ONNX Runtime
Software
Inference
Active
The portable format for shipping models across hardware.
You're looking at a software record: a tool or platform in the AI stack — who maintains it and what hardware it runs on.
ONNX defines an open interchange format for trained models; ONNX Runtime executes them across CPUs, GPUs and NPUs. It is the pragmatic layer for enterprise and edge deployment where the serving hardware is not chosen by the model's authors.
How this connects to the rest of the stack
This record's links at a glance — click a group to jump to it.
Maintained by
Sources
Record reviewed as of 2026-01-15 — see Methodology.