Back to Gap Map Infrastructure / Compilers & Model Optimization

tf2onnx

open source / Overall score: 2.5

tf2onnx converts TensorFlow, Keras, TensorFlow.js and TFLite models into ONNX, from the command line or a Python API, mapping framework operations onto ONNX operator sets. It is the ONNX project's TensorFlow-side converter and sits under the ONNX organization.

Openness

5 high confidence

5.0

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, so the core is ungated.

https://raw.githubusercontent.com/onnx/tensorflow-onnx/main/LICENSE recorded 2026-09-30
LICENSE file is the verbatim Apache-2.0 text, opening with "Apache License" and "Version 2.0, January 2004".
https://raw.githubusercontent.com/onnx/tensorflow-onnx/main/README.md recorded 2026-09-30
README states "SPDX-License-Identifier: Apache-2.0" and closes with "[Apache License v2.0](LICENSE)", and describes a public project seeking a maintainer, "looking for a new maintainer". Searched the body for enterprise, license key, paid, commercial and proprietary: none.
https://raw.githubusercontent.com/onnx/tensorflow-onnx/main/setup.py recorded 2026-09-30
setup.py opens with "# SPDX-License-Identifier: Apache-2.0" and builds the package with plain setuptools commands, "setuptools.command.build_py.build_py.run(self)", with no key or gated edition. Searched the body for enterprise, license key, paid, commercial and proprietary: none.

Adoption

3 high confidence

3.0

Adoption is measured from PyPI downloads of the `tf2onnx` package.

https://pypistats.org/api/packages/tf2onnx/recent recorded 2026-08-18
last_month downloads = 381,440 for tf2onnx

Capability

2 medium confidence

2.0

tf2onnx converts TensorFlow, Keras, TensorFlow.js and TFLite models to ONNX and runs ONNX-level graph cleanup passes such as transpose removal and constant folding. It compiles to no hardware target, so this single-purpose utility sits two steps below TensorRT, which fuses and quantizes for NVIDIA GPU engines.

https://raw.githubusercontent.com/onnx/tensorflow-onnx/main/README.md recorded 2026-09-30
README opens "tf2onnx converts TensorFlow (tf-2.x), keras, tensorflow.js and tflite models to ONNX via command" line or python api, and documents "We support and test ONNX opset-14 to opset-18."
https://raw.githubusercontent.com/onnx/tensorflow-onnx/main/tf2onnx/optimizer/__init__.py recorded 2026-09-30
Optimizer module registers ONNX graph cleanup passes including "optimize_transpose", "fold_constants" and "remove_identity", run by "def optimize_graph(graph, catch_errors=True, optimizers=None):".

Verified 2026-09-24