The latest Data Phoenix news tagged LFM2.5-2.6B.
Liquid AI released LFM2.5-2.6B, a 2.6-billion-parameter on-device agentic model the company says runs at 220 tokens per second on an Apple M5 Max.