Pull Requests microsoft/onnxruntime

Set workflow permissions to read

#29656 by pacowong was closed 2026-07-10 15:04

Fix CUDA AveragePool wrong results with asymmetric padding and dilation ep:CUDA

#29631 by titaiwangms was merged 2026-07-10 17:48

[WebGPU] Plugin EP 0.2.0 release cherry picks round 1

#29630 by edgchen1 was merged 2026-07-10 16:54

ORT Release 1.27.1 cherry pick 1

#29628 by tianleiwu was merged 2026-07-09 05:39

Fix JSEP pooling output shape to honor ceil_mode platform:web javascript ep:WebGPU

#29627 by titaiwangms was merged 2026-07-09 18:44

Avoid throwing C++ exceptions from WebGPU EP Dawn callbacks plugin-ep-webgpu/release:0.2.0

#29625 by edgchen1 was merged 2026-07-08 23:26

Hoist CropAndResize index offset to a named int64_t

#29623 by titaiwangms was merged 2026-07-09 20:34

Speed up CUDA build by splitting SM80 MoE GEMM generated files

#29614 by tianleiwu was merged 2026-07-08 17:06

Not-For-Review: Offline autotuner for the Intel subgroup-matrix MatMul

#29610 by jchen10 was closed 2026-07-08 01:07

Fix web e2e (npm/vite) and Python DML CI pipelines

#29609 by tianleiwu was merged 2026-07-08 02:37

Bump onnx to 1.22.0 and protobuf to 6.33.5 to fix security CVEs

#29606 by sushraja-msft was merged 2026-07-08 00:08

Handle non-finite ROI coordinates in CropAndResize

#29605 by titaiwangms was merged 2026-07-08 17:31

Skip MinLength logits processor construction when eos_token_id is negative

#29604 by titaiwangms was merged 2026-07-08 16:08

chore: bump version to 1.29.0

#29600 by sanaa-hamel-microsoft was merged 2026-07-08 03:41

Enhance build workflow with runtime context logging

#29597 by amitaysh was closed 2026-07-07 09:47

[CUDA] Fix fpA_intB internal test build: typed nullptr for TryMatMulNBits bias

#29596 by tianleiwu was merged 2026-07-07 16:01

webgpu: Fix GQA decode split-reduce head_size out-of-bounds race

#29593 by qjia7 was merged 2026-07-07 16:29

[WebGPU] Fix OrtReleaseEnv self-deadlock after failed adapter request on Linux plugin-ep-webgpu/release:0.2.0

#29591 by Copilot was merged 2026-07-08 03:36

Fix libcudart.so.13 hard dependency in pybind module breaking import on CPU-only Linux release:1.28.0

#29590 by Copilot was merged 2026-07-08 18:15

webgpu: Bump FlashAttentionDecodeQKV workgroup to 128, tile_size_k_vec to 32

#29586 by qjia7 was merged 2026-07-09 05:51

[CUDA] Enable native SM90, block_size=32, and fused bias for fpA_intB MatMulNBits

#29585 by tianleiwu was merged 2026-07-07 20:57

[CUDA] Fix QMoE profiler cross-stream race and CUDA-graph-capture safety

#29584 by tianleiwu was merged 2026-07-07 20:53

[CUDA] Fix QMoE profiler cross-stream race and CUDA-graph-capture safety

#29583 by tianleiwu was closed 2026-07-07 03:14

Fix CUDA 12 build errors with 4bit matmul

#29582 by skottmckay was closed 2026-07-07 00:07

Guard GridSample CUDA coordinate conversion

#29581 by GopalakrishnanN was merged 2026-07-09 01:07

[Build] Fix cuda packaging build error

#29580 by tianleiwu was merged 2026-07-07 02:20

Release outputHandlesArr in OrtTrainingSession.evalStep JNI binding

#29576 by Sammy-Dabbas was merged 2026-07-07 16:53

Don't echo command when setting VSO variable in mac-cpu-packing-jobs.yml.

#29575 by edgchen1 was merged 2026-07-06 22:16

Fix CustomOp forward compatibility: cap version instead of rejecting release:1.27.1

#29574 by chilo-ms was merged 2026-07-07 20:27

[ORT] onnxruntime_perf_test: Add multi-shape profiling support to perftest tool (--data_shape)

#29571 by ankitm3k was merged 2026-07-09 01:48