WebNN
A Web standard promoted by the W3C that provides a unified interface for accessing AI acceleration hardware such as device CPUs, GPUs and NPUs, analogous to how WebGL abstracts graphics hardware.
Timeline (last 90 days)
WebNN attempts to provide a unified web API for neural network inference.
The applicable boundaries of the edge-first architecture include: model capability is limited by size, and complex reasoning tasks still require large cloud models; the latency and bandwidth cost of the first load is shifted to the user; WebNN browser support is still evolving, and cross-device performance varies significantly.
Browser-side technologies such as WebGPU and WebAssembly have made it possible to run AI models in web pages.
All Facts (3)
WebNN attempts to provide a unified web API for neural network inference.
50%UnverifiedThe applicable boundaries of the edge-first architecture include: model capability is limited by size, and complex reasoning tasks still require large cloud models; the latency and bandwidth cost of the first load is shifted to the user; WebNN browser support is still evolving, and cross-device performance varies significantly.
50%UnverifiedBrowser-side technologies such as WebGPU and WebAssembly have made it possible to run AI models in web pages.
50%