Using Microsoft.Windows.AI.MachineLearning

The Microsoft.Windows.AI.MachineLearning package provides Windows Machine Learning APIs for on-device AI inference using ONNX Runtime.

Requirements

• Minimum OS: Windows 10 19H1 (build 18362) or later.
• For Windows 10 versions prior to 19H1 (build 18362), use the Microsoft.WindowsAppSDK.ML package instead, which provides framework-based deployment via the Windows App SDK runtime.

Getting Started

Add a reference to Microsoft.Windows.AI.MachineLearning in your project. No additional configuration is required for unpackaged apps on Windows 10 19H1+ and Windows 11.

Native C++

#include <winrt/Microsoft.Windows.AI.MachineLearning.h>

.NET

using Microsoft.Windows.AI.MachineLearning;

Deployment

Binaries are deployed alongside your application:

MyApp/
├── MyApp.exe
├── Microsoft.Windows.AI.MachineLearning.dll
├── onnxruntime.dll
└── DirectML.dll

WinRT activation relies on OS-native reg-free WinRT, available on Windows 10 19H1+.

Using ONNX Runtime Headers

The ONNX Runtime headers are included in a winml subdirectory:

#include <winml/onnxruntime_c_api.h>
#include <winml/onnxruntime_cxx_api.h>

To use headers without the winml/ prefix:

<PropertyGroup>
  <WinMLEnableDefaultOrtHeaderIncludePath>true</WinMLEnableDefaultOrtHeaderIncludePath>
</PropertyGroup>

WindowsAppSDK Integration

For WindowsAppSDK consumers or packaged (MSIX) applications, use Microsoft.WindowsAppSDK.ML instead. It depends on this package transitively and provides additional integration with the Windows App SDK runtime.

More Information

• Windows ML documentation
• Windows App SDK
• WindowsAppSDK-Samples