# datastax/jvector

JVector: the most advanced embedded vector search engine

Repository: https://github.com/datastax/jvector
Canonical: https://ross.abutalabs.com/products/jvector
Language: Java
License: Apache-2.0
License Family: permissive
Topics: ann, java, knn, machine-learning, search-engine, similarity-search, vector-search
Last push: 2026-08-26T16:03:33+00:00

## Health v2 (maintenance only)
Score: 89/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 82, longevity 78
- inputs: {"age_days": 1105, "days_push": 7, "days_rel": 43, "gap_med": 41, "n_releases_24m": 4}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1741, forks 158 (observed 2026-08-28T04:05:30.278643+00:00)

## What it is
JVector is an embedded approximate nearest neighbor (ANN) search engine for the JVM, implementing a graph index that merges HNSW's hierarchical structure with DiskANN's Vamana algorithm. It supports incremental, concurrent index construction and two-pass searches combining in-memory compressed vectors (PQ, BQ) with on-disk full-resolution vectors.

## Use cases
- add vector similarity search to a Java application
- build an embedded ANN index without running a separate vector database
- search embeddings at scale with memory-efficient quantization
- incrementally update a vector index as new data arrives
- perform disk-backed nearest neighbor search over large datasets
- power semantic search or RAG retrieval in a JVM service

## When to choose
- you need high-performance vector search embedded in a Java/JVM application
- you want incremental index updates rather than static batch-built indexes
- your dataset is too large to keep all full-resolution vectors in memory
- you need concurrent, multi-threaded index construction

## When to avoid
- you need a standalone vector database with a query API rather than an embedded library
- your stack is not on the JVM
- you require exact (non-approximate) nearest neighbor results
- you need multi-modal indexes or built-in persistence/replication features of full databases

## Facets
- artifact type: library
- maturity: active
- function: search-engine, vector-database, machine-learning
- domain: databases, machine-learning
- platform: jvm, cross-platform
- tags: ann, knn, similarity-search, hnsw, diskann, vamana, product-quantization, embedded, vector-search, search, algorithms

## Member repositories
- datastax/jvector (main) score 89

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:30.278643+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 2026-08-30T03:30:02.220259+00:00, confidence not recorded.
  - readme: https://github.com/datastax/jvector (fetched 2026-08-28T04:05:30.278643+00:00, sha 9ab5ee06bec5)
- Data as of 2026-08-30T08:39:29.467469+00:00.