Cohere Embed V4 is a multimodal embedding model that converts text, images, and complex business documents into vectors for RAG and semantic search with 128K context and multilingual support for 100+ languages.