LFM2.5-24B-A2B: Built for fast on-device agents

Liquid AI released LFM2.5-24B-A2B, a 24-billion parameter reasoning model with only 2 billion active parameters per token, designed for fast on-device agents that runs up to 1.7× faster than comparable models while matching or exceeding the performance of larger competitors on agentic benchmarks. The open-weight model features an expanded context window (131K tokens), doubled vocabulary, and was trained with reinforcement learning to excel at tool-use and reasoning tasks. The release completes the LFM2.5 family spanning from 230M to 24B parameters and is available on Hugging Face.