LFM2.5-24B-A2B: Fast On-Device AI Agents

Today, we release LFM2.5-24B-A2B, a 24B-parameter reasoning model with just 2B active parameters, built for fast, private agents on-device. It is the largest and final model in the LFM2.5 family, which now spans two orders of magnitude, from LFM2.5-230M to LFM2.5-24B-A2B.

LFM2.5-24B-A2B is competitive with Qwen3.6-35B-A3B, a model with ~50% more total and active parameters, on five of eight agentic and tool-use benchmarks. It scores 48.09 on Live Claw versus 38.19, and 27.69 on Automation Bench versus 8.72. Because only 2B parameters are active per token, it runs up to 1.7× faster than comparable MoEs on-device, reaching 184 tokens/s on an Apple M5 Max and up to 21K output tokens/s on a single NVIDIA B300 GPU.

The base (LFM2.5-24B-A2B-Base) and post-trained (LFM2.5-24B-A2B) models are open-weight and available today on Hugging Face and the company's Playground. Documentation on how to run and fine-tune the model locally is also available.

Compared with its predecessor, LFM2-24B-A2B, the new model features an expanded vocabulary and context window, scaled-up pretraining, and large-scale reinforcement learning. It is a reasoning-only model that produces an explicit chain of thought before its final answer, a strategy adopted to leverage its high decode speed and improve answer quality.

Source: Hacker News · Summarized by HeadlinesBriefing