Roadmap¶

Planned¶

Extend target-oriented validation guidance beyond the documented ONNX Runtime CPU and Web/WASM flows, especially for mobile deployments where practical.
Broaden capability-matrix coverage across dtype and shape variants, including BF16, dynamic dimensions, and non-square inputs.
Add focused end-to-end deployment examples for small vision and numerical models.
Add a realistic end-to-end RL deployment example based on a widely used RL library, loading a trained actor and exporting the inference-only obs -> action policy contract.
Continue targeted coverage work for JAX, Flax NNX/Linen, Equinox, SotA examples, and physics/simulation use cases, including classification of newly published upstream APIs before claiming them as supported.

Record trustworthy model provenance: Populate exported ONNX models with the active jax2onnx producer version while handling source checkouts safely, so metadata identifies the converter build without changing graph semantics.
Harden and modernize GitHub Actions: Pin third-party actions to immutable commit SHAs, declare least-privilege token access for CI and nightly jobs, and move workflows to the Node 24-based actions/checkout 7.0.1 and actions/setup-python 7.0.0 releases.
Automate dependency maintenance with bounded noise: Group minor and patch updates for GitHub Actions and npm, rate-limit update traffic across Actions, Python, npm, and pre-commit dependencies, keep major upgrades isolated for review, and defer separate uv automation until lockfile synchronization has a defined policy.
Protect both supported JAX stacks: Retain JAX/JAXLIB 0.10.2 for Python 3.11/3.12 and 0.11.0 for Python 3.13/3.14 in the Poetry lockfile, with a CI guard that verifies the modern stack remains present.
Refresh the validation and tooling stack: Validate against ONNX Runtime and onnxruntime-web 1.29.0, Playwright 1.62.1, pytest 9.1.1, and Ruff 0.16.4, with an explicit E4, E7, E9, and F lint baseline.

See Past Versions for the full release archive.