Issue 2026-11-15 · Industry · 开源 · 应用
LFM2.5-24B-A2B shines paired with Hermes, user says
LiquidAI's LFM2.5-24B-A2B is praised as one of few models balancing speed and capability on commodity hardware, suited as a task executor behind Hermes rather than a coding model. The user reports ~37 tokens/s at zero context on a 780M iGPU and expects it to be usable on CPU too.
Read original ↗