nvidia-cuda-nvcc-cu12

What it is and what it does

This package distributes the NVIDIA CUDA nvcc compiler as a Python wheel, making it installable via pip on Linux and Windows systems. It is a thin wrapper around the CUDA compiler toolchain, allowing developers to compile CUDA C/C++ code as part of their Python-based build workflows. The package has no runtime Python dependencies and serves primarily as a convenient distribution mechanism for the nvcc binary.

The package targets machine learning, deep learning, and scientific computing workflows where developers need to compile custom CUDA kernels. It supports Python 3.5 and later, though the actual compilation happens outside the Python runtime. The aging maintenance status (435 days since last release) suggests this may be a stable, infrequently-updated distribution rather than an actively developed project.

Use it for:

• Compile custom CUDA kernels as part of a Python package build pipeline.
• Set up a reproducible CUDA development environment in CI/CD workflows.
• Build GPU-accelerated libraries that require custom CUDA code compilation.
• Develop machine learning applications with hand-optimized CUDA kernels.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Provides the NVIDIA CUDA nvcc compiler for building CUDA applications, packaged as a Python distribution for easy installation across Linux and Windows platforms.

Yes, if you need to compile CUDA code in a Python environment and are comfortable with proprietary licensing. The package is stable, widely downloaded, and has no known vulnerabilities. However, verify that the proprietary license terms fit your use case, and confirm that CUDA 12.9.86 matches your GPU and toolkit requirements before committing to this version.

Install

nvidia-cuda-nvcc-cu12 on PyPI

pip install nvidia-cuda-nvcc-cu12

uv add nvidia-cuda-nvcc-cu12

poetry add nvidia-cuda-nvcc-cu12

Installing nvidia-cuda-nvcc-cu12

Before you install

Medium install friction due to platform-specific wheel distributions (manylinux2010, manylinux2014, Windows). Package is aging—last release was 435 days ago—but remains in the top 5000 by downloads with no known vulnerabilities.

License in practice

Licensed under a proprietary NVIDIA license (LicenseRef-NVIDIA-Proprietary) with unclear treatment. Users should review NVIDIA's licensing terms before deploying in production or commercial contexts.

Quickstart

pip install nvidia-cuda-nvcc-cu12==12.9.86

import subprocess
result = subprocess.run(['nvcc', '--version'], capture_output=True, text=True)
print(result.stdout)

Requires a compatible x86_64 or aarch64 Linux system (manylinux2010/2014) or Windows x64; the nvcc compiler itself is a system tool, not a Python library.

Verify before relying

• Whether this package includes the full CUDA toolkit or only the nvcc compiler binary.
• Compatibility with specific CUDA compute capabilities and GPU architectures.
• Whether updates to CUDA 12.9.86 are still released or if this version is end-of-life.

Package facts

Evidence: nvidia_cuda_nvcc_cu12-12.9.86-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl; nvidia_cuda_nvcc_cu12-12.9.86-py3-none-manylinux2014_aarch64.manylinux_2_17_aarch64.whl; nvidia_cuda_nvcc_cu12-12.9.86-py3-none-win_amd64.whl

Keywords: cuda, nvidia, runtime, machine learning, deep learning

Tags

More Software Development packages