onnx
Open Neural Network Exchange
Decision gist · record as of 2026-08-14
Yes. ONNX is a mature, widely-adopted standard (active maintenance, 21312 stars, top 5000 PyPI package) with no known vulnerabilities, permissive licensing, and broad platform support. Install it if you need to work with ONNX models, export models to ONNX format, or build cross-framework inference pipelines. The medium install friction is manageable given the availability of prebuilt wheels for common platforms and Python versions.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- Requires Python 3.10 or later.
- Model files must be in valid ONNX format.
- Medium install friction with prebuilt wheels for common platforms (macOS, Linux x86_64, Windows, ARM) and Python versions 3.10–3.14.
License · maintenance · safety
Apache-2.0 (permissive) — Apache-2.0 is permissive; you may use, modify, and distribute ONNX freely in commercial and open-source projects, provided you include a copy of the license and note any material changes.
last release 2026-06-15 (60 days) · last repo commit 2026-08-14 · 21,312 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 20,484,447 downloads/mo, #1,035 on PyPI
Alternatives
Verify before relying
pip install onnx
import onnx
model = onnx.load('model.onnx')
onnx.checker.check_model(model)- Whether the package includes a reference implementation or if optional dependencies are needed for full inference capability.
- Performance characteristics and inference speed compared to native framework execution.
- Supported ONNX opset versions and operator coverage for your specific models.
What it is and what it does
ONNX is a standardized, open-source format for representing machine learning models—both deep learning and traditional ML—along with a Python package for loading, inspecting, and manipulating those models. It defines an extensible computation graph model, built-in operators, and standard data types, with a focus on inference (scoring). The package lets you load ONNX model files, validate their structure, perform shape and type inference, and convert between opset versions.
ONNX is widely adopted across frameworks (PyTorch, TensorFlow, scikit-learn, and others) and hardware platforms, making it a bridge between research and production. By using ONNX, you can train a model in one framework and deploy it with a different runtime or hardware accelerator without rewriting inference code. The Python package depends on numpy, protobuf, typing_extensions, and ml_dtypes, and provides abi3-compatible wheels for Python 3.12 and later, allowing a single binary to work across multiple Python versions.
Use it for
- Export a model trained in PyTorch or TensorFlow to ONNX format for deployment on edge devices or inference servers.
- Load and validate ONNX models to ensure they conform to the specification before production use.
- Convert models between different ONNX opset versions to maintain compatibility across tools and runtimes.
- Perform shape and type inference on ONNX graphs to understand model I/O and intermediate tensor properties.
- Build model optimization and transformation pipelines that work across multiple training frameworks.
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes.
ONNX is a mature, widely-adopted standard (active maintenance, 21312 stars, top 5000 PyPI package) with no known vulnerabilities, permissive licensing, and broad platform support. Install it if you need to work with ONNX models, export models to ONNX format, or build cross-framework inference pipelines. The medium install friction is manageable given the availability of prebuilt wheels for common platforms and Python versions.
Install
onnx on PyPI
Before you install
Medium install friction with prebuilt wheels for common platforms (macOS, Linux x86_64, Windows, ARM) and Python versions 3.10–3.14. Active maintenance with a recent release and 21312 repository stars. Depends on numpy, protobuf, typing_extensions, and ml_dtypes.
Requires Python 3.10 or later. Model files must be in valid ONNX format.
License in practice
Apache-2.0 is permissive; you may use, modify, and distribute ONNX freely in commercial and open-source projects, provided you include a copy of the license and note any material changes.
Quickstart
pip install onnx
import onnx
model = onnx.load('model.onnx')
onnx.checker.check_model(model)Verify before relying
- Whether the package includes a reference implementation or if optional dependencies are needed for full inference capability.
- Performance characteristics and inference speed compared to native framework execution.
- Supported ONNX opset versions and operator coverage for your specific models.
Package facts
| License | Apache-2.0 permissive |
| Python support | Supports the current Python release >=3.10 |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 4 packagesnumpyprotobuftyping_extensionsml_dtypes |
| Maintenance | Actively maintained 60 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 20,484,447 / month, #1,035 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
| Classifiers | Programming Language :: Python :: 3 |
Evidence: onnx-1.22.0-cp310-cp310-macosx_12_0_universal2.whl; onnx-1.22.0-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; onnx-1.22.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; onnx-1.22.0-cp310-cp310-win32.whl; onnx-1.22.0-cp310-cp310-win_amd64.whl; onnx-1.22.0-cp311-cp311-macosx_12_0_universal2.whl; onnx-1.22.0-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; onnx-1.22.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; onnx-1.22.0-cp311-cp311-win32.whl; onnx-1.22.0-cp311-cp311-win_amd64.whl; onnx-1.22.0-cp311-cp311-win_arm64.whl; onnx-1.22.0-cp312-abi3-macosx_12_0_universal2.whl; onnx-1.22.0-cp312-abi3-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; onnx-1.22.0-cp312-abi3-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; onnx-1.22.0-cp312-abi3-pyemscripten_2025_0_wasm32.whl; onnx-1.22.0-cp312-abi3-win32.whl; onnx-1.22.0-cp312-abi3-win_amd64.whl; onnx-1.22.0-cp312-abi3-win_arm64.whl; onnx-1.22.0-cp314-cp314t-macosx_12_0_universal2.whl; onnx-1.22.0-cp314-cp314t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.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 › “model interoperability framework”
- onnxONNX provides an open-source format and runtime for representing and…
- onnxconverter-commonProvides common utilities and functions for converting machine…
- treeliteTreelite serializes and exchanges decision tree forest models in a…
Give your agent the search over MCP, or paste the wish link into any chat.
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to call 100+ LLM providers (OpenAI, Anthropic, Gemini, Bedrock, Azure, and others) using OpenAI-compatible API format, available as both a Python SDK and a self-hosted AI Gateway proxy server.
Install it if you need to work with multiple LLM providers or want to centralize LLM routing in your organization.
Client library and CLI tool for downloading, uploading, and managing models, datasets, and repositories on the Hugging Face Hub platform.
Install it if you work with Hugging Face Hub models or datasets.
LangChain provides a framework for building agents and LLM-powered applications by composing language models, tools, and memory through a unified API that abstracts over multiple model providers.
hf-xet provides chunk-based deduplication and efficient file transfer for the Hugging Face Hub, enabling faster uploads and downloads of large files with local disk caching.
Tokenizers converts raw text into token sequences for NLP models, with support for training custom vocabularies and using pre-built tokenizers (BPE, WordPiece) optimized for speed via Rust.
Transformers provides a unified framework for loading, fine-tuning, and running state-of-the-art pretrained models across text, vision, audio, video, and multimodal tasks using PyTorch, JAX, or TensorFlow.
Install it if you need to run or train any transformer-based model for NLP, vision, audio, or multimodal tasks.
See also onnx-weekly · onnxruntime · onnxconverter-common · skl2onnx · onnxruntime-gpu · onnxsim · onnx-ir · onnxruntime_extensions · onnxmltools · multi-model-server