Ray provides a small set of distributed computing primitives that allow developers to scale Python applications across heterogeneous clusters of compute. Ray also includes an ecosystem of higher-level libraries for scaling multimodal data processing, model training, model serving, reinforcement learning, and more.
Ray is used by Python developers and ML / AI practitioners and researchers who are building scalable Python and AI applications. Ray is also adopted by platform and infrastructure engineers who integrate it into their broader infrastructure.
Ray was contributed by Anyscale to the Linux Foundation in September 2025.
