pgvector

What is pgvector?

pgvector is an open-source extension for PostgreSQL that lets you store embeddings next to the rest of your data and search them by similarity with ordinary SQL. After running CREATE EXTENSION vector, you add a vector column to a table, insert embeddings and order rows by distance to find nearest neighbors, either exactly or with an approximate index. It supports single-precision, half-precision, binary and sparse vectors, and distance measures including L2, inner product, cosine, L1, Hamming and Jaccard. Approximate search uses HNSW or IVFFlat indexes, and filtering, iterative index scans, partitioning, binary quantization and subvector indexing are documented for larger collections. Because it lives inside Postgres, you keep ACID compliance, point-in-time recovery and JOINs, and any language with a Postgres client can use it. It can be combined with Postgres full-text search for hybrid retrieval. The extension is free to install on Postgres 13 or newer by compiling it or using Docker and common package managers, and it is preinstalled by many hosted Postgres providers.

Use cases

• Store embeddings in Postgres
• Run nearest neighbor search in SQL
• Build semantic search for an app
• Power retrieval for RAG pipelines
• Combine vector and full-text search
• Filter vector queries with SQL conditions

Who it's for

• Backend developers
• ML engineers
• Postgres users
• Data engineers

Why use pgvector

• Keeps vectors and relational data in one database, so you avoid syncing a separate vector store.
• Uses plain SQL, so existing queries, JOINs and Postgres tooling keep working.
• Costs nothing to install, with Docker, Homebrew, APT, Yum and other package routes documented.
• Offers a choice of index type to trade build time and memory against query speed and recall.
• Supports compact vector types and quantization to shrink storage for larger collections.

pgvector features

• Exact and approximate nearest neighbor search
• Single-precision, half-precision, binary and sparse vector types
• L2, inner product, cosine, L1, Hamming and Jaccard distances
• HNSW and IVFFlat approximate indexes
• Iterative index scans for filtered queries
• Binary quantization and subvector indexing
• Hybrid search alongside Postgres full-text search
• ACID compliance, point-in-time recovery and JOINs
• Works with any language that has a Postgres client

pgvector integrations

• PostgreSQL
• Docker
• Homebrew
• Postgres.app
• conda-forge
• PGXN
• GitHub Actions

How to get started with pgvector

1. Install the extension by compiling it, using Docker or a package manager, or pick a hosted Postgres that includes it.
2. Run CREATE EXTENSION vector once per database that should use it.
3. Add a vector column and insert embeddings from your model.
4. Query with ORDER BY on a distance operator and LIMIT to get nearest neighbors.
5. Add an HNSW or IVFFlat index and tune it as the data set grows.

pgvector limitations

• Requires PostgreSQL 13 or newer
• Approximate indexes can return different results than exact search, and filtered queries can return fewer rows unless iterative scans are enabled
• HNSW builds are slower and use more memory than IVFFlat
• You manage the database, tuning and scaling yourself unless a hosted provider does it

pgvector pricing

pgvector is open source, so you can use and self-host it for free.

Prices can change and may vary by country or exclude tax. Check the official pricing page.

pgvector alternatives

The main pgvector alternatives listed on Debuly are Qdrant, Weaviate, Pinecone, Chroma, LanceDB and Zilliz Cloud. The table compares pricing and platforms; order is A–Z.

1. 

   Chroma

   Open-source search database for AI apps, with vector, full-text, regex and metadata search and a serverless cloud option.

   Developer platformsResearch and searchAutomation and agents
   Open sourceVisit Chroma (official site)
2. 

   LanceDB

   Open-source vector database and multimodal lakehouse for AI search, retrieval, feature engineering and training data.

   Developer platformsResearch and search
   Open sourceVisit LanceDB (official site)
3. 

   Pinecone

   Managed vector database for AI search and RAG, with dense, sparse and full-text indexes, inference and an Assistant API.

   Developer platformsResearch and search
   FreemiumVisit Pinecone (official site)
4. 

   Qdrant

   An open-source vector database and search engine for AI workloads, available self-hosted or as Qdrant Cloud with a free tier.

   Developer platformsResearch and searchAutomation and agents
   FreemiumVisit Qdrant (official site)
5. 

   Weaviate

   Open-source vector database with hybrid search and a managed Weaviate Cloud for semantic search, RAG and recommendations.

   Developer platformsResearch and search
   FreemiumVisit Weaviate (official site)
6. 

   Zilliz Cloud

   Fully managed vector database from the team behind Milvus, with free, serverless and dedicated clusters plus on-demand lake-scale compute.

   Developer platformsResearch and search
   FreemiumVisit Zilliz Cloud (official site)

pgvector FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from avatars.githubusercontent.com · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.