Voyage AI / voyageai/rerank-2.5

Voyage Rerank 2.5 - access through LLMTR

Voyage Rerank 2.5 evaluates a query and a list of candidate documents together, producing a relevance score for each document. Embedding search vectorizes query and document separately for speed; a reranker reads both at once and therefore produces a more accurate ordering. The typical pattern is to retrieve the top 50-100 candidates with embeddings, rerank them, and pass only the most relevant few to the language model, which improves answer quality and lowers the generation model's token cost. A single request evaluates at most 1000 documents. Billing is on total processed tokens: (query tokens x document count) + the sum of tokens across all documents.