Install and deploy Windows ML

For framework-dependent (recommended), install:

• Microsoft.WindowsAppSDK.ML
• Microsoft.WindowsAppSDK.Runtime
• Then, see Deployment for framework-dependent apps for how to deploy the runtime

For self-contained (increases app size by ~41 MB):

• For apps targeting Windows 10 Build 18362 or higher, install Microsoft.Windows.AI.MachineLearning
• Or, for apps targeting Windows 10 Build 17763 or higher, install Microsoft.WindowsAppSDK.ML
• Do not install Microsoft.WindowsAppSDK.Runtime or the main Microsoft.WindowsAppSDK package

For framework-dependent (recommended), install:

• Microsoft.WindowsAppSDK.ML
• Microsoft.WindowsAppSDK.Runtime
• Then, see Deployment for framework-dependent apps for how to deploy the runtime

For self-contained (increases app size by ~41 MB):

• For apps targeting Windows 10 Build 18362 or higher, install Microsoft.Windows.AI.MachineLearning
• Or, for apps targeting Windows 10 Build 17763 or higher, install Microsoft.WindowsAppSDK.ML
• Do not install Microsoft.WindowsAppSDK.Runtime or the main Microsoft.WindowsAppSDK package

For self-contained (framework-dependent is not supported for C/C++), reference Microsoft.Windows.AI.MachineLearning and configure CMake to use its build/cmake directory. See Self-contained installation below for the full CMake setup.

For framework-dependent, install the required packages:

pip install wasdk-Microsoft.Windows.AI.MachineLearning[all] wasdk-Microsoft.Windows.ApplicationModel.DynamicDependency.Bootstrap onnxruntime-windowsml

And then install the Windows App SDK Runtime that matches the version of the wasdk- packages.

• .NET 8 or greater
• Target framework moniker net8.0-windows10.0.17763.0 or greater if using Microsoft.WindowsAppSDK.ML, or target framework moniker net8.0-windows10.0.18362.0 or greater if using Microsoft.Windows.AI.MachineLearning

• C++20 or later

• Visual Studio 2022 with C++ workload
• CMake 3.21 or later

• Python 3.10–3.13 on x64 or ARM64
• An unpackaged Python installation (python.org or winget — not the Microsoft Store)

In your project, install or update the following NuGet packages:

• Microsoft.WindowsAppSDK.ML (this adds the Windows ML APIs)
• Microsoft.WindowsAppSDK.Runtime (this enables framework-dependent mode)

Also, in your .csproj file, ensure that WindowsAppSDKSelfContained is either unspecified (not present), or false, so that it uses framework-dependent deployment.

In your project, install or update the following NuGet packages:

• Microsoft.WindowsAppSDK.ML (this adds the Windows ML APIs)
• Microsoft.WindowsAppSDK.Runtime (this enables framework-dependent mode)

Also, in your project file, ensure that WindowsAppSDKSelfContained is either unspecified (not present), or false, so that it uses framework-dependent deployment.

Framework-dependent C/C++ support is not currently available. Either use C++/WinRT, or self-contained as described below.

Install the required packages:

pip install wasdk-Microsoft.Windows.AI.MachineLearning[all] wasdk-Microsoft.Windows.ApplicationModel.DynamicDependency.Bootstrap onnxruntime-windowsml

And then install the Windows App SDK Runtime that matches the version of the wasdk- packages.

In your project, install or update the following NuGet packages:

• For apps targeting Windows 10 Build 18362 or higher:
  • Install Microsoft.Windows.AI.MachineLearning (this adds the Windows ML APIs and binaries, defaults to self-contained)
• Or, for apps targeting Windows 10 Build 17763 or higher:
  • Install Microsoft.WindowsAppSDK.ML (this adds the Windows ML APIs and binaries, defaults to self-contained, and adds RegFree WinRT support for 17763)
• Do NOT install Microsoft.WindowsAppSDK.Runtime (this would enable framework-dependent mode... remove it if you currently have it installed)
• Do NOT install the main Microsoft.WindowsAppSDK NuGet package (this would enable framework-dependent mode... remove it if you currently have it installed)
• Feel free to install other Windows App SDK component packages like Microsoft.WindowsAppSDK.AI, as long as you don't install the Runtime or the main package

In your project, install or update the following NuGet packages:

• For apps targeting Windows 10 Build 18362 or higher:
  • Install Microsoft.Windows.AI.MachineLearning (this adds the Windows ML APIs and binaries, defaults to self-contained)
• Or, for apps targeting Windows 10 Build 17763 or higher:
  • Install Microsoft.WindowsAppSDK.ML (this adds the Windows ML APIs and binaries, defaults to self-contained, and adds RegFree WinRT support for 17763)
• Do NOT install Microsoft.WindowsAppSDK.Runtime (this would enable framework-dependent mode... remove it if you currently have it installed)
• Do NOT install the main Microsoft.WindowsAppSDK NuGet package (this would enable framework-dependent mode... remove it if you currently have it installed)
• Feel free to install other Windows App SDK component packages like Microsoft.WindowsAppSDK.AI, as long as you don't install the Runtime or the main package

The C APIs ship in the Microsoft.Windows.AI.MachineLearning NuGet package. The package includes CMake config files under build/cmake/ for direct consumption.

To use the NuGet package...

1. Download or restore the NuGet package (e.g., via nuget restore or manual download)
2. Extract or reference the package root directory
3. Set CMAKE_PREFIX_PATH to the build/cmake subdirectory of the package

Example CMakeLists.txt

cmake_minimum_required(VERSION 3.21)
project(MyApp LANGUAGES CXX)

set(CMAKE_CXX_STANDARD 20)
set(CMAKE_CXX_STANDARD_REQUIRED ON)

find_package(microsoft.windows.ai.machinelearning CONFIG REQUIRED)

add_executable(MyApp main.cpp)
target_link_libraries(MyApp PRIVATE WindowsML::Api WindowsML::OnnxRuntime)

add_custom_command(TARGET MyApp POST_BUILD
    COMMAND ${CMAKE_COMMAND} -E copy_if_different
        $<TARGET_RUNTIME_DLLS:MyApp> $<TARGET_FILE_DIR:MyApp>
    COMMAND_EXPAND_LISTS
    VERBATIM
)

Example build commands

# Point CMAKE_PREFIX_PATH to the NuGet package's cmake directory
cmake -B build -S . -DCMAKE_PREFIX_PATH="C:/packages/Microsoft.Windows.AI.MachineLearning/build/cmake"
cmake --build build --config Release

The NuGet package ships its own cmake config under build/cmake/ that understands the NuGet directory structure (architecture-specific directories under lib/native/ and runtimes/). Architecture is determined from CMAKE_GENERATOR_PLATFORM, CMAKE_VS_PLATFORM_NAME, or CMAKE_SYSTEM_PROCESSOR.

The vcpkg portfile generates a separate, thin config file that maps to the standard vcpkg layout (lib/ and bin/). Both configs share the same targets file (microsoft.windows.ai.machinelearning-targets.cmake) which defines the IMPORTED targets.

The same targets (WindowsML::WindowsML, WindowsML::Api, WindowsML::OnnxRuntime, WindowsML::DirectML) are available in both consumption modes. Use the standard CMake 3.21 $<TARGET_RUNTIME_DLLS> generator expression to copy runtime DLLs to the build output directory.

Self-contained deployment is not applicable for Python. Use the framework-dependent pip packages described above.