[ospillinger]

Yes, Cortex uses ONNX Runtime (https://github.com/microsoft/onnxruntime) under the hood so any model that can be exported to ONNX can be deployed.

[solidasparagus]

Is it only able to handle ONNX models? That's a pretty massive limitation compared to a hosted SageMaker endpoint.

[vishalbollu]

Contributor here - Cortex supports Tensorflow saved models in addition to ONNX. PyTorch support is on the roadmap. Do you have specific frameworks in mind that you would like Cortex to support?

[solidasparagus]

Perfect. Nothing in particular other than TF.