ONNX

2 Articles

LFM2-MoE: An 8.3B Parameter Language Model Running Entirely in Your Browser via WebGPU

Liquid AI's LFM2-MoE loads 8.3B parameters (1.5B active) and generates text client-side through WebGPU. No server, no API key. A hybrid convolution-attention architecture running Mixture-of-Experts inference in a browser tab, powered by Transformers.js.

Mar 31, 2026

MasterSelects: A GPU-First Video Editor Built Entirely on WebGPU

A full video editor running in the browser with zero CPU roundtrip. 37 blend modes in one WGSL shader, GPU-accelerated scopes, optical flow scene detection, and SAM2 segmentation on-device. 13 dependencies. MIT licensed.

Mar 18, 2026