Voyage Rerank 3 and Rerank 3 Lite are now available on AI Gateway
Voyage's reranking models now live on Vercel's AI Gateway—faster relevance ranking for RAG pipelines without switching APIs.
Why it matters
Voyage AI's Rerank 3 and Rerank 3 Lite are now available through MongoDB's AI Gateway, offering practitioners a unified API for document reranking with trade-offs between accuracy and latency. This reduces integration friction for teams building retrieval-augmented generation (RAG) systems.
The key facts
13 to knowRerank 3 and Rerank 3 Lite now available on AI Gateway
Both models handle up to 32K tokens per query-document pair
Rerank 3 optimized for accuracy; Rerank 3 Lite for latency and cost
Largest improvements reported on long documents and code
Single API for model calls, usage tracking, cost, routing, retries, and failover
Models reorder search results before passing to language model
Available via AI Gateway API key or bring-your-own-provider setup
Voyage Rerank 3 and Rerank 3 Lite now available on AI Gateway
Rerank 3 focuses on accuracy; Rerank 3 Lite optimized for latency and cost
Model identifiers: voyage/rerank-3 and voyage/rerank-3-lite
AI Gateway provides unified API for model calling, usage tracking, routing, retries, failover
Integration available via AI Gateway API key or bring-your-own-provider key
Largest retrieval gains reported on long documents and code domains
Go to the source
Vercel Blogvercel.com
Publisher excerpt: and from Voyage AI by MongoDB are now available on AI Gateway.Rerank 3Rerank 3 Lite The models reorder search results by relevance before passing documents to a language model. The series improves retrieval quality across domains, with the largest gains on long documents and code. Rerank 3 focuses…