Implementation:Google deepmind Mujoco MjModel from xml path

Knowledge Sources	MuJoCo Python API MuJoCo
Domains	Physics_Simulation, Python_API, Model_IO
Last Updated	2026-02-15 06:00 GMT

Overview

Wrapper for loading MuJoCo models via the official Python bindings package used as the CPU-side entry point for MJX workflows.

Description

The mujoco.MjModel.from_xml_path class method loads an MJCF XML model file and returns a Python MjModel object. This wraps the C mj_loadXML function and provides Pythonic attribute access to all model fields. This is the standard way to load models in MJX pipelines before GPU transfer.

Usage

Call as the first step in MJX workflows. The returned MjModel is passed to mjx.put_model for GPU transfer.

External Reference

• MuJoCo Python Bindings Documentation

Code Reference

Source Location

• Repository: External (mujoco Python bindings pip package)
• File: Part of the mujoco pip package (not in this repository)

Signature

@classmethod
def MjModel.from_xml_path(path: str) -> mujoco.MjModel

Import

import mujoco

I/O Contract

Inputs

Name	Type	Required	Description
path	str	Yes	Filesystem path to MJCF XML model file

Outputs

Name	Type	Description
return	mujoco.MjModel	CPU-resident model object wrapping C mjModel struct

Usage Examples

import mujoco
import mujoco.mjx as mjx

# Load model on CPU
model = mujoco.MjModel.from_xml_path("humanoid.xml")

# Transfer to GPU for MJX
mx = mjx.put_model(model)

Related Pages

Implements Principle

• Principle:Google_deepmind_Mujoco_CPU_Model_Loading

Requires Environment

• Environment:Google_deepmind_Mujoco_Python_Bindings_Environment