New feature
GA
Managed MLflow
Managed MLflow is a Databricks ai / ml capability, introduced February 2019.
The fully managed, hosted version of open source MLflow on Databricks, built on Unity Catalog: experiment tracking, model evaluation, a production model registry, and deployment tools for ML models, plus tracing, evaluation, and prompt management for agents and LLM apps, all wired into the workspace security model.
- It went GA with two names in one breath: the April 2019 release note is headlined "MLflow on Databricks (GA)" while its first sentence announces that "Managed MLflow on Databricks is now generally available" - and the docs still use both.
- The Public Preview shipped as part of Databricks platform Version 2.92, and its headline trick was that logging a run from a notebook made Databricks quietly snapshot the notebook revision so the exact code could be replayed later.
- The docs now introduce MLflow as the largest open source AI engineering platform for agents, LLMs, and ML models, with over 30 million monthly downloads - quite a promotion for a tool that started out logging parameters and metrics.
- Category
- AI / ML
- Also known as
- MLflow on Databricks, Databricks-managed MLflow, hosted MLflow, MLflow 3 on Databricks
- Verified
- 2026-09-27
Sources
- Official Official Databricks / Microsoft docs
- Official Databricks release notes: Managed MLflow on Databricks Public Preview (February 26 - March 5, 2019)
- Official Databricks release notes: MLflow on Databricks GA (April 25, 2019)
- Official Databricks release notes: MLflow 3.0 is generally available (June 10, 2025) - Logged Models and deployment jobs
- Official Databricks docs: MLflow on Databricks - MLflow 3, and open source vs. Databricks-managed MLflow features