| PyTorch | Accelerator | Python | OS / Arch | Requirements | Install |
|---|
PyTorch versions at a glance
CUDA, ROCm and Python versions with official wheels for each stable PyTorch release.
Show all 77 releases
| PyTorch | CUDA | ROCm | Intel XPU | Python |
|---|---|---|---|---|
| 2.14.1 | 12.6, 13.0, 13.2 | 7.2, 7.14 | yes | 3.10, 3.11, 3.12, 3.13, 3.14, 3.15 |
| 2.14.0 | 12.6, 13.0, 13.2 | 7.2, 7.14 | yes | 3.10, 3.11, 3.12, 3.13, 3.14, 3.15 |
| 2.13.0 | 12.6, 12.9, 13.0, 13.2 | 7.1, 7.2 | yes | 3.10, 3.11, 3.12, 3.13, 3.14, 3.15 |
| 2.12.1 | 12.6, 12.9, 13.0, 13.2 | 7.1, 7.2 | yes | 3.10, 3.11, 3.12, 3.13, 3.14 |
| 2.12.0 | 12.6, 13.0, 13.2 | 7.1, 7.2 | yes | 3.10, 3.11, 3.12, 3.13, 3.14 |
| 2.11.0 | 12.6, 12.8, 12.9, 13.0 | 7.1, 7.2 | yes | 3.10, 3.11, 3.12, 3.13, 3.14 |
| 2.10.0 | 12.6, 12.8, 12.9, 13.0 | 7.0, 7.1 | yes | 3.10, 3.11, 3.12, 3.13, 3.14 |
| 2.9.1 | 12.6, 12.8, 12.9, 13.0 | 6.3, 6.4 | yes | 3.10, 3.11, 3.12, 3.13, 3.14 |
| 2.9.0 | 12.6, 12.8, 12.9, 13.0 | 6.3, 6.4 | yes | 3.10, 3.11, 3.12, 3.13, 3.14 |
| 2.8.0 | 12.6, 12.8, 12.9 | 6.3, 6.4 | yes | 3.9, 3.10, 3.11, 3.12, 3.13 |
| 2.7.1 | 11.8, 12.6, 12.8 | 6.2.4, 6.3 | yes | 3.9, 3.10, 3.11, 3.12, 3.13 |
| 2.7.0 | 11.8, 12.6, 12.8 | 6.2.4, 6.3 | yes | 3.9, 3.10, 3.11, 3.12, 3.13 |
| 2.6.0 | 11.8, 12.4, 12.6 | 6.1, 6.2.4 | yes | 3.9, 3.10, 3.11, 3.12, 3.13 |
| 2.5.1 | 11.8, 12.1, 12.4 | 6.1, 6.2 | — | 3.9, 3.10, 3.11, 3.12, 3.13 |
| 2.5.0 | 11.8, 12.1, 12.4 | 6.1, 6.2 | — | 3.9, 3.10, 3.11, 3.12, 3.13 |
| 2.4.1 | 11.8, 12.1, 12.4 | 6.0, 6.1 | — | 3.8, 3.9, 3.10, 3.11, 3.12 |
| 2.4.0 | 11.8, 12.1, 12.4 | 6.0, 6.1 | — | 3.8, 3.9, 3.10, 3.11, 3.12 |
| 2.3.1 | 11.8, 12.1 | 5.7, 6.0 | — | 3.8, 3.9, 3.10, 3.11, 3.12 |
| 2.3.0 | 11.8, 12.1 | 5.7, 6.0 | — | 3.8, 3.9, 3.10, 3.11, 3.12 |
| 2.2.2 | 11.8, 12.1 | 5.6, 5.7 | — | 3.8, 3.9, 3.10, 3.11, 3.12 |
| 2.2.1 | 11.8, 12.1 | 5.6, 5.7 | — | 3.8, 3.9, 3.10, 3.11, 3.12 |
| 2.2.0 | 11.8, 12.1 | 5.6, 5.7 | — | 3.8, 3.9, 3.10, 3.11, 3.12 |
| 2.1.2 | 11.8, 12.1 | 5.5, 5.6 | — | 3.8, 3.9, 3.10, 3.11 |
| 2.1.1 | 11.8, 12.1 | 5.5, 5.6 | — | 3.8, 3.9, 3.10, 3.11 |
| 2.1.0 | 11.8, 12.1 | 5.5, 5.6 | — | 3.8, 3.9, 3.10, 3.11 |
| 2.0.1 | 11.7, 11.8 | 5.3, 5.4.2 | — | 3.8, 3.9, 3.10, 3.11 |
| 2.0.0 | 11.7, 11.8 | 5.3, 5.4.2 | — | 3.8, 3.9, 3.10, 3.11 |
| 1.13.1 | 11.6, 11.7 | 5.1.1, 5.2 | — | 3.7, 3.8, 3.9, 3.10, 3.11 |
| 1.13.0 | 11.6, 11.7 | 5.1.1, 5.2 | — | 3.7, 3.8, 3.9, 3.10, 3.11 |
| 1.12.1 | 10.2, 11.3, 11.6 | 5.0, 5.1.1 | — | 3.7, 3.8, 3.9, 3.10 |
| 1.12.0 | 10.2, 11.3, 11.6 | 5.0, 5.1.1 | — | 3.7, 3.8, 3.9, 3.10 |
| 1.11.0 | 10.2, 11.3, 11.5 | 4.3.1, 4.5.2 | — | 3.7, 3.8, 3.9, 3.10 |
| 1.10.2 | 10.2, 11.1, 11.3 | 4.0.1, 4.1, 4.2 | — | 3.6, 3.7, 3.8, 3.9, 3.10 |
| 1.10.1 | 10.2, 11.1, 11.3 | 4.0.1, 4.1, 4.2 | — | 3.6, 3.7, 3.8, 3.9 |
| 1.10.0 | 10.2, 11.1, 11.3 | 4.0.1, 4.1, 4.2 | — | 3.6, 3.7, 3.8, 3.9 |
| 1.9.1 | 10.2, 11.1 | 4.0.1, 4.1, 4.2 | — | 3.6, 3.7, 3.8, 3.9 |
| 1.9.0 | 10.2, 11.1 | 4.0.1, 4.1, 4.2 | — | 3.6, 3.7, 3.8, 3.9 |
| 1.8.1 | 10.1, 10.2, 11.1 | 3.10, 4.0.1 | — | 3.6, 3.7, 3.8, 3.9 |
| 1.8.0 | 10.1, 10.2, 11.1 | 3.10, 4.0.1 | — | 3.6, 3.7, 3.8, 3.9 |
| 1.7.1 | 9.2, 10.1, 10.2, 11.0 | 3.7, 3.8 | — | 3.6, 3.7, 3.8, 3.9 |
| 1.7.0 | 9.2, 10.1, 10.2, 11.0 | — | — | 3.6, 3.7, 3.8 |
| 1.6.0 | 9.2, 10.1, 10.2 | — | — | 3.6, 3.7, 3.8 |
| 1.5.1 | 9.2, 10.1, 10.2 | — | — | 3.5, 3.6, 3.7, 3.8 |
| 1.5.0 | 9.2, 10.1, 10.2 | — | — | 2.7, 3.5, 3.6, 3.7, 3.8 |
| 1.4.0 | 9.2, 10.0, 10.1 | — | — | 2.7, 3.5, 3.6, 3.7, 3.8 |
| 1.3.1 | 9.2, 10.0, 10.1 | — | — | 2.7, 3.5, 3.6, 3.7 |
| 1.3.0.post2 | — | — | — | 2.7, 3.5, 3.6, 3.7 |
| 1.3.0 | 9.2, 10.0, 10.1 | — | — | 2.7, 3.5, 3.6, 3.7 |
| 1.2.0 | 9.2, 10.0 | — | — | 2.7, 3.5, 3.6, 3.7 |
| 1.1.0.post2 | — | — | — | 2.7, 3.5, 3.6, 3.7 |
| 1.1.0 | 9.0, 10.0 | — | — | 2.7, 3.5, 3.6, 3.7 |
| 1.0.1.post2 | 8.0, 9.0, 10.0 | — | — | 2.7, 3.5, 3.6, 3.7 |
| 1.0.1 | 8.0, 9.0, 10.0 | — | — | 2.7, 3.5, 3.6, 3.7 |
| 1.0.0 | 8.0, 9.0, 10.0 | — | — | 2.7, 3.5, 3.6, 3.7 |
| 0.4.1.post2 | 8.0, 9.0, 9.2 | — | — | 3.7 |
| 0.4.1 | 8.0, 9.0, 9.2 | — | — | 2.7, 3.5, 3.6, 3.7 |
| 0.4.0 | 8.0, 9.0, 9.1 | — | — | 2.7, 3.5, 3.6 |
| 0.3.1 | 8.0, 9.0, 9.1 | — | — | 2.7, 3.5, 3.6 |
| 0.3.0.post4 | 7.5, 8.0, 9.0 | — | — | 2.7, 3.5, 3.6 |
| 0.3.0.post3 | 7.5, 8.0, 9.0 | — | — | 2.7, 3.5, 3.6 |
| 0.3.0.post2 | 7.5, 8.0, 9.0 | — | — | 2.7, 3.5, 3.6 |
| 0.3.0 | 7.5, 8.0, 9.0 | — | — | 2.7, 3.5, 3.6 |
| 0.2.0.post3 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.2.0.post2 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.2.0.post1 | 7.5 | — | — | 2.7, 3.5, 3.6 |
| 0.1.12.post2 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.12.post1 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.11.post5 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.11.post4 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.10.post2 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.10.post1 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.9.post2 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.9.post1 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.8.post1 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.7.post2 | 7.5, 8.0 | — | — | 2.7, 3.5, 3.6 |
| 0.1.6.post22 | 7.5, 8.0 | — | — | 2.7, 3.5 |
| 0.1.6.post20 | 7.5, 8.0 | — | — | 2.7, 3.5 |
FAQ
Which CUDA versions does PyTorch 2.14.1 support?
PyTorch 2.14.1 ships official wheels for CUDA 12.6, 13.0, 13.2, ROCm 7.2, 7.14, Intel XPU and CPU.
Which Python versions does PyTorch 2.14.1 support?
PyTorch 2.14.1 has wheels for Python 3.10, 3.11, 3.12, 3.13, 3.14, 3.15.
Which NVIDIA driver version do I need for PyTorch?
It depends on the CUDA version of the wheel. For example, CUDA 13.2 needs driver 595.45.04 or newer on Linux. Pick a PyTorch version and CUDA version above to see the minimum driver for each build.
Does PyTorch support AMD GPUs?
Yes. PyTorch publishes ROCm wheels for Linux x86_64. Select ROCm as the accelerator above, or pick your GPU, to see which ROCm builds officially support your card's gfx architecture.
How do I check which CUDA version my PyTorch was built with?
Run python -c "import torch; print(torch.__version__, torch.version.cuda, torch.cuda.is_available())". torch.version.cuda is the CUDA toolkit the wheel was built with; your driver must be at least the minimum listed for that version.
Notes
- Minimum driver for CUDA rows comes from the NVIDIA CUDA release notes. Thanks to minor-version compatibility, a driver new enough for CUDA 11.x/12.x/13.x runs every toolkit of the same major version.
- ROCm GPU lists are the architectures officially supported per
AMD's documentation;
more consumer GPUs often work unofficially (e.g. via
HSA_OVERRIDE_GFX_VERSION). - GPU filter compares against toolkit-level support — official wheels are sometimes built for fewer
architectures (
TORCH_CUDA_ARCH_LIST), so treat results as an upper bound and check withtorch.cuda.is_available(). - macOS CPU wheels include MPS (Metal) acceleration on Apple Silicon for PyTorch ≥ 1.12.
- NVIDIA Jetson (Xavier, Orin) is Linux aarch64 only and uses NVIDIA's JetPack-specific wheels — the aarch64 CUDA wheels here target SBSA servers (e.g. GH200).
- Intel XPU wheels need Intel GPU drivers.