\n- 🚀 **New Supported Frameworks in TypeScript Tracing SDK**: Auto-tracing support for **Vercel AI SDK**, **Gemini**, **Anthropic**, **Mastra** in TypeScript, expanding MLflow's observability capabilities across popular JavaScript\u002FTypeScript frameworks. (#18402, @B-Step62)\r\n- 💰 **Tracking Judge Cost and Traces**: Comprehensive tracking of LLM judge evaluation costs and traces, providing visibility into evaluation expenses and performance with automatic cost calculation and rendering. (#18481, #18484, @B-Step62)\r\n- ⚙️ **Agent Server**: New agent server infrastructure for managing and deploying scoring agents with enhanced orchestration capabilities. (#18596, @bbqiu)\r\n\r\nBreaking changes:\r\n\r\n- Deprecate pmdarima, promptflow, diviner flavors (#18597, #18577, @copilot-swe-agent)\r\n- Drop numbering suffix (`_1`, `_2`, ...) from span names (#18531, @serena-ruan)\r\n\r\n## Features\r\n- [Evaluation] Support structured outputs in make_judge (#18529, @TomeHirata)\r\n- [Evaluation] Agent-as-a-judge support for default Databricks endpoint (#18709, @smoorjani)\r\n- [Evaluation] Frontend adjustments for handle judge traces (#18485, @B-Step62)\r\n- [Evaluation] Record judge traces (#18484, @B-Step62)\r\n- [Evaluation] [ML-57683] Add `search_traces` tool for agentic judge (#18228, @dbrx-euirim)\r\n- [Evaluation] Record and render LLM judge cost (#18481, @B-Step62)\r\n- [Evaluation] Add support for profile usage in Databricks Agents dataset API operat… (#18431, @BenWilson2)\r\n- [Evaluation] Add description property to Scorer interface (#18383, @alkispoly-db)\r\n- [Evaluation] Add mlflow scorers register-llm-judge CLI command (#18330, @alkispoly-db)\r\n- [Evaluation] Allow passing empty scorer list for manual result comparison (#18265, @B-Step62)\r\n- [Evaluation] Add CLI command to list registered scorers by experiment (#18255, @alkispoly-db)\r\n- [Evaluation] Log assessments to DSPy eval traces (#18136, @B-Step62)\r\n- [Evaluation] Add mlflow traces eval CLI command (#18069, @alkispoly-db)\r\n- [Tracing] Add documentation for new tracing integrations (Otel) (#18691, @B-Step62)\r\n- [Tracing] Display trace metadata (#18609, @B-Step62)\r\n- [Tracing] Support automatically tracking session ID for LangGraph (#18608, @B-Step62)\r\n- [Tracing \u002F Tracking] Add RLIKE operator support for trace search (#18591, @serena-ruan)\r\n- [Tracing] Attributes translation for OTel clients (#18532, @serena-ruan)\r\n- [Tracing] [Vercel #3] Implement auto-tracing logic for Vercel AI SDK (#18402, @B-Step62)\r\n- [Tracing] Minor clean up for the trace summary view (#18436, @B-Step62)\r\n- [Tracing] Support search by span details for traces in OSS MLflow server (#17918, @serena-ruan)\r\n- [UI] UI: Support filtering by span content \u002F type \u002F name (#18683, @dbczumar)\r\n- [UI] Add chat sessions tab (#18594, @daniellok-db)\r\n- [UI] Child Parent Link (#17248, @joelrobin18)\r\n- [Tracking] Make Pytorch lightning autologging support logging model signature (#18510, @WeichenXu123)\r\n- [Tracking] Add support for using the same DB for tracking and auth (#18384, @BenWilson2)\r\n- [Tracking] Job backend: Support creating virtual python environment for job execution (#18111, @WeichenXu123)\r\n- [Model Registry \u002F Tracking] Add deprecation warnings for filesystem backends (#18524, @harupy)\r\n- [Model Registry] Allow for skipping pip installation while packing environment for model serving (#18448, @juntai-zheng)\r\n- [Models] Support Langchain 1.x (#18490, @BenWilson2)\r\n- [Models] Use UBJSON format as default for XGBoost models (#18420, @harupy)\r\n- [Scoring] Introduce Agent Server (#18596, @bbqiu)\r\n- [Deployment] Add configuration option for long-running deployments client requests (#18363, @BenWilson2)\r\n- [Gateway] Make Openai provider supporting streamed function calling response (#18367, @WeichenXu123)\r\n- [Gateway] Make Gemini provider supporting function calling (#18328, @WeichenXu123)\r\n- [Gateway] AI-gateway revamp: Make anthropic provider supporting function calling (#18294, @WeichenXu123)\r\n- [Gateway] AI-gateway revamp: Add traffic route to multiple endpoints (#18064, @WeichenXu123)\r\n- [Build] Move fastmcp to optional mcp extra (#18422, @harupy)\r\n- [Do","2025-11-08T06:15:57",{"id":269,"version":270,"summary_zh":271,"released_at":272},280597,"v3.6.0rc0","MLflow 3.6.0rc0 includes several major features and improvements!\r\n\r\n### Major Features\r\n\r\n- 🔗 **Full OpenTelemetry Support in OSS Server**: MLflow now offers comprehensive OpenTelemetry integration, allowing you to use OpenTelemetry and MLflow SDK together for constructing unified traces with full OTLP span ingestion. (#18540, #18532, #18357, @B-Step62, @serena-ruan)\r\n- 💬 **Session-level View in Trace UI**: New chat sessions tab provides a dedicated view for organizing and analyzing related traces at the session level, making it easier to track conversational workflows. (#18594, @daniellok-db)\r\n- 🧭 **New experiment tab bar**: The experiment tab navigation bar has been moved from the top of the page to the left side. As MLflow continues to grow, this layout provides more room to add new tabs while keeping everything easy to find. (#18594, @daniellok-db)\r\n- 🚀 **Vercel AI Support in TypeScript Tracing SDK**: Auto-tracing support for Vercel AI SDK in TypeScript, expanding MLflow's observability capabilities across popular JavaScript\u002FTypeScript frameworks. (#18402, @B-Step62)\r\n- 💰 **Tracking Judge Cost and Traces**: Comprehensive tracking of LLM judge evaluation costs and traces, providing visibility into evaluation expenses and performance with automatic cost calculation and rendering. (#18481, #18484, @B-Step62)\r\n- ⚙️ **Agent Server**: New agent server infrastructure for managing and deploying scoring agents with enhanced orchestration capabilities. (#18596, @bbqiu)\r\n\r\n### Breaking Changes and deprecations\r\n\r\n- **[Tracking] Filesystem Backend Deprecation**: The filesystem backend is being deprecated in favor of SQLite. See [#18534](https:\u002F\u002Fgithub.com\u002Fmlflow\u002Fmlflow\u002Fissues\u002F18534) for details.\r\n- [Flavors] Deprecate promptflow flavor (#18597, @copilot-swe-agent)\r\n- [Flavors] Deprecate pmdarima and diviner flavors (#18577, @copilot-swe-agent)\r\n- [Tracing] Drop span name deduplication (#18531, @serena-ruan)\r\n\r\nStay tuned for the full release, which will be packed with more features and bugfixes.\r\n\r\nTo try out this release candidate, please run: `pip install mlflow==3.6.0rc0`","2025-11-04T02:19:05",{"id":274,"version":275,"summary_zh":276,"released_at":277},280598,"v3.5.1","MLflow 3.5.1 is a patch release that includes several bug fixes and improvements.\r\n\r\nFeatures:\r\n\r\n- [CLI] Add CLI command to list registered scorers by experiment (#18255, @alkispoly-db)\r\n- [Deployments] Add configuration option for long-running deployments client requests (#18363, @BenWilson2)\r\n- [Deployments] Create `set_databricks_monitoring_sql_warehouse_id` API (#18346, @dbrx-euirim)\r\n- [Prompts] Show instructions for prompt optimization on prompt registry (#18375, @TomeHirata)\r\n\r\nBug fixes:\r\n\r\n- [Evaluation] Validate if trace is None before accessing the value in mlflow.genai.evaluate (#18285, @srinathmkce)\r\n- [Evaluation] Revert \"Add atomicity to job_start API (#18226)\" (@serena-ruan)\r\n- [MCP] Move fastmcp to optional mcp extra (#18422, @harupy)\r\n- [Model Registry] Fix serialization bug in file store (#18365, @BenWilson2)\r\n- [Scoring] Pin uvloop\u003C0.22 to fix mlserver compatibility (#18370, @harupy)\r\n- [Tracing] Fix a forward-compatibility issue with Span to_dict (#18439, @serena-ruan)\r\n- [Tracing] Whitelist notebook trace UI renderer to allow display with default security settings (#18446, @TomeHirata)\r\n- [Tracing] Fix attribute error in StrandsAgent tracing (#18409, @B-Step62)\r\n- [Tracing] Adjust truncation logic in trace previews (#18412, @BenWilson2)\r\n- [Tracing] Revert \"Fix response handling in log_spans (#18280)\" (#18349, @serena-ruan)\r\n- [Tracking] Adjust util for remote tracking server declaration (#18411, @BenWilson2)\r\n- [Tracking] Handle Databricks FMAPI style in openai autolog (#18354, @TomeHirata)\r\n- [Tracking] Fetch config after adding first record (#18338, @serena-ruan)\r\n- [UI] Fix span ID parsing in the UI (#18419, @daniellok-db)\r\n- [UI] Fix Chat message parsing within the trace summary view modal (#18454, @daniellok-db)\r\n- [UI] Fix an issue with display of the assessments pane in the UI (#18333, @BenWilson2)\r\n\r\nDocumentation updates:\r\n\r\n- [Docs] Fix Kubernetes Deployment Tutorial Code (#18381, @Maeril)\r\n- [Docs] Update the documentation around requirements for optimize_prompts (#18398, @TomeHirata)\r\n- [Docs] Fix example FastAPI in track user sessions (#18388, @maxscheijen)","2025-10-22T19:07:21",{"id":279,"version":280,"summary_zh":281,"released_at":282},280599,"v3.5.0","MLflow 3.5.0 includes several major features and improvements!\r\n\r\n### Major Features\r\n\r\n- 🤖 **Tracing support for Claude Code SDK**: MLflow now provides a tracing integration for both the Claude Code CLI and SDK! Configure the autologging integration to track your prompts, Claude's responses, tool calls, and more. Check out this [doc page](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Ftracing\u002Fintegrations\u002Flisting\u002Fclaude_code\u002F) to get started. (#18022, @smoorjani)\r\n- 🎯 **Flexible Prompt Optimization API**: Introduced a new flexible API for prompt optimization with support for model switching and the GEPA algorithm, enabling more efficient prompt tuning with fewer rollouts. See the [documentation](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fgenai\u002Fprompt-registry\u002Foptimize-prompts\u002F) to get started. (#18183, #18031, @TomeHirata)\r\n- 🎨 **Enhanced UI Onboarding**: Improved in-product onboarding experience with trace quickstart drawer and updated homepage guidance to help users discover MLflow's latest features. (#18098, #18187, @B-Step62)\r\n- 🔐 **Security Middleware for Tracking Server**: Added a security middleware layer to protect against DNS rebinding, CORS attacks, and other security threats. Read the [documentation](https:\u002F\u002Fmlflow.org\u002Fdocs\u002Flatest\u002Fml\u002Ftracking\u002Fserver\u002Fsecurity\u002F) for configuration details. (#17910, @BenWilson2)\r\n\r\n### Features\r\n\r\n- [Tracing \u002F Tracking] Add `unlink_traces_from_run` batch operation (#18316, @harupy)\r\n- [Tracing] Add batch trace link\u002Funlink operations to DatabricksTracingRestStore (#18295, @harupy)\r\n- [Tracking] Claude Code SDK autologging support (#18022, @smoorjani)\r\n- [Tracing] Add support for reading trace configuration from environment variables (#17792, @joelrobin18)\r\n- [Tracking] Mistral tracing improvements (#16370, @joelrobin18)\r\n- [Tracking] Gemini token count tracking (#16248, @joelrobin18)\r\n- [Tracking] Gemini streaming support (#16249, @joelrobin18)\r\n- [Tracking] CrewAI token count tracking with documentation updates (#16373, @joelrobin18)\r\n- [Evaluation] Allow passing empty scorer list for manual result comparison (#18265, @B-Step62)\r\n- [Evaluation] Log assessments to DSPy evaluation traces (#18136, @B-Step62)\r\n- [Evaluation] Add support for trace inputs to built-in scorers (#17943, @BenWilson2)\r\n- [Evaluation] Add synonym handling for built-in scorers (#17980, @BenWilson2)\r\n- [Evaluation] Add span timing tool for Agent Judges (#17948, @BenWilson2)\r\n- [Evaluation] Allow disabling evaluation sample check (#18032, @B-Step62)\r\n- [Evaluation] Reduce verbosity of SIMBA optimizer logs when aligning judges (#17795, @BenWilson2)\r\n- [Evaluation] Add `__repr__` method for Judges (#17794, @BenWilson2)\r\n- [Prompts] Add prompt registry support to MLflow webhooks (#17640, @harupy)\r\n- [Prompts] Prompt Registry Chat UI (#17334, @joelrobin18)\r\n- [UI] Delete parent and child runs together (#18052, @joelrobin18)\r\n- [UI] Added move to top, move to bottom for charts (#17742, @joelrobin18)\r\n- [Tracking] Use sampling data for run comparison to improve performance (#17645, @lkuo)\r\n- [Tracking] Add optional 'outputs' column for evaluation dataset records (#17735, @WeichenXu123)\r\n- [Tracking] Job backend execution (#17676, #18012, #18070, #18071, #18112, #18049, @WeichenXu123)\r\n\r\n### Bug Fixes\r\n\r\n- [Tracing] Fix parent run resolution mechanism for LangChain (#17273, @B-Step62)\r\n- [Tracing] Add client-side retry for `get_trace` to improve reliability (#18224, @B-Step62)\r\n- [Tracing] Fix OpenTelemetry dual export (#18163, @B-Step62)\r\n- [Tracing] Suppress false warnings from span logging (#18092, #18276, @B-Step62)\r\n- [Tracing] Fix OpenTelemetry resource attributes not propagating correctly (#18019, @xiaosha007)\r\n- [Tracing] Fix DSPy prompt display (#17988, @B-Step62)\r\n- [Tracing] Fix usage aggregation to avoid ancestor duplication (#17921, @TomeHirata)\r\n- [Tracing] Fix double counting in Strands tracing (#17855, @joelrobin18)\r\n- [Tracing] Fix `to_predict_fn` to handle traces without tags field (#17784, @harupy)\r\n- [Tracing] URL-encode trace tag keys in `delete_trace_tag` to prevent 404 errors (#18232, @copilot-swe-agent)\r\n- [Tracking] Fix Claude Code autologging inputs not displaying (#17858, @smoorjani)\r\n- [Tracking] Fix runs with 0-valued metrics not appearing in experiment list contour plots (#17916, @WeichenXu123)\r\n- [Tracking] Fix DSPy run display (#18137, @B-Step62)\r\n- [Tracking] Allow list of types in tools JSON Schema for OpenAI autolog (#17908, @fedem96)\r\n- [Tracking] Set tracking URI environment variable for job runner (#18073, @WeichenXu123)\r\n- [Evaluation] Add atomicity to `job_start` API (#18226, @BenWilson2)\r\n- [Evaluation] Fix trace ingest for outputs in `merge_records()` API (#18047, @BenWilson2)\r\n- [Evaluation] Fix judge regression (#18039, @B-Step62)\r\n- [Evaluation] Fix judges to use non-empty user messages for Anthropic model compatibility (#17935, @dbczumar)\r\n- [Evaluation] Fix endpoints error in judge (#18048, @joelrobin18)\r\n- [Model Registry] Fix creating model versions from non-Databricks tracking","2025-10-16T15:17:59"]