Agentic Kernel Optimization at the scale of the web

LFM2.5 230M from Liquid AI running entirely locally in the browser on WebGPU, with every kernel optimized by Fable 5 and Opus 4.8. Private by design: no data ever leaves your machine.

INITIALIZING LFM2.5
Fetching LFM2.5 weights0%
Checking WebGPU…

Ready · on-device

What's on your mind today?

LFM2.5 runs entirely on your device.