Deploy local agents everywhere with LFM2.5-2.6B

LFM2.5-2.6B is a small artificial intelligence model designed to run agents directly on local devices like laptops and phones rather than relying on cloud computing. The model supports tool calling and multi-step workflows while using minimal memory and processing power, which allows developers to keep user data private on devices and avoid costs associated with cloud inference. According to benchmark testing, it performs competitively on tasks like tool use and instruction following despite being smaller than comparable models, though larger models maintain an advantage on coding tasks. The model was trained using specialized techniques including supervised fine-tuning, teacher specialization, multi-domain distillation, and reinforcement learning within agent environments to improve its ability to work with different tools and multi-turn interactions.