Skip to content

Windows AI

The windows provider support AI for Windows Apps which provides state-of-the-art local models, with NPU hardware support.

  1. Install the AI Toolkit for Visual Studio Code extension.

  2. Open the Model Catalog and add a model from the ONNX Models runtime section.

  3. Right click on the model in the Explorer view and select Copy model name

  4. Set the model name in your script to the model name you copied.

    script({
    model: "windows:Phi-4-mini-gpu-int4-rtn-block-32",
    })

See Azure AI Toolkit getting started guide.