Voyage AI / voyageai/rerank-3

Voyage Rerank 3 - access through LLMTR

Voyage Rerank 3 reads a query and a list of candidate documents together, produces a relevance score for each document and orders the list by that score. Voyage positions it as the most accurate member of its reranker family and the default choice for most applications. Reranking the top 50-100 candidates from embedding or keyword search with this model and passing only the most relevant few to the language model improves answer quality and lowers the generation model's token cost. The query can be at most 8,000 tokens; the query plus any single document cannot exceed 32,000 tokens, and a single request evaluates at most 1000 documents. Total processed tokens, (query tokens x document count) + the sum of tokens across all documents, can be at most 600,000 per request, and billing is on that total.