MongoDB Atlas Vector Search

MongoDB Atlas Vector Search: semantic search built into your database

MongoDB Atlas Vector Search adds vector search directly to MongoDB Atlas, so AI applications can store embeddings alongside the operational data they already use. Instead of running a separate vector database, developers add a $vectorSearch aggregation stage to query by semantic similarity — powering retrieval-augmented generation (RAG), recommendations, and AI agents against the same data, indexes, and security model as the rest of their MongoDB app.

Production credibility: MongoDB (NASDAQ: MDB) is one of the most widely deployed databases in the world, and Atlas Vector Search is its native AI-retrieval layer. In 2025 MongoDB acquired embedding-model company Voyage AI to bring high-accuracy embeddings into the platform, and added automated embedding so teams can enable semantic search in a single step.

Key Features

• $vectorSearch stage with Approximate (ANN) and Exact (ENN) nearest-neighbor search
• Automated embedding — generate and manage vector embeddings for text in one click
• Embeddings from any provider under 4096 dimensions, including Voyage AI
• Vectors stored beside operational data — no separate vector database to sync
• Pre-filtering with the full MongoDB Query API for hybrid search
• Integrations with LangChain, LlamaIndex, and major RAG frameworks

Ideal Use Case

Teams already building on MongoDB who want to add semantic search, RAG, or AI-agent retrieval without standing up and syncing a separate vector store. It is the path of least resistance for putting AI on top of data that already lives in MongoDB.

How MongoDB Atlas Vector Search differentiates

The pitch is consolidation: keep your vectors, metadata, and operational data in one place, queried through one API, under one security and scaling model. Dedicated vector databases like Pinecone or Weaviate can offer more specialized tuning, but they add a second system to operate and keep in sync. Atlas Vector Search trades some specialization for the simplicity of one database doing both jobs.

FAQ

Is MongoDB Atlas Vector Search free? There is a free path: MongoDB Atlas offers a free-tier (M0) cluster, and vector search runs within Atlas. Production workloads use paid dedicated clusters (starting around $0.08/hour) plus usage, so costs scale with data and traffic.

Do I need a separate vector database? No — that is the point. Vector embeddings are stored and indexed inside MongoDB Atlas alongside your existing data, removing the need to run and sync a standalone vector database.

Can it power RAG and AI agents? Yes. Atlas Vector Search is designed for retrieval-augmented generation and agent retrieval, with integrations for LangChain, LlamaIndex, and other frameworks, plus automated embeddings via Voyage AI.

tl;dr

MongoDB Atlas Vector Search brings semantic vector search into MongoDB Atlas, letting developers store embeddings beside their data and power RAG and AI agents without a separate vector database — with a free tier to start.