LFM2.5-2.6B Released: Small Model Beats Larger Counterparts

The newly released LFM2.5-2.6B model outperforms larger models in instruction following and tool use, utilizing a training pipeline of SFT, online distillation, and agentic RL.

2026-08-04 ~ 2026-08-04 · 3 related posts

• LFM2.5-2.6B released: full agent training pipeline compressed into 2.6B parameters — SergioPaniego · 2026-08-04
• LFM2.5 training details: MOPD and agentic RL are the two most interesting stages — SergioPaniego · 2026-08-04
• 2.6B Model Beats Larger Models: Agent Training Insights — SergioPaniego · 2026-08-04