models for turning text, images, audio (and combos) into useful vectors or feature maps. Ideal for search/RAG, clustering, recommendation, retrieval.