Ray: Distributed AI compute framework
Unified framework for scaling AI and Python applications from laptops to clusters with distributed runtime.
Learn more about Ray
Ray is a distributed computing framework designed for scaling Python applications and AI workloads across multiple machines. The framework consists of a core distributed runtime that manages tasks, actors, and objects across a cluster, along with specialized libraries for data processing, training, hyperparameter tuning, reinforcement learning, and model serving. Ray uses a distributed object store and task scheduler to coordinate work across nodes, with support for both stateless functions (tasks) and stateful processes (actors). The framework is commonly used for distributed machine learning training, hyperparameter optimization, reinforcement learning experiments, and serving large language models at scale.
Unified Runtime
Single framework handles diverse workloads from data processing to model serving. Core abstractions of tasks, actors, and objects work consistently across all AI libraries.
Seamless Scaling
Same Python code runs unchanged from laptop to multi-node clusters. Built-in distributed object store and scheduler handle resource management automatically.
Specialized Libraries
Includes purpose-built libraries for data processing, distributed training, hyperparameter tuning, reinforcement learning, and model serving. Each library integrates with the core runtime for optimal performance.
Ray-2.53.0
- –Ray plans to drop support for Pydantic V1 starting version 2.56.0. Please see this RFC for details.
- –Ray Data now has support for bounded reading from Kafka and improved Iceberg support.
- –Autoscaling**: New utilization-based cluster autoscaler for Ray Data workloads (#59353, #59362, #59366). To use this new autoscaler set RAY\DATA\CLUSTER\_AUTOSCALER=V2.
- –Kafka Datasource**: Add Kafka as a native datasource for data ingestion (#58592)
- –Dataset summary API**: Add `Dataset.summary()` API for quick dataset inspection (#58862)
Ray-2.51.2
- –Fix for CVE-2025-62593: reject Sec-Fetch-* other browser-specific headers in dashboard browser rejection logic
Ray-2.52.1
- –More robust handling for CVE-2025-62593: test for more browser-specific headers in dashboard browser rejection logic
See how people are using Ray
Top in AI & ML
Related Repositories
Discover similar tools and frameworks used by developers
langchain
Modular framework for chaining LLMs with external data.
Chat SDK
Open-source Next.js template for AI chatbots with auth, persistence, and multi-provider support.
tesseract
LSTM-based OCR engine supporting 100+ languages.
StabilityMatrix
Multi-backend inference UI manager with embedded dependencies.
vllm
Fast, memory-efficient LLM inference engine with PagedAttention for production deployments at scale.