Unlock a new era of innovation with Windows Copilot Runtime

Build intelligent AI experiences on Windows and scale across a breadth of Windows hardware.

Utilize powerful AI functionality with Windows Copilot Runtime APIs

Transform your Windows experiences with a powerful set of on-device APIs.

Get started with Windows Copilot Runtime APIs

  • Phi Silica

    Enable your Windows app to connect with the on-device Phi model for natural language processing tasks (chat, math, code, reasoning, etc.) using the Windows App SDK.

    Learn more

  • Text recognition (OCR)

    Detect, recognize, and extract text from images.

    Learn more

  • Recall

    Recall utilizes AI to help users find almost anything they’ve seen on their PC by tapping into Windows UserActivity API.

    Learn more

  • Studio Effects

    Apply special effects to the device camera with the Studio Effects API.

    Learn more

Deploy custom AI models into your Windows app

From native to web apps, leverage DirectML and WebNN to bring AI innovation to your customers across Windows

Get started with custom models

  • ONNX and ONNX Runtime

    The ONNX format is an open standard that enables ML models to run on Windows devices with ONNX Runtime.

  • PyTorch

    PyTorch with DirectML enables training and inference of complex machine learning models on a wide range of DirectX 12-compatible hardware.

  • WebNN

    The WebNN API leverages the DirectML API on Windows to access the native hardware capabilities and optimize the execution of neural network models.

  • Samples

    Get started with our wide-range of C# and Python samples.

Abstract image of a Windows wallpaper

Scale AI innovations with ease with DirectML

DirectML is a single, cross-hardware DirectX API for hardware accelerated machine learning on Windows. DirectML provides GPU and NPU acceleration across a broad range of supported hardware from AMD, Intel, Nvidia, and Qualcomm. DirectML also connects with ONNX Runtime, PyTorch, and WebNN to enable amazing AI experiences on native and web apps.

Get started with DirectML

Train and fine-tune models on Windows

The AI Toolkit for Visual Studio Code provides tools and access to a model catalog to help jump-start local AI development and deployment.

Get started with AI Toolkit
Image of the VS Code AI Toolkit

Learn how others are successful with AI

Image of the Topaz Labs logo

Topaz Labs empowers creative professionals with innovative AI-powered photo and video enhancement software. Leveraging ONNX Runtime, Olive, and DirectML, we tap into the increased processing speeds offered by AMD GPUs. Our strong collaboration with AMD and Microsoft enables us to locally optimize large models like Stable Diffusion for AMD laptops equipped with integrated GPUs. Thanks to this powerhouse trio, our stable diffusion models achieve a remarkable 4X speed improvement on AMD GPUs. - Dr. Suraj Raghuraman, Head of AI Engine, Topaz Labs