Open Neural Network Exchange (ONNX)
Definition
ONNX facilitates the exchange of models between compatible tools.
An open format and ecosystem for representing and transferring models between compatible tools and runtimes.
Simply explained
A common model format helps to use a trained AI component in another suitable program.
Why it matters in practice
Helps develop and operate digital systems robustly, securely, and maintainably.
Example from business
A trained model is exported and tested on target hardware in another runtime.
Limits and pitfalls
- Equating exportability with identical behavior or ignoring operator and opset compatibility.
Origin & history
ONNX was announced in 2017 by Microsoft and Facebook and is further developed as an open project.
Use in mid-sized companies
- Possible example: A trained model is exported and tested on target hardware in another runtime.
Frequently asked questions
What does Open Neural Network Exchange mean?
An open format and ecosystem for representing and transferring models between compatible tools and runtimes.
Which mistakes should be avoided?
Equating exportability with identical behavior or ignoring operator and opset compatibility.
Sources
Source checked on 10.10.2026. Editorial application examples are hypothetical; no evidence of real customer results.
Editorially reviewed on: