cuda-pathfinder
Pathfinder for CUDA components
Decision gist · record as of 2026-08-14
Yes. If your project needs to programmatically locate CUDA components across different systems and CUDA versions, this is the official NVIDIA solution for that problem. Low friction, no dependencies, permissive license, and active maintenance make it a straightforward addition. Not relevant if CUDA paths are already known or hardcoded.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Python 3.10 or later; CUDA Toolkit installation on the system is expected for meaningful operation.
- Low install friction with no runtime dependencies.
- Actively maintained by NVIDIA with recent commits and a stable release cadence supporting the two most recent CUDA Toolkit major versions.
License · maintenance · safety
Apache-2.0 (permissive) — Apache-2.0 permissive license allows commercial and private use with minimal restrictions.
last release 2026-07-21 (24 days) · last repo commit 2026-08-14 · 3,341 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 52,035,930 downloads/mo, #564 on PyPI
Alternatives
Verify before relying
pip install cuda-pathfinder
import cuda.pathfinder
# Locate CUDA libraries and headers on the system- Whether the package can successfully locate libraries on all major operating systems (Windows, Linux, macOS).
- Performance characteristics when searching across multiple CUDA Toolkit installations.
- Exact scope of 'other artifacts' mentioned as in-progress support.
What it is and what it does
cuda-pathfinder is a utility library from NVIDIA for programmatically discovering CUDA components installed on a system. It abstracts away the platform-specific and version-specific logic needed to find CUDA dynamic libraries (.so, .dll files) and CTK header directories, returning paths that can be used by downstream code to load or compile against CUDA.
The package is version-agnostic with respect to the CUDA Toolkit itself, following NVIDIA's support policy of maintaining compatibility with the two most recent major CUDA versions. It has no runtime dependencies and installs as a pure Python wheel, making it lightweight to add to projects that need to dynamically discover CUDA on end-user systems.
Use it for
- Build tools and installers that need to detect CUDA availability and version at runtime before compiling extensions.
- Python bindings and wrappers that load CUDA libraries dynamically without hardcoding paths.
- CI/CD pipelines testing against multiple CUDA Toolkit versions installed on the same machine.
- GPU-accelerated libraries that need to locate and link CUDA headers during package installation.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes.
If your project needs to programmatically locate CUDA components across different systems and CUDA versions, this is the official NVIDIA solution for that problem. Low friction, no dependencies, permissive license, and active maintenance make it a straightforward addition. Not relevant if CUDA paths are already known or hardcoded.
Install
cuda-pathfinder on PyPI
Before you install
Low install friction with no runtime dependencies. Actively maintained by NVIDIA with recent commits and a stable release cadence supporting the two most recent CUDA Toolkit major versions.
Requires Python 3.10 or later; CUDA Toolkit installation on the system is expected for meaningful operation.
License in practice
Apache-2.0 permissive license allows commercial and private use with minimal restrictions.
Quickstart
pip install cuda-pathfinder
import cuda.pathfinder
# Locate CUDA libraries and headers on the systemVerify before relying
- Whether the package can successfully locate libraries on all major operating systems (Windows, Linux, macOS).
- Performance characteristics when searching across multiple CUDA Toolkit installations.
- Exact scope of 'other artifacts' mentioned as in-progress support.
Package facts
| License | Apache-2.0 permissive |
| Python support | Supports the current Python release >=3.10 |
| Install friction | Low. Pure-Python wheel |
| Runtime dependencies | None |
| Maintenance | Actively maintained 24 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 52,035,930 / month, #564 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
Evidence: cuda_pathfinder-1.6.0-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 › “cuda library path finder”
- cuda-pathfinderLocates and loads CUDA dynamic libraries and finds CTK header…
- findlibsLocates shared libraries on the system by searching Python packages,…
- findpythonLocates Python interpreters on your system by version, architecture,…
Give your agent the search over MCP, or paste the wish link into any chat.
Similar packages
Provides low-level Python bindings for CUDA host APIs, offering 1:1 access to CUDA functionality from Python code.
cuda-python is a metapackage providing Pythonic access to NVIDIA's CUDA platform, bundling low-level bindings, runtime abstractions, and utilities for GPU-accelerated computing from Python.
However, verify that the proprietary NVIDIA license aligns with your project's requirements before committing to it in production or redistributed code.
Provides Pythonic APIs to access CUDA core functionalities including runtime control, compiler, and linker operations from Python.
A meta-package that installs NVIDIA CUDA Toolkit components via pip, allowing selective installation of GPU-accelerated libraries, compilers, and runtime tools through package extras.
However, verify license terms beforehand and confirm your GPU and driver support the toolkit version—this is not a substitute for proper CUDA environment setup on…
Provides NVIDIA's CUDA C++ Core Compute Libraries (CCCL) for GPU-accelerated computing on CUDA 12 platforms.
Provides NVIDIA CUDA 12 runtime native libraries for GPU-accelerated computing on Linux, Windows, and ARM64 systems.
See also nvidia-cuda-runtime · nvidia-cuda-runtime-cu11 · nvidia-cuda-nvcc · findpython