rerank-2.5 is a high-accuracy reranking model that improves search and retrieval results by reordering candidate documents based on relevance. It is well-suited for production search, RAG, recommendation systems, and enterprise retrieval workflows.
The model supports instruction-following reranking, allowing teams to guide relevance scoring with natural language instructions. It also supports a 32K token context length, making it effective for longer documents and complex retrieval tasks.
On-demand DeploymentDocs | On-demand deployments allow you to use voyage-rerank-2.5 on dedicated GPUs with Fireworks' high-performance serving stack with high reliability and no rate limits. |
