AI company Cohere has added multimodal capabilities to its Embed 3 embedding model, allowing images to be included in RAG-based company searches. This is reported by Emilia David for VentureBeat. The new version can create embeddings for both images and text, with both formats stored in a unified database. According to Cohere, this allows companies to search even complex reports, product catalogs, and design files accurately and quickly. The model is available in more than 100 languages and can be used through both the Cohere platform and Amazon SageMaker.