Liquid AI: New Multilingual Search Models for 11 Languages

Liquid AI has released two new retrieval models: LFM2.5-ColBERT-350M and LFM2.5-Embedding-350M. Both models have 350 million parameters and are the first bidirectional members of the LFM family. What's interesting is that these models are designed for fast multilingual and cross-lingual search across 11 languages. They have a small footprint, meaning they can run almost anywhere. LFM2.5-Embedding-350M is a dense bi-encoder, converting each document into a single vector for the fastest search. LFM2.5-ColBERT-350M is a late-interaction model, converting each token into a vector for higher accuracy, but with a larger index. The bottom line is that these new models offer flexible options for efficient and accurate search in various applications, from product catalogs to support documentation.