# MLflow

An open source platform for debugging, evaluating, monitoring and optimizing production AI applications built with agents, LLMs and machine learning models, while tracking cost and access.

MLflow is an open source AI engineering platform for agents, LLM applications and traditional machine learning models. It helps teams debug, evaluate, monitor and optimize production AI applications while controlling costs and managing access to models and data.

For agents and LLM applications it provides production-grade observability built on OpenTelemetry, capturing complete traces for any LLM provider or agent framework; systematic evaluation with many built-in metrics and LLM judges to track quality over time and catch regressions; and prompt management, prompt optimization, and an AI Gateway for managing model costs and access. It supports Python, TypeScript/JavaScript and Java, and integrates natively with OpenTelemetry and MCP.

It runs as a local or self-hosted server started with a single command, or connects to a Databricks workspace; a setup wizard can also configure tracing for an existing project. It is written in Python and licensed Apache-2.0.

Website: https://mlflow.org