Ray version 2.0 is here!

:boom::tada: Ray version 2.0 is here!


  • Ray AIR, a scalable and unified toolkit for ML applications, is now in Beta.
  • ​​Ray now supports natively shuffling 100TB or more of data with the Ray Datasets library.
  • KubeRay, a toolkit for running Ray on Kubernetes, is now in Beta. This replaces the legacy Python-based Ray operator.
  • Ray Serve’s Deployment Graph API is a new and easier way to build, test, and deploy an inference graph of deployments. This is released as Beta in 2.0.

And there’s more! Head over to the release blog for the deep dive.


Below is an aggregated list of different migration guides for each Ray component.

Core APIs

Ray Core - Ray Core 2.0 Changes

Library APIs

Ray RLlib - Ray RLlib 2.0 Changes
Ray Tune - Ray Tune 2.0 Changes
Ray Train - Ray Train 2.0 Changes
Ray Serve - Ray Serve 2.0 Migration Guide
XGBoost Integrations -Ray XGBoost Integrations 2.0 Changes
Misc Util - Ray Util 2.0 Changes

Cluster APIs

Ray Autoscaler - Ray Autoscaler 2.0 Changes
KubeRay - KubeRay 0.3.0 Migration Guide

1 Like