google/brax
A fast, fully differentiable physics engine written in JAX for accelerating reinforcement learning and robotics research.
Not currently ranked — collecting fresh signals.
star history
Brax provides massively parallel rigidbody physics simulation on accelerator hardware (TPU/GPU). Written entirely in JAX, it enables efficient single-device and distributed simulation without datacenter requirements. The library includes baseline RL algorithms (PPO, SAC, ARS) that can train agents in seconds to minutes, and is specifically designed for robotics, human perception, and reinforcement learning research applications.
Frequently asked
- What is google/brax?
- A fast, fully differentiable physics engine written in JAX for accelerating reinforcement learning and robotics research.
- Is brax open source?
- Yes — google/brax is open source, released under the Apache-2.0 license.
- What language is brax written in?
- google/brax is primarily written in Jupyter Notebook.
- How popular is brax?
- google/brax has 3.2k stars on GitHub.
- Where can I find brax?
- google/brax is on GitHub at https://github.com/google/brax.