← All terms
Infrastructure

Vector database

A database optimized for storing embeddings and finding nearest neighbors.

Vector databases index embeddings using approximate nearest-neighbor algorithms (HNSW, IVF) so similarity search stays fast at scale. Popular 2026 choices: pgvector (Postgres extension, default for most), Qdrant, Pinecone, Weaviate, Turbopuffer. Selection criteria: throughput, latency, hybrid-search support, multi-tenancy model, and cost-per-vector at your expected scale.

Related terms

Building with Vector database?

We ship production AI systems built around concepts like this every quarter. Send a brief and get a written proposal in 48 hours.

Send a brief →