LFM2.5 – Advanced on-device AI model optimized for edge deployment

Summary: LFM2.5 is Liquid AI's latest edge AI model family, integrating text, vision, and audio capabilities within a 1 billion parameter range. It delivers an 8x speedup in audio processing and includes optimizations for AMD and Qualcomm NPUs, enabling efficient on-device AI performance.

What it does

LFM2.5 runs a full modal stack optimized for edge devices, improving latency and reliability through hardware-specific enhancements for AMD and Qualcomm processors.

Who it's for

It targets developers and manufacturers deploying AI agents on edge devices, including AI PCs and mobile hardware.

Why it matters

It addresses edge AI constraints by delivering multimodal AI with low latency and hardware efficiency for scalable on-device applications.