# Cohere launches Rerank 5 models for enterprise search

> Cohere’s new Rerank 5 Pro and Fast models are now generally available, offering passage‑level relevance and lower token costs.

Oossa · 2026-10-21 · https://oossa.com/en/cohere-launches-rerank-5-models-for-enterprise-search

Cohere announced today that its Rerank 5 models are generally available. The two tiers – Rerank 5 Pro and Rerank 5 Fast – let enterprises retrieve the most relevant passages from documents, cutting the number of tokens sent to downstream LLMs. Pro delivers the highest benchmark scores across finance, healthcare and multilingual tasks, while Fast provides about three times higher throughput for high‑volume workloads. Both models can be accessed via Cohere’s API, Model Vault, Microsoft Foundry, Amazon SageMaker, or deployed privately.

## The facts

- Rerank 5 Pro costs $0.05 per 1M tokens; Rerank 5 Fast costs $0.02 per 1M tokens.
- Pro ranked first on 39 of 46 public benchmark datasets, with a score of 82.2 on the BRIGHT benchmark.

## Why it matters

Enterprises can lower AI inference costs and improve answer quality by sending only the most relevant text passages to their LLMs.

## Sources & references

1. [Introducing Rerank 5: precise, efficient enterprise document retrievalGive your AI the most relevant context for each request, now with passage-level precision.Oct 21, 20268 min read](https://cohere.com/blog/rerank-5) – Cohere, 2026-10-21

Last updated: 2026-10-21