AReaL
Apache-2.0 fully asynchronous reinforcement-learning framework for large reasoning models, with a public Python package and docs.
Tags
Description
Open RL training system (areal-project/AReaL; not archived). GitHub and PyPI license Apache-2.0. PyPI package areal 1.0.4 (uploaded 2026-06-10, not yanked; currently the only PyPI release) verified 2026-11-12. The GitHub repository's latest release is v2.1.1 (published 2026-11-04, not a draft or prerelease). No 2.x release is published on PyPI; the PyPI package lags the GitHub releases. This is an established research/training framework, not presented as a brand-new hosted product launch. GPU clusters, model weights and any third-party rollouts are separate. The maintained English docs at https://areal-ai.io/docs/en/intro.html return HTTP 200 directly, with no redirect (re-checked 2026-11-12). The documentation URL listed on PyPI for areal 1.0.4, https://areal-project.github.io/AReaL/en/intro.html, responds 301 to http://areal-ai.io/AReaL/en/intro.html, and that Location currently returns HTTP 500.
Agent access
Related
- verl: Distributed reinforcement-learning training library for LLMs, with GRPO/PPO dataflows and pluggable training and rollout engines for agentic post-training.
- SkyRL: Modular reinforcement-learning stack with a training engine, long-horizon agent layer and Gymnasium-style tool-use environments for LLMs.
- rLLM: Reinforcement-learning framework that reuses agent harnesses for evaluation and training, collecting rollouts and rewards across sandbox backends.
- OpenRLHF: Apache-2.0 high-performance RLHF / post-training framework built on Ray, with support for multiple reward and policy-optimization recipes.
Guides & comparisons
- Post-training a tool-use model — TRL, PEFT, and scale-out RL (Stack): A recommended post-training stack for chat and tool-use models: TRL and PEFT for SFT, LoRA/QLoRA, DPO and GRPO, with OpenRLHF and AReaL as the distributed and asynchronous RL options. Cross-links the fine-tuning frameworks and agent RL stacks comparisons.
Sources
- https://areal-ai.io
- https://github.com/areal-project/AReaL
- https://github.com/areal-project/AReaL/blob/main/LICENSE
- https://pypi.org/project/areal/
- https://github.com/areal-project/AReaL/releases/tag/v2.1.1
- https://areal-ai.io/docs/en/intro.html
Machine-readable
- JSON
- Markdown
- Source file on GitHub (edit via pull request)