Registry / ai-ml / jax-cuda13-plugin

jax-cuda13-plugin

JSON →
library0.10.0pypypi✓ verified 83d ago

JAX plugin providing NVIDIA GPU support for CUDA 13.x. Version 0.10.0 is compatible with JAX v0.10.0. Release cadence follows JAX releases.

ai-ml
pip install jax-cuda13-plugin
github ↗package ↗homepage ↗
INSTALL
IMPORT
SIG · JAX-CUDA13-PLUGIN
J
jax-cuda13-plugin
ai-mlpythonv0.10.0
harness data pending
Install & Compatibility
Where this runs

No compatibility data collected yet for this library.

Code
Verified usage

Verified import paths — ran on the pinned version, not inferred.

jax_cuda13_plugin
✓ import jax_cuda13_plugin
Plugin is imported as a side effect; no top-level API.

Verify GPU device is detected after installing plugin.

import jax import jax.numpy as jnp x = jnp.array([1., 2., 3.]) print(jax.devices())
Debug
Known issues
breakingPlugin version must exactly match JAX major.minor version (e.g., 0.10.x). Mismatch causes import errors.
fix
Install jax-cuda13-plugin==0.10.0 alongside jax==0.10.0.
affects: all
gotchajax-cuda13-plugin only supports CUDA 13.x; it will not work with other CUDA versions. Ensure your system has CUDA 13 runtime.
fix
Verify CUDA version with `nvidia-smi`; install appropriate plugin (e.g., jax-cuda12-plugin for CUDA 12).
affects: all
gotchaImporting jax_cuda13_plugin is not required; merely installing it enables GPU support. The import statement is a no-op and may cause confusion.
fix
Do not import plugin explicitly; just install and import jax.
affects: all
Errors
Common errors & fixes
ImportError: libcuda.so.1: cannot open shared object file: No such file or directory
CUDA runtime library not found; plugin requires CUDA 13.x installed.
fix
Install CUDA 13.x from NVIDIA and ensure LD_LIBRARY_PATH includes /usr/local/cuda-13/lib64.
jax._src.plugins.PluginNotInstalledError: jax_cuda13_plugin not installed
Plugin package not installed or version mismatch with JAX.
fix
Run `pip install jax-cuda13-plugin==0.10.0` matching your JAX version.
RuntimeError: Failed to initialize XLA: CUDA driver version is insufficient for CUDA runtime version
NVIDIA driver is too old for CUDA 13.
fix
Update NVIDIA driver to version that supports CUDA 13 (e.g., driver >= 545).
Upgrade
Version history
0.10.0latest on PyPI · released Apr 16, 2026
Audit
Dependencies
jaxrequiredRequired; plugin provides CUDA support for JAX.
jaxlibrequiredRequired; JAX binary library.
Agent activity
8 hits · last 30 days
node
8
Resources
githubhttps://github.com/jax-ml/jax ↗
packagehttps://pypi.org/project/jax-cuda13-plugin/ ↗
homepagehttps://jax.readthedocs.io ↗
jax-cuda13-plugin — pip install jax-cuda13-plugin · libregistry