# alibaba/zvec

A lightweight, lightning-fast, in-process vector database

Repository: https://github.com/alibaba/zvec
Canonical: https://ross.abutalabs.com/products/zvec
Homepage: https://zvec.org
Language: C++
License: Apache-2.0
License Family: permissive
Topics: rag, agent-skills, embedded, faiss, hnsw, llm-memory, search-engine, semantic-search, similarity-search, vector-database, local, db, vector-db
Last push: 2026-08-26T08:55:58+00:00

## Health v2 (maintenance only)
Score: 83/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 99, longevity 19
- inputs: {"age_days": 271, "days_push": 7, "days_rel": 9, "gap_med": 23.0, "n_releases_24m": 11}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 15518, forks 980 (observed 2026-08-28T04:11:09.986638+00:00)

## What it is
Zvec is an open-source, in-process vector database by Alibaba that embeds directly into applications with no server or daemon required. It provides fast dense and sparse vector similarity search with hybrid filtering, durable WAL-backed storage, and Python/Node.js SDKs.

## Use cases
- run a local vector database without a server
- build semantic search over my documents
- store embeddings for a RAG pipeline
- add similarity search to an existing application
- give AI agents long-term memory with vector search
- search billions of vectors with low latency
- replace faiss with a persistent embedded vector store
- hybrid search combining vectors and metadata filters

## When to choose
- you want an embedded, zero-config vector database like SQLite but for vectors
- you need sub-millisecond similarity search at local or billion scale
- you're building RAG, semantic search, or agent memory in Python or Node.js
- you need durable storage with WAL and multi-process reads

## When to avoid
- you need a distributed, multi-writer vector database cluster
- your stack requires a language without an official SDK
- you need full-text search as the primary workload rather than vectors

## Facets
- artifact type: library
- maturity: active
- function: vector-database, search-engine, rag, database
- domain: databases, machine-learning, artificial-intelligence, developer-tools
- platform: python, cpp, windows, cross-platform, self-hosted
- tags: in-process, embedded-database, hnsw, diskann, similarity-search, semantic-search, llm-memory, faiss-alternative, local-first, sparse-vectors, hybrid-search, agent-skills, mcp-server, search, retrieval-augmented-generation, nodejs, linux, macos

## Member repositories
- alibaba/zvec (main) score 83

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:09.986638+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:05:41.236499+00:00, confidence not recorded.
  - readme: https://github.com/alibaba/zvec (fetched 2026-08-28T04:11:09.986638+00:00, sha d03d16385d23)
  - homepage: https://zvec.org (fetched 2026-08-29T08:03:49.824839+00:00, sha f95295253291)
  - site_page: https://zvec.org/en/docs/db (fetched 2026-08-29T08:03:49.833808+00:00, sha cdcc354a41c3)
  - site_page: https://zvec.org/en/docs/db/quickstart (fetched 2026-08-29T08:03:49.835823+00:00, sha e0cde62392db)
  - site_page: https://zvec.org/en/docs/db/benchmarks (fetched 2026-08-29T08:03:49.837666+00:00, sha 166fc6fdc61b)
- Data as of 2026-08-30T08:39:29.467469+00:00.