Model details
Cohere Rerank 4 Pro
Cohere Rerank 4 Pro serves as a second-stage relevance reranker rather than a generative model, designed to reorder candidate documents after an initial retrieval step. Vercel AI Gateway describes it as built for state-of-the-art relevance on complex queries, with explicit support for both English and non-English content as well as semi-structured JSON payloads, which makes it useful for multilingual search systems, enterprise knowledge bases, and retrieval-augmented pipelines that need to surface the most pertinent passages from heterogeneous sources.
In practical terms, developers access the model through Vercel AI Gateway using the AI SDK's rerank helper, supplying a query string and an array of candidate documents and receiving a relevance-ranked result back. The gateway lists Cohere as the underlying provider for this model, and billing is structured around the number of queries processed rather than token volume, which can simplify cost forecasting for high-traffic retrieval workloads where rerank calls scale with search volume.
Quick Info
Powered by- Provider
- Vercel AI Gateway
- Model key
- cohere/rerank-v4-pro
- Release date
- Dec 11, 2025
- Last updated
- Dec 11, 2025
- Input modalities
- Output modalities
- Capabilities
Limits
- Output tokens
- 32,000 tokens
- Context window
- 32,000 tokens
Latest news about Cohere Rerank 4 Pro
No articles yet. Fetch the latest news to show it here.