Anyscale
Paid
The managed platform from the creators of Ray for scaling AI training and inference.
About MLflow
MLflow is an open-source platform that covers the life of a machine learning project. It tracks experiments, packages models, stores them in a registry and helps deploy them, all with a consistent interface.
In recent years it has added support for generative AI, including tracing for LLM apps and agents, prompt versioning and evaluation. It works with nearly every ML library and cloud, and Databricks offers a managed version.
MLflow is free to run yourself. Its wide adoption makes it a safe, vendor-neutral choice for teams building ML systems.
Standout features
- Experiment tracking
- Model registry
- Model packaging and serving
- GenAI tracing
- Evaluation tools
- Open source
Tags
More in MLOps
View all →Comet
Freemium
Experiment tracking and the open-source Opik platform for evaluating LLM apps.
Modal
Freemium
Serverless cloud for running AI workloads, with GPUs available in seconds.
Replicate
Paid
Run open-source and custom AI models in the cloud with a few lines of code.
Weights and Biases
Freemium
Tools for tracking experiments, managing models and evaluating LLM apps.
Hugging Face
Freemium
The main hub for open models, datasets and demos, plus tools to train and deploy them.
