#26722·onnxruntime

[Meta] WebNN - Windows ML / ONNX Runtime Issues and Feature Requests

Author: ibelemCreated Dec 4, 2025Updated Sep 17, 2026
Labelsep:OpenVINOep:DMLplatform:windowsplatform:webep:WebNNep:WebGPU

Describe the issue

This is a meta issue to track WebNN / ONNX Runtime relevant open issues.

OT Blocker

ORT Core VF

ORT Core

ORT OpenVINO EP NPU

ORT WebGPU EP // Required PRs for 0.3 branch release

item | unsupported_reason | detail |   |   -- | -- | -- | -- | -- Add | Missing type: int64 | v14 kernel exists, supports: float,float16,int32,uint32 | ✅ | https://github.com/microsoft/onnxruntime/pull/29844 AveragePool | Unsupported opset: v19 | No compatible kernel version registered | ✅ | https://github.com/microsoft/onnxruntime/pull/29835 Cast | Missing type: uint8 | v21 kernel exists, supports: float16,float,int32,uint32,int64,bool | ✅ | https://github.com/microsoft/onnxruntime/pull/29839 Clip | Missing type: int32 | v13 kernel exists, supports: float | ✅ | https://github.com/microsoft/onnxruntime/pull/29835 Clip | Missing type: int64 | v13 kernel exists, supports: float | ✅ | https://github.com/microsoft/onnxruntime/pull/29834 Expand | Missing type: uint8 | v13 kernel exists, supports: float16,float,int32,uint32,int64,bool | ✅ | https://github.com/microsoft/onnxruntime/pull/29854 Gather | Missing type: uint8 | v13 kernel exists, supports: float,float16,int32,uint32,bool | ✅ | https://github.com/microsoft/onnxruntime/pull/29861 HardSwish | No matching kernel | No compatible kernel version registered | ✅ | https://github.com/microsoft/onnxruntime/pull/29828 Max | No matching kernel | No compatible kernel version registered | ✅ | https://github.com/microsoft/onnxruntime/pull/29833 Min | No matching kernel | No compatible kernel version registered | ✅ | Same PR to Max Reshape | Missing type: int64 | v21 kernel exists, supports: float,float16,int32,uint32 | ✅ | https://github.com/microsoft/onnxruntime/pull/29830 Reshape | Missing type: uint8 | v21 kernel exists, supports: float,float16,int32,uint32 | ✅ | https://github.com/microsoft/onnxruntime/pull/29897 Tile | Missing type: int32 | v13 kernel exists, supports: float,float16 | ✅ | https://github.com/microsoft/onnxruntime/pull/29847 Tile | Missing type: int64 | v13 kernel exists, supports: float,float16 | ✅ | https://github.com/microsoft/onnxruntime/pull/31049 Trilu | No matching kernel | No compatible kernel version registered | ✅ | https://github.com/microsoft/onnxruntime/pull/29845 Concat | Missing type: int64 | supports: float,float16,int32,uint32 | ✅| https://github.com/microsoft/onnxruntime/pull/31049 Gru | Op missing | WebNN noise supression sample | ✅ | https://github.com/microsoft/onnxruntime/pull/29840 CumSum | Missing type: int32 | v14 kernel exists, supports: float,float16 | ✅| https://github.com/microsoft/onnxruntime/pull/31702 Gather | int64 |   | ✅ | https://github.com/microsoft/onnxruntime/pull/31714 Clip |   | Gate int64 Clip kernel behind enable_int64 flag | ✅ | https://github.com/microsoft/onnxruntime/pull/29918 Min/Max | int64 |   | ✅ | https://github.com/microsoft/onnxruntime/pull/31709 Support device-free compile-only sessions | | | ✅ | https://github.com/microsoft/onnxruntime/pull/29681 Robustness provider option | | | ✅ | https://github.com/microsoft/onnxruntime/pull/31971 Zero-initialize writable device allocator buffers | | | ✅ | https://github.com/microsoft/onnxruntime/pull/32063

Chromium

  • 8124054 skip SetFakeGdiInit() for the compiler process sandbox
  • 8091227 [WebGPU EP] add WebGpuExecutionProvider to kKnownEPs
  • 8127421 Disable ORT graph transforms in the GPU process for security

Feature Request

ep:WebGPU

Bug

ep:WebGPU

ep:DML

ORT Default CPU