Self-host VectorChord
1.1.1PostgreSQL extension for billion-scale vector search
- Stars
- 1 832
- RAM
- 128GB for 1B-vector indexes
- Requires
- PostgreSQL
ghcr.io/tensorchord/vchord-postgresWhat is VectorChord?
VectorChord is a PostgreSQL extension built for scalable, high-performance vector search at billion-scale. It uses RaBitQ compression — offering native 4-bit and 8-bit vector types — to drastically cut storage costs while keeping recall loss under 1%. The extension works with the same data types and syntax as pgvector, so existing applications can migrate without code changes. Indexing 100 million vectors takes roughly 20 minutes thanks to hierarchical K-means and optimized disk I/O, and the system can scale to 1 billion vectors on machines with 128 GB of RAM. It ships as Docker images: a base Postgres image with the extension (ghcr.io/tensorchord/vchord-postgres) and an all-in-one suite image that also bundles VectorChord-bm25 and pg_tokenizer.rs. The project is dual-licensed under AGPLv3 and Elastic License v2.
Should you self-host VectorChord?
VectorChord scores 57/100 on selfhost.directory (how scoring works) and ranks #4 of 5 in Vector Database. It has 1 832 GitHub stars.
It shipped 3 stable releases in the past year. The latest, 1.1.1, came out on 28 February 2026. New versions land roughly every 2 months.
Setup is moderate: expect some configuration beyond a single command. An official Docker image is available. It is Other-licensed and written mostly in Rust.
Its closest alternative on selfhost.directory is Typesense (79/100, 27 061 stars). It scores higher overall, so compare the two before you pick. See VectorChord vs Typesense side by side.
Features
- Accelerated index build
- Billion-scale vector support
- External index precomputation
- Graph index
- Monitoring and Observability
- Multi-vector retrieval (MaxSim)
- Pgvector compatibility
- Prefilter and prefetch
- RaBitQ quantization (4-bit and 8-bit)
- Similarity filter / range queries
- Vector Similarity Search
VectorChord README
More in Vector Database
weaviateSelf-hosted open-source vector database that keeps objects and embeddings together, mixing semantic, BM25 and hybrid search with RAG, reranking, multi-tenancy and RBAC.
AntflyAntfly is a self-hosted, open-source Zig database unifying full-text, vector, and graph search with built-in ML for embeddings, reranking, and RAG.
ArcadeDBOpen-source, self-hosted multi-model database: one ACID engine for graph, document, key-value, vector, full-text and time-series data, queried in SQL, Cypher or Gremlin.
brain-dbSelf-hosted memory database for AI agents with fused semantic, lexical, graph, and temporal retrieval. Single Rust core, binary wire protocol, Apache-2.0.
chromaOpen-source vector database for AI retrieval: dense, sparse, full-text, regex and metadata search. Self-host it with pip, the chroma CLI, or Docker on port 8000.
databendSelf-hosted, open-source data warehouse in Rust for large-scale analytics plus vector and full-text search, with Python UDF sandboxes and git-like data branching.
GelOpen-source graph-relational database that builds on Postgres with EdgeQL, migrations, branching, built-in auth and AI; self-hosted via Docker or the gel-server binary.
helixdbA Rust-based graph database that unifies vector similarity and keyword search on object storage for AI memory workloads.
Manticore SearchSelf-hosted, open-source search database that speaks SQL and HTTP JSON, with full-text, vector, hybrid and conversational search pitched as an Elasticsearch alternative.
oceanbaseSelf-hosted, open-source distributed SQL database that is MySQL compatible and handles OLTP plus real-time HTAP analytics, vector and full-text search in one engine.
qdrantQdrant is an open-source, self-hosted vector database in Rust for fast AI similarity search with filtering and quantization.
RedisRedis is a self-hosted, open-source in-memory data structure server used as a database, cache, message broker, and vector query engine with sub-millisecond latency.
0 comments