NGT-labs/

NGT

GitHub

NGT is a C++ library and CLI for fast approximate nearest neighbor search on high-dimensional data, supporting graph and tree-based indexing with multiple data types and distance metrics.

1.4kstars

+17 days

131forks

30issues

Apache-2.0license

2016since

Star historydaily snapshots by VibeCrowd

Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (2 so far).

Alternatives & relatedmatched by topic overlap

infiniflow/

infinity

▲ 10/wk

The AI-native database built for LLM applications, providing incredibly fast hybrid search of dense vector, sparse vector, tensor (multi-vector), and full-text.

4.7k435C++

weaviate/

weaviate

▲ 35/wk

Weaviate is an open-source vector database that stores both objects and vectors, allowing for the combination of vector search with structured filtering with the fault tolerance and scalability of a cloud-native database.

17k1.4kGo

lancedb/

lancedb

▲ 58/wk

Developer-friendly OSS embedded retrieval library for multimodal AI. Search More; Manage Less.

11k985Rust

milvus-io/

milvus

▲ 105/wk

Milvus is a high-performance, cloud-native vector database built for scalable vector ANN search

46k4.2kGo

qdrant/

qdrant

▲ 144/wk

Qdrant - High-performance, massive-scale Vector Database and Vector Search Engine for the next generation of AI. Also available in the cloud https://cloud.qdrant.io/

34k2.6kRust

MariaDB/

server

▲ 24/wk

MariaDB server is a community developed fork of MySQL server. Started by core members of the original MySQL team, MariaDB actively works with outside developers to deliver the most featureful, stable, and sanely licensed open SQL server in the industry.

7.9k2.1kC++

CompareNGT vs infinity NGT vs weaviate NGT vs lancedb

Reviewgenerated from repository data · Jul 28, 2026

What it is

NGT provides commands and a library for performing high-speed approximate nearest neighbor searches against a large volume of data in high dimensional vector data space.

How it works

Method: Graph and tree-based indexing (NGT) with optional QG (Quantized graph-based) and QBG (Quantized blob graph-based) methods. Supported distance functions include L1, L2, Cosine similarity, Angular, Hamming, Jaccard, Poincare, Lorentz, and Inner product. Data types include 4-byte and 2-byte floating point numbers, and 1-byte unsigned integers. Build parameters cover shared memory usage, large data optimization, and disabling QG/QBG.

Getting started

Installation steps are provided for Linux, CentOS, Ubuntu, and macOS. Build commands include cmake and make sequences, with options such as -DNGT_QBG_DISABLED=ON and -DNGT_SHARED_MEMORY_ALLOCATOR=ON. Example commands from README:

$ unzip NGT-x.x.x.zip
$ cd NGT-x.x.x
$ mkdir build
$ cd build
$ cmake -DNGT_QBG_DISABLED=ON ..
$ make
$ make install
$ ldconfig /usr/local/lib

Build parameters include:

Shared memory use: -DNGT_SHARED_MEMORY_ALLOCATOR=ON ..
Large-scale data use: -DNGT_LARGE_DATASET=ON ..
Disable QG and QBG: -DNGT_QBG_DISABLED=ON ..

Supported languages include Python, Ruby, PHP, Rust, Go, C, and C++ (sample code).

Recent releases

Latest releases (as listed):

v2.7.4 Add PrimitiveComparatorArm.h (2026-04-20)
v2.7.3 Add support for ARM NEON (2026-04-20)
v2.7.2 Add an option for normalization errors (2026-03-18)
v2.7.1 Fix normalization issues (2026-03-16)
v2.7.0 Improve Forest and QG (2026-03-11)

Traction

Stars: 1368 (as of provided data).

Behind the repo

Public company/organization: Yahoo Japan (contact via contributor page). Publisher pages reference ONNG, PANNG, ANNG, and related publications in the README.

Caveats

License: Apache-2.0. Created in 2016. Last push: 2026-07-27. Issues: 30 open. Original license and CLA requirements noted in repository.

SharePost on X LinkedIn