Show HN: Kiln – A WebGPU alternative to transformers.js, 2.2× faster (median)
What happened
Kiln is a JavaScript library that lets web browsers run large language, vision and speech models directly on the client using WebGPU. It reads model files such as safetensors, pytorch_model.bin or GGUF from the Hugging Face Hub or any static server, quantizing them to 4‑bit or 8‑bit on load, and requires no ONNX conversion, compile step, or inference server. The API mirrors transformers.js, so existing code can switch by changing the import and model ID. Kiln caches downloads, streams results from a worker, and falls back to CPU‑based WASM SIMD when WebGPU isn’t available, all within a 241 KB gzipped module.
Why it matters
For ordinary readers, this means AI features like chat, translation or image analysis can run locally in the browser without sending data to external servers, improving privacy and speed on modest hardware. It also lowers the barrier for developers to embed AI into web apps without complex backend setups.
Reality check
An automated, sceptical read of the sourceIncremental. The project adapts existing transformer APIs to WebGPU and offers modest speed gains on older laptops, but it builds on known concepts rather than introducing fundamentally new technology.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More AI news
Every Way to Get 192GB for Local AI (Without a $9,500 Mac) [video]Industry news · 14 Nov 2026DeepSeek V4 Pro 0813DeepSeek · 14 Nov 2026DeepSeek V4 Pro 0813DeepSeek · 14 Nov 2026Show HN: Portal2Bench – Can AI agents solve Portal 2?Industry news · 14 Nov 2026Anthropic Upgrades Claude Code Projects to Run Parallel AI Workers Hands-FreeAnthropic · 14 Nov 2026This page was collected from github.com and summarised automatically by AIPulsee's pipeline. See how it runs.