Pull Requests microsoft/onnxruntime

[WIP] Fix AMX U8S8 QGEMM results on VMs without TILEDATA

#33258 opened 2026-10-11 01:12 by Copilot

[WebGPU] Support fixed subgroup size in im2col_matmul

#33249 opened 2026-10-10 12:22 by Jiawei-Shao

[CUDA] Add opt-in SM121 L2 prefetch for quantized decode

#33248 opened 2026-10-10 08:51 by kunal-vaishnavi

webgpu: Fix Python factory selection for plugin EP hosts

#33241 opened 2026-10-10 02:31 by qjia7

[CUDA] Plumb GQA workspace envelope into Level-2 declaration

#33240 opened 2026-10-09 23:27 by chilo-ms

[WIP] Fix float16 weights quantization to avoid zeroing out tensors

#33239 opened 2026-10-09 21:17 by Copilot

Add opt-in platform-specific onnxruntime-node packages

#33237 opened 2026-10-09 18:03 by avens19

Reject overflowing Concat axis dimensions

#33235 opened 2026-10-09 14:28 by xadupre

Remove stale Intel MKL third-party notice

#33233 opened 2026-10-09 10:03 by xadupre

[CUDA][WebGPU] Add PackedSparseAttentionIndexerMerge for MTP index sharing

#33224 opened 2026-10-09 01:42 by kunal-vaishnavi

[Objective-C] Add encrypted compiled EPContext inference coverage

#33223 opened 2026-10-08 22:11 by GopalakrishnanN

[React Native] Add opt-in encrypted compiled EPContext device workflow

#33222 opened 2026-10-08 22:11 by GopalakrishnanN

Test C# encrypted compiled-model inference

#33221 opened 2026-10-08 21:57 by GopalakrishnanN

Test Rust encrypted compiled-model inference

#33220 opened 2026-10-08 21:57 by GopalakrishnanN

Test Node.js encrypted compiled-model inference

#33219 opened 2026-10-08 21:57 by GopalakrishnanN

Test Python encrypted compiled-model inference

#33218 opened 2026-10-08 21:57 by GopalakrishnanN

Test Java encrypted compiled-model inference

#33217 opened 2026-10-08 21:57 by GopalakrishnanN

Add executable EPContext fixture for encrypted binding tests

#33216 opened 2026-10-08 21:57 by GopalakrishnanN

[WebGPU] Add GatedDeltaNetStateReplay contrib operator

#33213 opened 2026-10-08 20:51 by aciddelgado

[WebGPU EP] fuse layer norm projections into gated delta net

#33211 opened 2026-10-08 19:57 by prathikr

Update libprotobuf-mutator to handle protobuf api changes

#33210 opened 2026-10-08 19:28 by eserscor

Make CPUinfo DLL-unload ownership checks deterministic

#33205 opened 2026-10-08 17:14 by GopalakrishnanN

Finish React Native workers before publishing promise results

#33204 opened 2026-10-08 16:42 by GopalakrishnanN

[WebGPU] Skip unsupported FP16 execution in operator tests

#33201 opened 2026-10-08 12:10 by gyagp

Enable 1DS telemetry in exception-free builds

#33199 opened 2026-10-08 10:49 by bmehta001

Fix Windows float16 model-path conversion

#33193 opened 2026-10-08 02:55 by aswanth-07

Guard normalized WebGPU GEMM dispatch bounds

#33174 opened 2026-10-07 18:46 by danielsongmicrosoft

cuda plugin: add native sm_120 SASS to the LLM target under MSVC

#33173 opened 2026-10-07 18:40 by tianleiwu

Add opt-in quantized LoraMulAdd fusion

#33172 opened 2026-10-07 18:34 by mustjab

[WebGPU] Handle empty-rank MatMul and DequantizeLinear correctly

#33171 opened 2026-10-07 18:34 by mustjab