ORT 1.24 release candidate available for testing

[tianleiwu]

ORT 1.24.0 will be released soon, pending successful validation. Release candidate builds are available now for testing. If you encounter issues, please report them by responding in this issue.

Release branch: rel-1.24.0
Release manager: @tianleiwu

Python Packages

CPU: https://aiinfra.visualstudio.com/PublicPackages/_artifacts/feed/ORT-Nightly/PyPI/onnxruntime/overview/1.24.0.dev20260123005
CUDA 12: https://aiinfra.visualstudio.com/PublicPackages/_artifacts/feed/ORT-Nightly/PyPI/onnxruntime-gpu/overview/1.24.0.dev20260127002
CUDA 13: https://aiinfra.visualstudio.com/PublicPackages/_artifacts/feed/ort-cuda-13-nightly/PyPI/onnxruntime-gpu/overview/1.24.0.dev20260127002

Installation

First, you might need uninstall existed packages like

pip uninstall onnxruntime onnxruntime-gpu -y

Then choose only one of the following based on cuda version (or CPU only) to install:

CUDA 13.x

pip install flatbuffers numpy packaging protobuf sympy coloredlogs
pip install --pre --index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ort-cuda-13-nightly/pypi/simple/ onnxruntime-gpu==1.24.0.dev20260127002

CUDA 12.x

pip install flatbuffers numpy packaging protobuf sympy coloredlogs
pip install --pre --index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT-Nightly/pypi/simple/ onnxruntime-gpu==1.24.0.dev20260127002

CPU Only

pip install flatbuffers numpy packaging protobuf sympy coloredlogs
pip install --pre --index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/ORT-Nightly/pypi/simple/ onnxruntime==1.24.0.dev20260123005

Nuget packages

CUDA 13

Please use the version 1.24.0-dev-20260127-1643-50d4c84724:
Microsoft.ML.OnnxRuntime.Gpu
Microsoft.ML.OnnxRuntime.Gpu.Linux
Microsoft.ML.OnnxRuntime.Gpu.Windows
Microsoft.ML.OnnxRuntime.Managed

CUDA 12

Please use the version 1.24.0-rc.2:
Microsoft.ML.OnnxRuntime.Gpu
Microsoft.ML.OnnxRuntime.Gpu.Linux
Microsoft.ML.OnnxRuntime.Gpu.Windows
Microsoft.ML.OnnxRuntime.Gpu.Managed

[stonebig]

the pyreadline3-3.5.4 bug (for python -3.13+) can be slightly workarounded, if not patched, by using ptpython instead of python, as the standard command line....

[tianleiwu]

You can now install 1.24.1 official packages:

Python:
onnxruntime 1.24.1
onnxruntime-gpu 1.24.1 for CUDA 12.x
onnxruntime-webgpu 1.24.1
onnxruntime-directml 1.24.1
onnxruntime-gpu 1.24.1 for CUDA 13.x

How to install python package for CUDA 13.x:

pip install flatbuffers numpy packaging protobuf sympy
pip install --index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/onnxruntime-cuda-13/pypi/simple/ onnxruntime-gpu==1.24.1

Nuget:
Microsoft.ML.OnnxRuntime 1.24.1
Microsoft.ML.OnnxRuntime.Managed 1.24.1
Microsoft.ML.OnnxRuntime.Gpu 1.24.1 (for cuda 12.x)
Microsoft.ML.OnnxRuntime.Gpu.Windows 1.24.1 (for cuda 12.x)
Microsoft.ML.OnnxRuntime.Gpu.Linux 1.24.1 (for cuda 12.x)
Microsoft.ML.OnnxRuntime.QNN 1.24.1

Nuget for CUDA 13:
Microsoft.ML.OnnxRuntime.Gpu 1.24.1 (for cuda 13.x)
Microsoft.ML.OnnxRuntime.Gpu.Linux 1.24.1 (for cuda 13.x)
Microsoft.ML.OnnxRuntime.Gpu.Linux 1.24.1 (for cuda 13.x)
Microsoft.ML.OnnxRuntime.Gpu.Windows 1.24.1
Microsoft.ML.OnnxRuntime.Managed 1.24.1

JavaScript:
https://www.npmjs.com/package/onnxruntime-common
https://www.npmjs.com/package/onnxruntime-web
https://www.npmjs.com/package/onnxruntime-node
https://www.npmjs.com/package/onnxruntime-react-native

Java:
Maven: com/microsoft/onnxruntime/onnxruntime/1.24.1
Maven: com/microsoft/onnxruntime/onnxruntime_gpu/1.24.1 (for cuda 12.x)
Maven: com/microsoft/onnxruntime/onnxruntime-android/1.24.1
Maven: com/microsoft/onnxruntime/onnxruntime-android-qnn/1.24.1

[Craigacp]

The Java CPU jar is missing macOS arm64 binaries - https://repo1.maven.org/maven2/com/microsoft/onnxruntime/onnxruntime/1.24.1/onnxruntime-1.24.1.jar, it only has linux-arm64, linux-x64 and windows-x64.

[tianleiwu]

The Java CPU jar is missing macOS arm64 binaries - https://repo1.maven.org/maven2/com/microsoft/onnxruntime/onnxruntime/1.24.1/onnxruntime-1.24.1.jar, it only has linux-arm64, linux-x64 and windows-x64.

Thanks for reporting, @Craigacp. We will fix it in next patch release 1.24.2.

[Craigacp]

The Java CPU jar is missing macOS arm64 binaries - https://repo1.maven.org/maven2/com/microsoft/onnxruntime/onnxruntime/1.24.1/onnxruntime-1.24.1.jar, it only has linux-arm64, linux-x64 and windows-x64.

Thanks for reporting, @Craigacp. We will fix it in next patch release 1.24.2.

Cool, thanks.

This is much less important as it's not actually a correctness thing, but my name wasn't listed in the contributors for 1.24.1 yet there are two of my PRs listed in the features for the Java interface. Is it possible to update the release note?

[ChrisAReed]

For the onnxruntime-linux-x64-gpu-1.24.1.tgz package available on the release assets, the TensorRT/CUDA provider libraries are only linking against CUDA 12 packages.
Is there a similar package available that enables CUDA 13 support to download for Linux? I see there are different versions to download for CUDA 12 vs 13 for python.

[tianleiwu]

For the onnxruntime-linux-x64-gpu-1.24.1.tgz package available on the release assets, the TensorRT/CUDA provider libraries are only linking against CUDA 12 packages. Is there a similar package available that enables CUDA 13 support to download for Linux? I see there are different versions to download for CUDA 12 vs 13 for python.

@ChrisAReed, I've added onnxruntime-linux-x64-gpu_cuda13-1.24.1.tgz for cuda 13.