MongoDB and Voyage AI

State-of-the art embedding models and rerankers made for building, scaling, and deploying intelligent applications.

An illustration of developer writing code and inputting data into computer.

Power more accurate and trustworthy AI applications at scale

Retrieving highly accurate and relevant data is critical for building and deploying mission-critical AI applications. Voyage AI’s embedding and reranking models help deliver highly accurate and relevant information retrieval to power sophisticated AI use cases.

Supercharging search and retrieval for unstructured data

Text embedding models

Voyage AI’s models help you retrieve the most relevant data for any purpose and language out-of-the-box.

Retrieval reranking

Voyage AI’s rerankers help reorder retrieved results from your documents and end-user queries so applications can deliver the most relevant responses.

Multimodal data processing

Voyage AI processes text and images together to derive fuller meaning from context-rich information sources.

How does MongoDB + Voyage AI Work

Chunk your data into smaller parts—including text documents, PDFs, videos, and more—to ensure that it fits within an LLM’s context window.

Voyage AI creates embeddings that are stored in MongoDB Atlas, enabling efficient search and retrieval of accurate and relevant information.

Atlas Vector Search enables retrieval-augmented generation (RAG) use cases or semantic search to find context-aware answers.

Domain-specific models

Finance

Optimized for finance retrieval, voyage-finance-2 enhances retrieval quality for data spanning financial news, public filings, finance advice, and financial reports.

Legal

Equipped with 16K-context length and trained on massive long-context legal documents, voyage-law-2 excels in long-context retrieval across domains.

Code retrieval

Tailored for semantic retrieval of code and related text data from both natural language and code queries, voyage-code-2 is specifically optimized for code-related applications.

Use Cases

Semantic search

A technique designed to understand the meaning behind a text query rather than matching it against just text itself. Semantic search allows developers to find relevant records much more accurately.

Retrieval-augmented generation (RAG)

An approach that combines information retrieval and text generation to efficiently use relevant and proprietary knowledge alongside LLMs to generate accurate responses. This allows outputs that are contextually rich and factually grounded.

Start building AI applications with MongoDB

With an operational database built for AI use cases and dedicated expert support to quickly go from prototype to production, the possibilities are limitless.

Get Started

GET STARTED WITH:

LLM and AI framework integrations
Comprehensive reference architectures
115+ regions across clouds
Expert support and tailored solutions
On-demand AI training