MongoDB - Build AI That Scales
Mini toolsLFM2.5-Embedding-350M
By Liquid AI
Model family: LFM
LFM2.5-Embedding-350M is a bidirectional encoder adapted from LFM2.5-350M-Base using CLS-style pooling to produce a single dense embedding per document. It shares the same backbone and three-stage training recipe as LFM2.5-ColBERT-350M: large-scale English contrastive pretraining, multilingual and cross-lingual distillation with additional cross-lingual data, and hard-negative fine-tuning. Receives slightly more cross-lingual training data than the ColBERT variant. Supports Arabic, German, English, Spanish, French, Italian, Japanese, Korean, Norwegian, Portuguese, and Swedish. Achieves competitive results on NanoBEIR Multilingual and MKQA-11, trailing only the ColBERT variant. Optimized for fastest search speed and smallest index footprint. Well-suited for product catalogs, FAQ knowledge bases, and support documentation. GGUF available for llama.cpp on CPUs, laptops, and edge devices.
New
Text
Gen 7
Overview
350M-parameter bidirectional multilingual bi-encoder that maps each document to a single dense vector for fast, cost-efficient multilingual and cross-lingual search. Supports 11 languages. Built from LFM2.5-350M-Base. Produces the smallest, cheapest index among LFM2.5 retrieval models. Available in GGUF format for CPU and edge deployment.
About Liquid AI
Liquid AI is an MIT spin-off building efficient general-purpose AI models (Liquid Foundation Models, or LFMs) that run on edge devices with less memory and power.
They recently raised $250M in Series A funding to scale model development and deployment.
Tools using LFM2.5-Embedding-350M
No tools found for this model yet.
Last updated: June 19, 2026
