mujoco-mjx
MuJoCo XLA (MJX)
Decision gist · record as of 2026-08-14
Yes, if you are doing machine learning with physics simulation and have access to GPU/TPU hardware. MJX is actively maintained by DeepMind, has no known vulnerabilities, and is permissively licensed. Install only if you need JAX integration and can tolerate that some MuJoCo features are not yet supported; if you need full MuJoCo compatibility, use the base mujoco package instead.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Python 3.10 or later.
- JAX and jaxlib must be installed and configured for your target hardware (CPU, GPU, or TPU).
- Low friction install with a pure-Python wheel.
License · maintenance · safety
Apache-2.0 (permissive) — Licensed under Apache-2.0 (permissive), allowing commercial and private use with attribution. No restrictions on modification or distribution.
last release 2026-07-28 (17 days) · last repo commit 2026-08-14 · 14,550 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 99,309 downloads/mo, #13,029 on PyPI
Alternatives
Verify before relying
pip install mujoco-mjx
from mujoco import mjx
# Load and step a physics model on accelerator
model = mjx.Model.from_xml(xml_string)
data = mjx.Data(model)- Feature parity with MuJoCo: documentation notes some features are missing; specifics of what is unsupported are not in the fact sheet.
- Performance characteristics: whether acceleration actually improves simulation speed depends on workload and hardware; not quantified in the fact sheet.
- Batch simulation capability: whether the package supports vectorized/batched physics steps is implied but not explicitly confirmed.
What it is and what it does
MJX is a physics simulation library that reimplements MuJoCo using JAX, enabling physics computations to run on GPUs and TPUs. It maintains API compatibility with MuJoCo but is not a drop-in replacement—some features are missing. The package is developed and maintained by Google DeepMind and is kept synchronized with MuJoCo releases.
You use MJX when you need differentiable physics for reinforcement learning, want to exploit accelerator hardware for batch simulations, or need to integrate physics into JAX-based machine learning pipelines. It depends on JAX, jaxlib, the base mujoco package, scipy, trimesh, absl-py, and etils.
Use it for
- Train reinforcement learning policies with differentiable physics gradients flowing through the simulation.
- Run batched physics simulations in parallel on GPU/TPU for robotics research or control optimization.
- Integrate physics simulation into JAX-based neural network training pipelines.
- Prototype physics-based learning algorithms that require automatic differentiation through dynamics.
- Accelerate large-scale multi-environment simulations for population-based training.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you are doing machine learning with physics simulation and have access to GPU/TPU hardware.
MJX is actively maintained by DeepMind, has no known vulnerabilities, and is permissively licensed. Install only if you need JAX integration and can tolerate that some MuJoCo features are not yet supported; if you need full MuJoCo compatibility, use the base mujoco package instead.
Install
mujoco-mjx on PyPI
Before you install
Low friction install with a pure-Python wheel. Actively maintained by Google DeepMind with a recent release (17 days old) and high repository activity (14550 stars). Requires modern Python (3.10+) and brings in substantial dependencies including JAX, jaxlib, and the base mujoco package.
Requires Python 3.10 or later. JAX and jaxlib must be installed and configured for your target hardware (CPU, GPU, or TPU).
License in practice
Licensed under Apache-2.0 (permissive), allowing commercial and private use with attribution. No restrictions on modification or distribution.
Quickstart
pip install mujoco-mjx
from mujoco import mjx
# Load and step a physics model on accelerator
model = mjx.Model.from_xml(xml_string)
data = mjx.Data(model)Verify before relying
- Feature parity with MuJoCo: documentation notes some features are missing; specifics of what is unsupported are not in the fact sheet.
- Performance characteristics: whether acceleration actually improves simulation speed depends on workload and hardware; not quantified in the fact sheet.
- Batch simulation capability: whether the package supports vectorized/batched physics steps is implied but not explicitly confirmed.
Package facts
| License | Apache-2.0 permissive |
| Python support | Supports the current Python release >=3.10 |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | 7 packagesabsl-pyetilsjaxjaxlibmujocoscipytrimesh |
| Maintenance | Actively maintained 17 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 99,309 / month, #13,029 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Development Status :: 5 - Production/StableIntended Audience :: DevelopersIntended Audience :: Science/ResearchNatural Language :: EnglishProgramming Language :: Python :: 3Programming Language :: Python :: 3.10Programming Language :: Python :: 3.11Programming Language :: Python :: 3.12Programming Language :: Python :: 3.13Programming Language :: Python :: 3.14Topic :: Scientific/Engineering |
Evidence: mujoco_mjx-3.11.0-py3-none-any.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “jax physics simulation”
- mujoco-mjxA JAX-based reimplementation of the MuJoCo physics engine that runs…
- mujoco-warpGPU-accelerated physics simulation for robotics using NVIDIA Warp,…
- warp-langWarp is a Python framework that JIT-compiles regular Python functions…
Give your agent the search over MCP, or paste the wish link into any chat.
More Scientific/Engineering packages
NumPy provides an N-dimensional array object and a comprehensive suite of mathematical, linear algebra, Fourier transform, and random number functions for scientific computing in Python.
pandas provides fast, flexible data structures (Series and DataFrame) for loading, cleaning, transforming, and analyzing labeled or relational data in Python.
scipy provides numerical algorithms for mathematics, science, and engineering—including optimization, integration, linear algebra, Fourier transforms, signal and image processing, and ODE solvers—built on numpy arrays.
scikit-learn provides a comprehensive Python library for supervised and unsupervised machine learning, including classification, regression, clustering, dimensionality reduction, and model evaluation tools built on NumPy and SciPy.
Install it if you need to train, evaluate, or deploy supervised or unsupervised learning models.
dill extends Python's pickle module to serialize and deserialize a much wider range of Python objects, including functions, lambdas, classes, and interpreter sessions, to byte streams for storage or network transmission.
Multiprocess is an enhanced fork of Python's standard multiprocessing library that uses dill for better serialization, allowing you to spawn processes with a threading-like API and share complex objects between them.
Install it if you use multiprocessing and encounter pickle serialization limits with lambdas or complex objects.
See also dm-control · mujoco · mujoco-warp · mjviser · mjlab · mujoco-usd-converter · newton · chex · jax-cuda13-pjrt · jax-cuda12-pjrt