---
id: nvidia-cuda-crt
version: "13.3.73"
license: unclear
license_treatment: unclear
maintenance: active
---

# nvidia-cuda-crt — CUDA C Runtime

License: unclear · Maintenance: active · Downloads: 2.1M/mo

## What it is and what it does

nvidia-cuda-crt packages the CUDA C Runtime libraries from NVIDIA for use in Python environments. It is a low-level runtime dependency typically installed as part of a larger GPU-accelerated software stack rather than used directly by application code. The package targets developers and researchers working with machine learning, deep learning, and scientific computing who need GPU acceleration.

The package supports Python 3.5 through 3.11 and is distributed as platform-specific wheels for Windows, Linux x86_64, and Linux aarch64. It has no Python runtime dependencies of its own, functioning as a pure library distribution. Installation requires selecting the correct wheel for your platform and architecture.

Use it for:

- Providing CUDA runtime dependencies for deep learning frameworks in containerized or virtual environments.
- Setting up GPU-accelerated scientific computing environments where lower-level CUDA libraries are needed.
- Packaging GPU applications that depend on CUDA C runtime without requiring users to install NVIDIA CUDA Toolkit separately.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Provides NVIDIA CUDA C Runtime libraries for Python, enabling GPU-accelerated compute on supported platforms.

Yes, if you need CUDA runtime support for GPU-accelerated Python applications and your platform matches the available wheels. The package is actively maintained with no known vulnerabilities. Install only if you have a specific dependency requirement from another package or a documented need for CUDA 13.3 runtime support; it is not typically installed standalone.

## Install

pip install nvidia-cuda-crt
uv add nvidia-cuda-crt
poetry add nvidia-cuda-crt

## Installing nvidia-cuda-crt

Before you install: Medium install friction due to platform-specific wheels for Windows, Linux x86_64, and Linux aarch64. Package is actively maintained with recent release activity.

Quickstart:

pip install nvidia-cuda-crt==13.3.73

# Typically imported indirectly as a dependency of other packages
# rather than used directly in application code.

Requires a platform matching one of the available wheels: Windows AMD64, Linux x86_64 (manylinux2014), or Linux aarch64 (manylinux2014).

Verify before relying:

- Whether this package requires a separate NVIDIA CUDA Toolkit installation or provides complete runtime independently.
- Whether installation succeeds on systems without NVIDIA GPUs or CUDA drivers present.
- Compatibility with specific CUDA versions beyond what version 13.3.73 implies.

## Package facts

- License: not declared (unclear)
- Python support: supports_current
- Install friction: medium
- Maintenance: active
- Downloads: 2.1M/month (top 5,000 on PyPI)
- Known vulnerabilities: none known

## Tags

cuda runtime libraries python, nvidia gpu acceleration, cuda c runtime, gpu compute environment, nvidia cuda python bindings, gpu-acceleration, cuda

[View on SkillFed](https://skillfed.io/packages/nvidia-cuda-crt) · [View on PyPI](https://pypi.org/project/nvidia-cuda-crt/)