ONNX
Shares rank 16 of 27 inference engines by recommenders
Free, open model format and runtime for running trained models across CPUs, GPUs and browsers, with many models available. Common caveats: tooling is raw, iOS use is hard, and exports can differ from originals.
Free, open model format and runtime for running trained models across CPUs, GPUs and browsers, with many models available. Common caveats: tooling is raw, iOS use is hard, and exports can differ from originals.
15 people gave an opinion
- recommend it
- 9
- criticize it
- 6
- approval share
- 60%
Recommended in 7 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: recommended
- February 2026: not recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: not recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 9 people recommend itNeeds 10 or more, not met.
- 1.5 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 7 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
Other inference engines people recommend
The four most recommended other inference engines, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| vLLM | 112 | 56 | 67% | 12 | — | Approved |
| Strata | 31 | 15 | 67% | 5 | — | Approved |
| DwarfStar 4 | 26 | 2 | 93% | 6 | — | Approved |
| SGLang | 19 | 13 | 59% | 8 | — | — |
| ONNXThis page | 9 | 6 | 60% | 7 | — | — |
What people said
All 15 comments that mention it, newest first. Each links to the original comment.
“considering you now have semantic scoring "at your finger tips" without resorting to levenshteins”
“ONNX never got within spitting distance of llama.cpp”
“dealing with ONNX is kind of annoying”
“the cross platform alternatives (ONNX) are terrible”
“Packaging for deployment as ONNX is pretty simple as well.”
“A torch model exported as onnx is a different model.”
“I only wish that they weren't so frustratingly difficult to use on Apple iOS.”
“This format continues to impress me.”
“onnx and coreml are secure but they can only do the actual model not all the code around it”
“I already thought that ONNX (CPU at least) was already low level code and already very optimized.”
“ONNX decided that it wants to link against my ROCm instance specifically”
“Microsoft is doing the same with ONNX”
“ONNX and DirectML seem sort of promising right now, but it's all super raw.”
“has the most models available and can use webgpu which is available everywhere”
“none of them are quite there yet in terms of efficiency”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.