What is a vector database? (and which free ones to use)

What it is

A vector database stores high-dimensional embeddings (numeric representations of text, images, or audio) and retrieves the nearest ones to a query vector — i.e. semantic similarity search. It's the storage-and-retrieval engine behind RAG, recommendations, and semantic search.

Why not a normal database?

Traditional databases match exact values or keywords. A vector database uses approximate nearest-neighbor (ANN) search to find items that are semantically close, even when the words don't match. That's what lets "How do I reset my password?" retrieve a doc titled "Account recovery steps."

What to look for

• Recall vs. speed — ANN indexes trade a little accuracy for big latency wins.
• Metadata filtering — combine semantic search with structured filters.
• Scaling model — embedded/local vs. server vs. managed.

Free options on ai-supply

Several open-source vector stores and the embedding models that feed them are published here, each security-scanned with a score and grade. Pair an embedding model with a vector store from the data category, then validate retrieval quality with an eval.

Building RAG or semantic search? Start with a vetted vector store — browse the catalog.