Liquid AI has unveiled LFM3, a family of six specialized foundation models spanning text, audio, vision, and omni modalities. The models use a tri-hybrid architecture optimized for on-device deployment and can be fine-tuned on a single GPU.
FireBrief
FireBrief