Liquid AI Introduces LFM2.5-24B-A2B for Fast, Private On-Device Agents

Summary

Liquid AI’s page presents LFM2.5-24B-A2B as a model for running fast, private agents directly on devices. The accompanying model metadata identifies LiquidAI/LFM2-24B-A2B as a 24B text-generation model and says it was updated on Aug. 5. The metadata also shows 2.91k and 354, without explaining what those figures represent. The page title and model metadata use slightly different model-name forms, so the available extract does not establish whether they refer to exactly the same release.