Native embedding and reranking models now on Atlas!

The Embedding and Reranking API on MongoDB Atlas is now available in public preview, giving you direct access to Voyage AI’s state-of-the-art embedding and reranking models through a native, serverless API, including 200M free tokens on the latest models.

As more AI applications move into production, accurate retrieval has become essential for grounding LLMs in relevant, trustworthy data. Embedding and reranking models sit at the core of this pipeline, directly influencing relevance, accuracy, and hallucination reduction in search, recommendations, and RAG workflows, making access to high-quality models critical. Voyage AI offers a suite of embedding and reranking models that consistently deliver the highest accuracy across categories on the Retrieval Embedding Benchmark (RTEB).

Learn more

With the Embedding and Reranking API on MongoDB Atlas, you can:

Access Voyage AI models natively:

Generate API keys in Atlas and call Voyage AI embedding and reranking models, including the latest Voyage 4 model series, through a serverless API with simple token-based pricing—no additional infrastructure to manage.

Improve retrieval accuracy for AI workloads:

Increase semantic relevance and reranking precision to return higher-quality results with less context noise.

Operate a unified AI stack on Atlas:

Build and run your data, retrieval, and models under a single control plane with Atlas’ enterprise-grade security, access controls, and unified billing.

When combined with Atlas capabilities like Atlas Vector Search, the Embedding and Reranking API brings together the core building blocks required to ship reliable, production-ready AI applications and agents.

Dive into the documentation and get started today.

Try the API