Microsoft Releases Windows ML: A Unified AI Platform for Windows 11

Microsoft has just announced the general availability of Windows ML – a runtime platform optimized for on-device AI model inference. This is a tool that helps developers build AI applications for Windows 11 PCs without worrying about hardware differences.

 

Previously, to develop AI applications on Windows, programmers were forced to build custom support layers to ensure software compatibility with many different chip lines, from Qualcomm, Intel, AMD to NVIDIA. Windows ML was born to completely solve this problem by providing a unified framework, allowing applications to run smoothly on millions of Windows 11 PCs without the need for individual hardware optimization.

According to Microsoft, Windows ML offers several notable benefits:

• Support for a wide range of hardware architectures through infrastructure APIs and execution providers (EPs) available or automatically downloadable.
• Enables model precompilation (AOT) to provide a more seamless end-user experience.
• Runtime and EP do not need to be packaged directly in the application, thanks to Windows ML automatically downloading depending on the machine hardware.
• Always maintain long-term compatibility, keeping your apps running smoothly even as hardware continues to evolve.
• Allows developers to assign device policies, optimizing performance according to needs: battery saving (NPU), high performance (GPU) or choosing a specific silicon type.

Microsoft said many major software companies have already adopted Windows ML for their AI applications, including Adobe, BUFFERZONE, Dot Inc., McAfee, Reincubate, Topaz Labs, and Wondershare.

Windows ML is now included in the Windows App SDK, supporting all PCs running Windows 11 24H2 or later. Developers can start building production-ready AI apps right away.