Release

ONNX Runtime CUDA Plugin EP 0.3.0 adds sparse attention and INT2 QMoE support

Event: Oct 9, 2026 · Published: Oct 12, 2026

On October 9, 2026 Microsoft published version 0.3.0 of the ONNX Runtime CUDA Plugin EP on the ONNX Runtime GitHub repository. This is a separate release from ONNX Runtime 1.31.0, which the notes give as the matched core version. The release notes list sparse attention and packed continuous-batching support, INT2 and mixed-width QMoE execution, new QSA/CSA attention operators, expanded paged speculative decoding, and opt-in MatMul auto-tuning. The notes warn that the core and the plugin must use the same contributed-operator schemas and recommend building both from the same ONNX Runtime revision. USASI has not tested this release.1

Sources

1.
ONNX Runtime CUDA Plugin EP 0.3.0 (GitHub release notes) (external site: github.com)
Microsoft (ONNX Runtime) · Release notes · published Oct 9, 2026 · accessed Oct 12, 2026

Email a correction Report a correction Edit this entry

All news

Cite this page

APA style

USASI. (2026, October 12). ONNX Runtime CUDA Plugin EP 0.3.0 adds sparse attention and INT2 QMoE support. United States of America Superintelligence. https://unitedstatesofamericasuperintelligence.com/news/onnx-runtime-cuda-plugin-ep-0-3-0-released/

BibTeX

@misc{usasi_news_onnx_runtime_cuda_plugin_ep_0_3_0_released,
  author = {{USASI}},
  title = {ONNX Runtime CUDA Plugin EP 0.3.0 adds sparse attention and INT2 QMoE support},
  year = {2026},
  month = oct,
  howpublished = {\url{https://unitedstatesofamericasuperintelligence.com/news/onnx-runtime-cuda-plugin-ep-0-3-0-released/}},
  note = {United States of America Superintelligence. Last updated 2026-10-12}
}

The date is when this page was last updated. Add the date you read it if your style needs one.

Related catalog entries

Sources

Share or cite this page

Support Us