Cohere offers the Rerank model, which takes a search query and a list of documents and re-sorts them by relevance so the most useful results appear first. It is used inside enterprise search systems and AI agent pipelines to improve the precision of information retrieval.

The Rerank API receives a query and a list of candidate documents, then scores and reorders them by semantic relevance using a cross-encoder architecture that processes the query and each document together. The current version, Rerank 4 (released December 2025), supports a 32K context window, over 100 languages, and a self-learning capability that allows users to customize ranking preferences without additional labeled data. It is available on the Cohere Platform, AWS SageMaker, Microsoft Azure AI Foundry, and as a private deployment.