A developer known as MAAN built a browser-based demo that reimplements Nvidia's DLSS 5 neural rendering network using WebGPU compute shaders, hosted on Cloudflare Workers with adjustable settings and sample scenes. The port reportedly works on non-Nvidia GPUs and even macOS, using roughly 147MB of extracted model weights alongside a small JavaScript runtime, though each rendered frame currently takes one to two seconds. MAAN says the weights came from a leaked DLSS 5 library file and plans to release the source code on GitHub soon.
tomshardware.com
· 2026-09-18
A security researcher published a proof-of-concept called the Deathray, a tiny WebGPU program that locks up a Mac's graphics pipeline after a single click on a malicious webpage. The bug reproduces across Chrome, Firefox and Safari on Apple Silicon Macs running macOS Tahoe, but does not cause the same lasting freeze on other operating systems. An infinite compute-shader loop feeding a shared buffer into a render shader appears to be the trigger, forcing victims to hard-restart their machine.
auberon.xyz
· 2026-09-10
A developer released gpu-lexer, a 27.5KB JavaScript package that uses a small WebGPU-run neural model to perform syntax highlighting without relying on language-specific grammars. It tokenizes code into words, whitespace and symbols, then predicts each token's category—such as keyword, string or function—based on surrounding context, aiming to work even on languages it has never seen before.
gpu-lexer.vercel.app
· 2026-09-09
WebLLM is an open-source engine that runs large language models entirely inside a web browser, using WebGPU for hardware acceleration instead of relying on a server. It offers full compatibility with the OpenAI API, including streaming, JSON-mode structured output, and support for models like Llama 3, Phi 3, Gemma, Mistral, and Qwen. Developers can install it as an npm package and build applications on top of it, or integrate custom models in MLC format.
github.com
· 2026-09-02
An independent developer has released pngine, a declarative language and runtime built over 2.5 years that lets programmers describe WebGPU shaders and pipelines using S-expressions mirroring the WebGPU spec directly. The tool validates configurations via WGSL reflection without needing a live GPU context, offers an LSP for real-time error checking, and can export projects as a single HTML file, a ZIP, or even a self-contained PNG image holding compiled bytecode.
hugodaniel.com
· 2026-08-23
GoKawiil