Cette page est affichée en anglais. Une traduction française est en cours.
ConfigsMar 31, 2026·2 min de lecture

MLflow — Open Source AI Engineering Platform

MLflow is the largest open-source AI engineering platform for tracing, evaluation, prompt management, and model deployment. 25K+ GitHub stars. 60M+ monthly downloads. Apache 2.0.

Introduction

MLflow is the largest open-source platform for AI engineering, enabling teams to debug, evaluate, monitor, and optimize production AI applications for agents, LLMs, and ML models. With 25,000+ GitHub stars, 60+ million monthly downloads, and Apache 2.0 license, MLflow provides observability and tracing (OpenTelemetry-based), 50+ built-in evaluation metrics, prompt management with version control, an AI Gateway for unified multi-provider API access with rate limiting and cost control, experiment tracking, model registry, and deployment tools. Supports 60+ frameworks including LangChain, OpenAI, and Anthropic.

Best for: ML/AI teams needing end-to-end experiment tracking, model management, and production monitoring Works with: Claude Code, OpenAI Codex, Cursor, Gemini CLI, Windsurf Integrations: 60+ frameworks (LangChain, OpenAI, Anthropic, PyTorch, etc.)


Key Features

  • Observability: OpenTelemetry-based tracing for LLM applications
  • 50+ evaluation metrics: Built-in LLM judges for quality assessment
  • Prompt management: Version, test, deploy prompts with lineage
  • AI Gateway: Unified API across providers with rate limiting and cost control
  • Model registry: Version, stage, and deploy models
  • 60+ integrations: LangChain, OpenAI, Anthropic, PyTorch, and more
  • Multi-language: Python, TypeScript, Java, OpenTelemetry

FAQ

Q: What is MLflow? A: MLflow is the largest open-source AI engineering platform with 25K+ stars. Tracing, 50+ eval metrics, prompt management, AI Gateway, model registry. 60M+ monthly downloads. Apache 2.0.

Q: How do I install MLflow? A: pip install mlflow then mlflow server. Or uvx mlflow server without installing. Access UI at localhost:5000.


🙏

Source et remerciements

Created by Databricks. Licensed under Apache 2.0. mlflow/mlflow — 25,000+ GitHub stars

Discussion

Connectez-vous pour rejoindre la discussion.
Aucun commentaire pour l'instant. Soyez le premier à partager votre avis.

Actifs similaires