lfm3

A dependency-free C inference engine for Liquid AI's LFM2 / LFM2.5 models. Terminal chat, an OpenAI-compatible server, a web agent and a terminal agent, all in one executable.

Run LiquidAI's LFM3 collection fully in your browser, on the CPU: chat, images, speech in and out, and d1's structured answers, at up to 97 tokens/s. Rust compiled to WebAssembly with fusor-ai. No server, no GPU.

在安卓（骁龙 8 Gen 1）上跑 Liquid AI LFM3-1.7B：预编译 arm64 二进制、内置聊天页、实测性能与调优记录

Liquid AI Just Dropped LFM3: The 1GB Local AI Agent That Beats Qwen 9B! - 256K context tri-hybrid edge agent

Public install-and-run distribution for the local-first Audio Report AI Ubuntu desktop app

Liquid AI LFM3の全6モデルをColab T4で試す実験ノート。日英ガイド、実測VRAM・速度・失敗例、再現用CLIを公開。