Last updated: April 5, 2026 · Model Architecture · by Daniel Ashford
What is Vector Database?
A specialized database for storing and searching embeddings — the backbone of RAG systems.
Definition
A vector database is optimized for storing, indexing, and searching high-dimensional vector embeddings. It is where embeddings are stored and similarity searches are performed in RAG systems.
How It Works
Popular options include Pinecone (managed cloud), Weaviate (open source), Qdrant, Chroma (lightweight), and pgvector (PostgreSQL extension). They use approximate nearest neighbor algorithms to search millions of vectors in milliseconds.
Example
A legal RAG system: 50,000 contract clauses embedded and stored in Pinecone. When a lawyer asks about indemnification terms, the top 5 most similar clauses are retrieved for the LLM prompt.
Related Terms
See How Models Compare
Understanding vector database is important when choosing the right AI model. See how 12 models compare on our leaderboard.