Tag
The article highlights the new WebGPU backend in llama.cpp/ggml, enabling GPU-accelerated local AI model inference in browsers, developed by Reese Levine and team at USCS over the past year and a half.
Created an in-browser WebGPU version of the classic char-rnn demo training on Shakespeare, paying homage to Karpathy joining Anthropic.
A developer demonstrates running the Qwen3.6-27B AI model entirely on WebGPU in a browser, though speed is not optimal.
Demonstrates running Gemma 4 offline in the browser using WebGPU and Transformers.js to control a Reachy Mini robot via WebSerial.
Opus 4.7 auto-generated a custom WebGPU kernel that accelerates Qwen3.5 inference up to 13× via fused LinearAttention, now shipping in Transformers.js v4.2.0.
A 3D building editor built with React Three Fiber and WebGPU, structured as a Turborepo monorepo with core, viewer, and editor packages for schema definitions, 3D rendering, and interactive tools.
Brush is an open-source 3D reconstruction engine using Gaussian Splatting, built in Rust and compatible with WebGPU for cross-platform real-time rendering on desktop, mobile, and browser.
OpenReel Video is an open-source, browser-based professional video editor built with React, TypeScript, WebCodecs, and WebGPU that runs entirely client-side without uploads or installations.