Databricks · siet Erste Alpha-Version: Juni 2018 (Databricks); aktuelle Hauptversion MLflow 3, aktuellstes Minor-Release 3.14.0 (2026) · 2× · tolest 29. Juni 2026
12
Momentum
MLflow is an open-source AI engineering platform originally created by Databricks in 2018, now stewarded under the Linux Foundation. It covers experiment tracking, model registry and deployment for classical ML, plus observability (tracing), LLM-judge-based evaluation, and prompt management for LLM applications and agents. Built on OpenTelemetry, it supports any LLM provider or agent framework and is available both self-hosted and as a Databricks-managed cloud offering ("Managed MLflow"). The core software is fully free under the Apache 2.0 license.
Momentum-Verloop
16.05.14.08.
Features
| Deployment (Self-Hosted/Cloud) | Self-hosted (including official Helm chart for Kubernetes) or as Managed MLflow cloud service on Databricks |
| License | 100% open source under Apache 2.0 license, forever free |
| Open-Source Option | Yes – 100% open source under Apache 2.0 license, permanently free. Under Linux Foundation governance; over 900 community contributors on GitHub. |
| Platform | Local, on-premises, cloud, Kubernetes, and as managed service on Databricks, AWS SageMaker, Azure |
| Price | Open-source core is free; Managed MLflow on Databricks billed by usage (compute/storage) |
| Pricing Model | Open source (Apache 2.0) – free and royalty-free. Managed service on Databricks paid (Databricks platform pricing); Databricks Free Edition available for getting started. |
| Protocol Compatibility | Tracing natively built on OpenTelemetry, plus MCP support |
| Release Date | First released June 2018 (alpha) by Databricks; MLflow 3 since June 2025; MLflow 3.14.0 latest version |
| Supported Models/Providers | Any LLM provider and agent framework; 100+ integrations; SDKs for Python, TypeScript/JavaScript, Java, R |