WeMM-Embedding: Multimodal Embedding Family up to 9B Parameters
August 24, 2026
WeMM-Embedding provides universal multimodal embeddings for text, images, video, and interleaved inputs. The model family includes 2B, 4B, and 9B variants trained via large-scale alignment and fine-grained relevance refinement.
HOW THIS AFFECTS YOU
●
builderYou can use these embeddings for cross-modal retrieval, recommendation, and agentic systems across various media types.
●
founderThis provides a scalable foundation for building multimodal search and RAG applications.