Liquid AI launches d1-3B and d1-omni-600M for multimodal decisions

First reported by Hugging Face at · Updated · 3 sources

The models accept different input combinations: d1-3B handles text and images, while the experimental 600M model pairs text with either images or audio. According to MarkTechPost, both have open weights and return calibrated, typed answers in a single forward pass rather than generating text, using zero output tokens.

  • Hugging Face describes the models as designed for edge use.

Covered by 3 publishers within 1 hour of the first report: 2 news reports and 1 from the company.

Reporting2

RuntimeWire Liquid AI releases open d1 models for multimodal edge decisions · Ryan Merket
MarkTechPost Liquid AI Releases Open-Weight d1-3B and d1-omni-600M: Multimodal Decision Models With Zero Output Tokens · Asif Razzaq

From the company1

Hugging Face Multimodal open d1 decision models for the edge