Liquid AI LFM

Liquid Foundation Models, or LFM, are Liquid AI’s open-weight models designed for efficient inference on phones, laptops, edge devices, CPUs, and cloud hardware. The family includes compact text, vision-language, audio-language, reasoning, retrieval, encoder, and mixture-of-experts variants. LFM models emphasize low memory use, fast local execution, tool calling, privacy, and offline agent workflows. Developers can download the weights, test them in Liquid’s playground, or deploy them through the LEAP edge platform and open-model runtimes.