Cohere Embed v4 released: Multimodal embeddings and 128k context

AI Impact Summary

Cohere has released Embed v4, a significant update to their embedding models that introduces multimodal capabilities (text and images) and a substantially increased context length of 128k tokens. This expands Cohere's offerings and improves performance across various retrieval tasks, particularly those involving mixed modalities. While this is an additive release, users should evaluate the benefits of this new capability for their applications.