Mlflow

MLflow is an open source platform for building, debugging, evaluating and monitoring LLM applications, agents and traditional machine learning models. It is backed by the Linux Foundation and is used by developers, data scientists and ML platform teams, with support for Python, TypeScript and JavaScript, Java and R and integrations across a wide range of agent frameworks and ML libraries. The platform covers several jobs. Observability captures full traces of LLM applications and agents, built on OpenTelemetry and working with any LLM provider. Evaluation runs systematic checks with more than 50 built in metrics and LLM judges, and can analyze traces for issues across correctness, latency, relevance and safety. Prompt management versions, tests and deploys prompts with lineage tracking and offers automatic prompt optimization. An AI Gateway provides one interface for multiple LLM providers with routing, rate limits, fallbacks and cost controls, and an Agent Server hosts agents as production endpoints with streaming and built in tracing. Classic model training and experiment tracking remain part of the product. MLflow is free to use under the Apache 2.0 license. The site publishes no paid plans or prices.