Use Cohere Rerank 4.0 in OCI Generative AI

You can now host Cohere Rerank 4.0 on a dedicated AI cluster in OCI Generative AI. Rerank 4.0 improves enterprise retrieval and reranking workflows by providing a larger context window, improved handling for multilingual and semi-structured content, and variants optimized for either higher-quality reranking or lower-latency workloads.

Key Features

• 32,000-token context window: Supports much larger inputs than Rerank v3.5, improving handling for long enterprise documents and larger retrieval candidate sets.
• Improved reranking quality: Enhances relevance ranking for enterprise retrieval workloads, including business, finance, and technical content.
• Self-learning support: Lets you adapt reranking behavior to enterprise terminology and domain-specific data without annotated training datasets.
• Two model variants:
  • Rerank 4 Pro (cohere.rerank-v4.0-pro) for higher-precision and more complex reranking tasks.
  • Rerank 4 Fast (cohere.rerank-v4.0-fast) for lower-latency and higher-throughput workloads.
• Multilingual and semi-structured support: Improves reranking for multilingual content and semi-structured data, including JSON, tables, and code-like content.
• Mode: Available on-demand and through hosting on dedicated AI clusters.

  For available regions, and whether each variant is offered on-demand or through dedicated AI clusters, see Generative AI Models by Region.

For information about the service, see the Generative AI documentation.