Hey @brunocaballero,
That's cool! Experimenting with ONNX is a great way to make models portable.

While ONNX itself doesn't have a dedicated C++ API for programmatic model creation like onnx.helper, there's a work-around:

• ONNX Runtime C++ API: This lets you load and run pre-built ONNX models in C++. It's not for creating them directly, but it's a powerful tool for integrating models into C++ applications. Check it out here: https://github.com/microsoft/onnxruntime

• Alternative Libraries: Some libraries like PyTorch or TensorFlow offer C++ APIs that you can use to train and define models. These can then be exported to the ONNX format for deployment.

Bottom line: No C++ equivalent of onnx.helper for pure model creation, but ONNX Runtime C++ API and libraries like PyTorch/TensorFlow C++ can help integrate models in your C++ project.

If it helped, let me know. Happy Coding!