Introducing @huggingface/kernels: 200+ WebGPU Kernels for Local AI

Hugging Face announced @huggingface/kernels, a library containing over 200 WebGPU‑optimized kernels aimed at running AI models locally. The release targets developers who want GPU acceleration directly in browsers or other WebGPU‑compatible environments. The initiative is intended to simplify deployment and boost performance for local AI applications.

Summary written by Kernelia from the original article by Hugging Face Blog. The story and its rights belong to its author.