MODELS
by Cohere
Cohere’s multimodal embedding model for text, images, and mixed documents, with 128K context and configurable embedding dimensions for enterprise search/RAG.
embed-v4.0 is Cohere’s multimodal embedding model for creating embeddings from text, images, and mixed documents. The supplied Cohere launch page and model documentation describe it as supporting a 128K context window and configurable embedding dimensions, positioning it for enterprise search and retrieval-augmented generation workflows where content may include both language and visual document inputs.
Use embed-v4.0 when you need embeddings for text, images, or mixed documents from a Cohere model. Consider it for enterprise search or RAG workflows that benefit from long-context embedding support and configurable embedding dimensions.
Last updated Jun 12, 2026
