Liquid AI releases LFM3, a family of small multimodal models
Liquid AI has released LFM3, its new generation of Liquid Foundation Models, with text, vision, audio, omni, and decision models that share one backbone, all in the roughly 2B parameter range. The company says LFM3-1.7B outperforms its predecessor, LFM2.5-2.6B, while using 35% fewer parameters. The text model fits in about 1 GB at 4-bit precision and has a 256K context window, and every model can be fine-tuned on a single GPU. All LFM3 models are open-weight and available on Hugging Face.