Liquid AI Ships Two 350M Multilingual Search Models

Liquid AI released LFM2.5-Embedding-350M and LFM2.5-ColBERT-350M, both 350M-parameter bidirectional models for fast multilingual search across 11 languages. The dense bi-encoder offers the fastest search and smallest index. The late-interaction ColBERT variant matches queries word-by-word for higher accuracy. Both are available on Hugging Face under the LFM Open License v1.0.