3 articles found
LFM3-1.7B fits in 1GB, runs on a phone, and beats models 5x its size at agentic tasks. The edge AI race just got a new frontrunner.
A developer runs LFM2.5-2.6B on a OnePlus 13 at 17 tok/s with pure CPU inference. This is what happens when edge AI stops being a demo and becomes production.
How Qwen 3.5 0.8B manages complex spatial reasoning and action execution on smartwatch-grade hardware, and what it means for the future of edge AI.