Discover the new WebNN Developer Preview! It’s a cross-platform web standard for running machine learning models in browsers with hardware acceleration. Now supporting DirectML and ONNX Runtime Web, it promises enhanced web app capabilities without compromising privacy.-

Introducing the Future of Web Applications

The tech world is buzzing with the latest announcement from the Windows Developer Blog. The launch of the WebNN Developer Preview, leveraging DirectML and ONNX Runtime Web, marks a significant leap forward in web development. This innovative technology enables cross-platform and hardware-accelerated neural network inference directly in the browser.

What’s New with WebNN?

WebNN stands as a groundbreaking web standard designed for running machine learning models in the browser, utilizing the local device’s GPU or NPU for hardware acceleration. This advancement opens the door to a new era of web applications, including generative AI, object recognition, and natural language processing, without the need for additional software or plugins.

“WebNN is a game-changer for web development.”

Major Updates: DirectML and ONNX Runtime Web Integration

One of the most exciting aspects of WebNN is its integration with DirectML, offering performant, cross-hardware ML acceleration across Windows devices. This integration not only enhances performance but also ensures reliability and scalability across different hardware platforms.

“With WebNN, you can unleash the power of ML models in your web app.”

What’s Important to Know

For developers eager to dive into this new technology, the WebNN Developer Preview is available on DirectX 12 compatible GPUs, requiring Windows 11 version 21H2 or newer, ONNX Runtime Web minimum version 1.18, and Microsoft Edge Canary channel with the WebNN flag enabled. This setup paves the way for running ONNX models in the browser with minimal code changes and maximum hardware acceleration.

Support from Major Hardware Vendors

Major hardware vendors including AMD, Intel, and NVIDIA have expressed their excitement and support for the launch of WebNN with DirectML. This collaborative effort underscores the industry’s commitment to advancing web development technologies and enhancing the performance of web applications on a wide range of hardware.

Getting Started with WebNN

Developers looking to explore the capabilities of WebNN and DirectML can start by checking out various Build sessions and downloading the latest drivers from their preferred hardware vendors. This initial step into the WebNN ecosystem promises to revolutionize the way web applications are developed and experienced.

As we stand on the brink of this new web development era, the possibilities seem endless. The WebNN Developer Preview with DirectML not only signifies a major technological advancement but also a new chapter in how we interact with web applications.

• WebNN enables machine learning directly in the browser using GPUs or NPUs without extra software.
• Supports Intel® Core™ Ultra processors with AI Boost and Qualcomm® Hexagon™ NPUs.
• DirectML backend enhances cross-hardware ML acceleration across Windows devices.
• Combines with ONNX Runtime Web for running ONNX models in browsers efficiently.
• Requires Windows 11, version 21H2 or newer, and Microsoft Edge Canary with WebNN flag enabled.

From the Windows Blog

---