MLflow

MLflow 3.10.0 introduces major features like organization support for multi-workspace tracking, advanced multi-turn conversation evaluation, and automatic LLM trace cost tracking. The release also includes a significant navigation bar redesign and a new one-click demo experiment.

MLflow 3.10.0rc0 introduces major features like Organization Support for multi-workspace tracking, enhanced conversation simulation, and automatic LLM cost tracking. This release candidate also includes a new demo command and navigation redesign for better usability.

MLflow 3.9.0 introduces significant enhancements to GenAI capabilities, including the MLflow Assistant chatbot, Trace Overview Dashboard, and a revamped AI Gateway integrated into the tracking server. This release also focuses heavily on evaluation tooling with new LLM judges, judge builders, and distributed tracing support.

MLflow 3.9.0rc0 is a pre-release introducing significant GenAI features like the MLflow Assistant, Trace Overview Dashboard, integrated AI Gateway, and LLM Judge capabilities.

MLflow 3.8.1 primarily delivers several bug fixes across tracking, models, prompts, and UI components, along with minor documentation updates.

MLflow 3.8.0 introduces major new capabilities such as prompt model configuration, in‑progress trace display, DeepEval/RAGAS judges integration, and two new conversational scorers, while also adding numerous tracking, tracing, and evaluation enhancements and fixing a wide range of bugs.

MLflow 3.8.0rc0 introduces prompt model configuration, in‑progress trace display, DeepEval judges integration, and two new conversational scorers, with no breaking changes.

MLflow 3.7.0 adds major GenAI observability features such as an Experiment Prompts UI, multi-turn evaluation, trace comparison, and new auto-tracing SDKs, while introducing breaking changes like SQLite becoming the default tracking backend and removal of deprecated model flavors. It also includes numerous bug fixes and enhancements across tracking, evaluation, tracing, and UI components.

Patch 2.22.4 backports fixes for Unity Catalog Volumes path handling and Spark UDFs, and adds a regex constant for model version source validation.

MLflow 3.7.0rc0 introduces trace comparison, multi‑turn evaluation, full‑text UI search, and a Gemini TypeScript SDK, while changing the default backend to SQLite and removing the deprecated `diviner` and `promptflow` flavors.

MLflow 3.6.0 introduces full OpenTelemetry support, a new Agent Server, extensive tracing and evaluation enhancements, and numerous UI and framework integrations, while deprecating several flavors and changing span naming conventions.

MLflow 3.6.0rc0 adds full OpenTelemetry support, session‑level trace UI, a new experiment tab layout, Vercel AI TypeScript tracing, cost tracking for LLM judges, and an Agent Server, while deprecating the filesystem backend and removing span name deduplication.

MLflow 3.5.1 is a patch release that adds new CLI and deployment features, introduces a monitoring API, and includes numerous bug fixes across evaluation, tracing, tracking, and UI components.

MLflow 3.5.0 adds major tracing, prompt optimization, UI onboarding, and security middleware features, along with numerous enhancements and bug fixes across tracing, tracking, evaluation, and model registry.

MLflow 3.5.0rc0 introduces Claude Code SDK tracing, UI enhancements, evaluation datasets UI, GEPA prompt optimization, and a default security middleware layer for the tracking server.

MLflow 3.4.0rc0 adds extensive new capabilities—including OpenTelemetry metrics export, MCP server integration, a custom judges API, and experiment types UI—while delivering numerous feature enhancements and bug fixes across evaluation, tracing, CLI, tracking, and model registry.

MLflow 3.4.0rc0 introduces major new capabilities including OpenTelemetry metrics, MCP server integration, a custom judges API, correlations backend, evaluation datasets, Databricks backend support, Claude autologging, and Strands agent tracing.

A lightweight patch release backports the fix from issue #15970 to v2.22.2.

MLflow 3.3.2 adds dataset name persistence for evaluation and includes several bug fixes and documentation updates.

MLflow 3.3.1 primarily delivers bug fixes, including corrections to the mlflow.genai.datasets attribute, UI tag display, and dspy tracing performance.

MLflow 3.3.0 adds webhooks, Agno tracing, open‑source GenAI evaluation, a revamped trace UI, and switches the default tracking server to FastAPI + Uvicorn while delivering numerous bug fixes and new feature enhancements.

MLflow 3.3.0rc0 introduces model registry webhooks, Agno tracing integration, open‑source GenAI evaluation, a revamped trace table UI, and switches the default Tracking Server to FastAPI + Uvicorn.

MLflow 3.2.0 introduces major GenAI tracing enhancements, feedback tracking, UI redesigns, Polars dataset support, and a suite of new features and bug fixes, while adding optional usage tracking.

MLflow 3.2.0 introduces extensive GenAI tracing enhancements, new feedback tracking, UI improvements, Polars dataset support, and several feature additions and bug fixes, all without breaking changes.

MLflow 3.1.4 is a minor release that adds several major features and improvements along with small bug fixes and documentation updates.

MLflow 3.1.3 introduces artifact permission handling and OpenAI ChatCompletions parsing support while fixing bugs in deployments, model registry, evaluation, and tracking.

MLflow 3.1.2 is a patch release that fixes several bugs in tracking, models, and tracing components.

Automated nightly build of MLflow (commit bdc041eb85cb37856e16b14a22c5b09c92c18e5a) containing the latest changes from the master branch; no specific changelog items are listed.

MLflow 3.1.1 adds a larger prompt limit, pagination for history retrieval, and several bug fixes including artifact download support and security hardening.

MLflow 3.0.1 adds a larger prompt text limit (5K→100K) and fixes bedrock provider compatibility, along with assorted small bug fixes and documentation updates.

The provided release notes only reference the v3.1.0 tag on the MLflow GitHub repository and contain no specific change details.

MLflow 3 introduces a model‑centric GenAI architecture with new LoggedModel entity, prompt optimization, enhanced tracing, and many integrations, while also delivering several breaking changes that necessitate migration steps.

MLflow 2.22.1 adds DBR 15.4 support for spark_udf in the DBConnect client and includes several bug fixes across Model Registry, Tracking, and documentation.

MLflow 3.0.0rc1 release candidate is now available; upgrade with `pip install mlflow==3.0.0rc1` and refer to the documentation for details.

MLflow 3.0.0rc3 release candidate is now available; upgrade via pip and refer to the documentation for details.

MLflow 3.1.0rc0 release candidate is now available; upgrade with pip install mlflow==3.1.0rc0.

MLflow 3.0.0rc2 release candidate is now available; upgrade with `pip install mlflow==3.0.0rc2` and refer to the documentation for details.

MLflow 2.22.0 adds new tracing capabilities, Gemini embeddings, ADLS artifact support, and several bug fixes and UI improvements.

MLflow 3.0.0rc0 release candidate is now available for upgrade via pip.

MLflow 2.21.3 adds a `return_type` argument to the search_traces API and includes several bug fixes for Spark model handling and other minor issues.

MLflow 2.21.2 introduces a bug fix for Databricks trace export connection exhaustion and adds result‑table logging for DSPy optimizer tracking.

MLflow 2.21.1 adds DSPy evaluation logging, run creation on DSPy compile, and a Java‑less SageMaker container option, while fixing several tracing, tracking, and UI bugs.

MLflow 2.21.0 introduces a redesigned documentation site, a Prompt Registry, FastAPI‑based inference server, enhanced tracing, and numerous new integrations and bug fixes.

MLflow 2.21.0rc0 introduces a redesigned documentation site, migrates the scoring server to FastAPI, and adds enhanced tracing capabilities, while deprecating Recipes and fastai/f2o flavors.

MLflow 2.20.3 adds GPU metrics for AMD/HIP GPUs, txtai tracing, Google GenAI SDK support, and Anthropic Claude 3.7 thinking block, while fixing LangGraph tracing and numerous minor bugs.

MLflow 2.20.2 adds tracing support for generator functions, enhances ChatAgent and Langgraph connectors, introduces VariantType in spark_udf, and includes several bug fixes and documentation updates.

MLflow 2.20.1 is a patch release that adds Spark_udf support for model signatures, helper connectors for ChatAgent with LangChain/LangGraph, and default RUC/Lift curves for CatBoost classifiers, while fixing several bugs including Pydantic 1.x compatibility and LiteLLM tracing issues.

MLflow 2.20.0 introduces type‑hint‑based model signatures, Bedrock/Groq tracing, inline Jupyter trace rendering, uv‑powered model validation, a new chat panel, and several API enhancements along with numerous bug fixes.

MLflow 2.20.0rc0 introduces type‑hint based model signatures, Bedrock/Groq tracing, inline notebook trace rendering, uv‑accelerated model validation, a new chat panel in the Trace UI, and several enhancements such as the ChatAgent base class and improved Spark UDF handling.