ONNX Simplifier project

license

Apache-2.0(OSI)

source

public

core features withheld

no

The LICENSE file contains the Apache-2.0 text. The repository is public and the README describes no paid tier, enterprise edition or license-gated build beyond optional third-party provider integrations, so the core is ungated.

• https://raw.githubusercontent.com/onnxsim/onnxsim/master/LICENSE recorded 2026-09-30

  LICENSE file is the verbatim Apache-2.0 text, opening with "Apache License" and "Version 2.0, January 2004" and "TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION".

• https://raw.githubusercontent.com/onnxsim/onnxsim/master/pyproject.toml recorded 2026-09-30

  pyproject.toml declares license = "MIT AND (Apache-2.0 OR BSD-2-Clause)" and a build from the repository through requires = ["setuptools>=77", "wheel", "nanobind>=2.12.0"], with onnx as the only required dependency.

• https://raw.githubusercontent.com/onnxsim/onnxsim/master/README.md recorded 2026-09-30

  README says "onnxruntime is optional; onnxsim falls back to the onnx reference" evaluator and that "`onnx` is the only required dependency." Searched the body for enterprise, license key, paid, commercial, proprietary, pricing, subscription and closed source: none.

PyPI downloads of `onnxsim` are the adoption signal here.

• https://pypistats.org/api/packages/onnxsim/recent recorded 2026-08-18

  last_month downloads = 535,030 for onnxsim

ONNX Simplifier runs shape inference, optimizer passes and constant folding to a fixed point and bridges the result to MLIR, Core ML and TFLite, with quantization-aware fine-tuning beside them. That is real coverage across several passes and routes, but the final lowering to a device is left to other tools, so it sits one step below TensorRT, which compiles and runs the model on NVIDIA hardware.

• https://raw.githubusercontent.com/onnxsim/onnxsim/master/README.md recorded 2026-09-30

  README features list says "At its core onnxsim runs a fixed point of shape inference, graph optimization and" constant folding, and adds a Core ML export "via a built-in ONNX-to-MIL translator and" a TensorFlow Lite export "via a built-in ONNX-to-TensorFlow translator and" tf.lite.TFLiteConverter.

• https://raw.githubusercontent.com/onnxsim/onnxsim/master/README.md recorded 2026-09-30

  README says "a bridge into MLIR-based compiler stacks" and "Recover accuracy a quantization lost with `onnxsim.apply_qat()`:" and "A zero-install [web version](#web-version), a Python"