What it does
This is Liquid AI's own model family, not Jev. The company reports d1-3B scoring 48.57 on Decision Index v0.2.1 and answering one question in 8 ms on an RTX 4090; the experimental d1-omni-600M also accepts audio. Weights are on Hugging Face under Liquid's own lfm1.0 licence rather than a standard open-source licence.