torch
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Decision gist · record as of 2026-08-14
Yes, if you need GPU-accelerated deep learning or tensor computation. PyTorch is actively maintained, permissively licensed, and has no known security vulnerabilities. Install friction is real (large CUDA dependencies) but manageable with prebuilt wheels. Not necessary if you only do CPU-bound NumPy work or need a simpler inference-only framework.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- CUDA toolkit and compatible GPU drivers required for GPU acceleration; CPU-only wheels available but slower.
- Requires Python >=3.10.
- Medium install friction due to large compiled dependencies (CUDA toolkit, cuDNN, NCCL, Triton).
License · maintenance · safety
Apache-2.0 AND Apache-2.0 WITH LLVM-exception AND BSD-2-Clause AND BSD-3-Clause AND BSL-1.0 AND MIT (permissive) — Permissive multi-license (Apache-2.0, Apache-2.0 WITH LLVM-exception, BSD-2-Clause, BSD-3-Clause, BSL-1.0, MIT) allows commercial and private use with minimal restrictions.
last release 2026-07-08 (37 days) · last repo commit 2026-08-14 · 102,361 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 102,480,941 downloads/mo, #335 on PyPI
Alternatives
Verify before relying
pip install torch
import torch
x = torch.randn(3, 4)
y = x.sum()- Specific performance benchmarks or memory efficiency claims relative to alternatives
- Compatibility details for AMD ROCm or Intel GPU support beyond CPU/NVIDIA CUDA
- Whether all 14 runtime dependencies are mandatory or conditionally installed
What it is and what it does
PyTorch is a machine learning library that provides tensors (multidimensional arrays) with GPU acceleration and a tape-based automatic differentiation system for computing gradients. It sits between NumPy-like tensor operations and deep learning frameworks, letting you write neural network code imperatively in Python without building a static computation graph first. The library integrates with NVIDIA CUDA, cuDNN, and NCCL for GPU compute, and depends on 14 runtime packages including Triton (GPU code generation), Jinja2 (templating), SymPy (symbolic math), and NetworkX (graph operations).
Most developers use PyTorch either as a NumPy replacement when GPU acceleration is needed, or as a research platform for building custom neural networks with maximum flexibility. The tape-based autograd system means you can change network structure on the fly, making it natural for dynamic architectures and debugging. Installation involves medium friction due to compiled CUDA dependencies, but prebuilt wheels cover modern Python versions (3.10–3.14) and major platforms.
Use it for
- Train convolutional or recurrent neural networks on GPU-accelerated tensor operations
- Prototype dynamic neural architectures where network structure changes per forward pass
- Replace NumPy for scientific computing tasks that benefit from GPU acceleration
- Build custom layers and loss functions using Python and automatic differentiation
- Deploy serialized models via TorchScript for inference in production systems
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you need GPU-accelerated deep learning or tensor computation.
PyTorch is actively maintained, permissively licensed, and has no known security vulnerabilities. Install friction is real (large CUDA dependencies) but manageable with prebuilt wheels. Not necessary if you only do CPU-bound NumPy work or need a simpler inference-only framework.
Install
torch on PyPI
Before you install
Medium install friction due to large compiled dependencies (CUDA toolkit, cuDNN, NCCL, Triton). Actively maintained with recent release and strong repository health (102361 stars). Supports Python 3.10–3.14 with prebuilt wheels for major platforms.
CUDA toolkit and compatible GPU drivers required for GPU acceleration; CPU-only wheels available but slower. Requires Python >=3.10.
License in practice
Permissive multi-license (Apache-2.0, Apache-2.0 WITH LLVM-exception, BSD-2-Clause, BSD-3-Clause, BSL-1.0, MIT) allows commercial and private use with minimal restrictions.
Quickstart
pip install torch
import torch
x = torch.randn(3, 4)
y = x.sum()Verify before relying
- Specific performance benchmarks or memory efficiency claims relative to alternatives
- Compatibility details for AMD ROCm or Intel GPU support beyond CPU/NVIDIA CUDA
- Whether all 14 runtime dependencies are mandatory or conditionally installed
Package facts
| License | Apache-2.0 AND Apache-2.0 WITH LLVM-exception AND BSD-2-Clause AND BSD-3-Clause AND BSL-1.0 AND MIT permissive |
| Python support | Supports the current Python release >=3.10 |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 14 packagesfilelocktyping-extensionssetuptoolssympynetworkxjinja2fsspeccuda-toolkitcuda-bindingsnvidia-cudnn-cu13nvidia-cusparselt-cu13nvidia-nccl-cu13nvidia-nvshmem-cu13triton |
| Maintenance | Actively maintained 37 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 102,480,941 / month, #335 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Development Status :: 5 - Production/StableIntended Audience :: DevelopersIntended Audience :: EducationIntended Audience :: Science/ResearchProgramming Language :: C++Programming Language :: Python :: 3 :: OnlyProgramming Language :: Python :: 3.10Programming Language :: Python :: 3.11Programming Language :: Python :: 3.12Programming Language :: Python :: 3.13Programming Language :: Python :: 3.14Topic :: Scientific/EngineeringTopic :: Scientific/Engineering :: Artificial IntelligenceTopic :: Scientific/Engineering :: MathematicsTopic :: Software DevelopmentTopic :: Software Development :: LibrariesTopic :: Software Development :: Libraries :: Python Modules |
Evidence: torch-2.13.0-cp310-cp310-macosx_14_0_arm64.whl; torch-2.13.0-cp310-cp310-manylinux_2_28_aarch64.whl; torch-2.13.0-cp310-cp310-manylinux_2_28_x86_64.whl; torch-2.13.0-cp310-cp310-win_amd64.whl; torch-2.13.0-cp311-cp311-macosx_14_0_arm64.whl; torch-2.13.0-cp311-cp311-manylinux_2_28_aarch64.whl; torch-2.13.0-cp311-cp311-manylinux_2_28_x86_64.whl; torch-2.13.0-cp311-cp311-win_amd64.whl; torch-2.13.0-cp312-cp312-macosx_14_0_arm64.whl; torch-2.13.0-cp312-cp312-manylinux_2_28_aarch64.whl; torch-2.13.0-cp312-cp312-manylinux_2_28_x86_64.whl; torch-2.13.0-cp312-cp312-win_amd64.whl; torch-2.13.0-cp313-cp313-macosx_14_0_arm64.whl; torch-2.13.0-cp313-cp313-manylinux_2_28_aarch64.whl; torch-2.13.0-cp313-cp313-manylinux_2_28_x86_64.whl; torch-2.13.0-cp313-cp313-win_amd64.whl; torch-2.13.0-cp314-cp314-macosx_14_0_arm64.whl; torch-2.13.0-cp314-cp314-manylinux_2_28_aarch64.whl; torch-2.13.0-cp314-cp314-manylinux_2_28_x86_64.whl; torch-2.13.0-cp314-cp314t-macosx_14_0_arm64.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 › “gpu-accelerated numpy alternative”
- torchPyTorch provides GPU-accelerated tensor computation and automatic…
- cupy-cuda13xCuPy provides a NumPy and SciPy-compatible array library that runs…
- cupy-cuda12xCuPy provides a NumPy- and SciPy-compatible array library that runs…
Give your agent the search over MCP, or paste the wish link into any chat.
More Software Development packages
Provides backported and experimental type hints for Python 3.9+, allowing use of newer typing features on older Python versions and enabling early experimentation with type system PEPs before they enter the standard library.
NumPy provides an N-dimensional array object and a comprehensive suite of mathematical, linear algebra, Fourier transform, and random number functions for scientific computing in Python.
FastAPI is a Python web framework for building REST APIs using type hints, with automatic request validation, serialization, and interactive API documentation.
Provides a way to document function parameters, class attributes, return types, and variables inline using Python's `Annotated` type hint syntax instead of traditional docstrings.
Typer builds command-line applications from Python functions using type hints, automatically generating help text, argument parsing, and shell completion.
Install it if you are building CLIs in Python.
Distlib provides low-level packaging utilities for building, distributing, and managing Python software—including metadata handling, version specifiers, wheel support, script installation, and dependency resolution.
See also hoptorch · nvidia-cudnn-cu12 · pytorch · pytorch-ignite · pytorch_revgrad · tinygrad · torch-complex · torch-directml · torch-geometric · torch-npu