Distributed engine for plugin-based graph and vector search

Vector Graph indeX is a high-performance, distributed engine for building custom search and recommendation services using Python plugins. Developers can implement service logic using the PyVGX C-extensions, expose it as HTTP endpoints, and automatically scale across a sharded, replicated back-end.

Install PyVGX

pip install pyvgx

This package includes native C extensions and has the following platform requirements:

  • Python: 3.9 or higher

  • macOS: 14 (Sonoma) or higher

  • Linux: glibc 2.34 or higher (e.g. Ubuntu 22.04+)

  • Windows: 10 or higher

Ensure your environment meets these requirements before installing.

Build Instructions

You can build the package yourself if needed: Instructions here.

Getting Started

PyVGX Overview

High level overview of VGX/PyVGX

PyVGX Tutorial

Introduction to graphs and graph queries

PyVGX Reference

Full Reference

A complete description of the VGX Python API.

Compact Reference

A compact summary of the VGX Python API.

VGX Server

Create scalable, multi-node services built on pyvgx plugins

…​ VGX Server …​ Plugins …​ Multi-Node …​ Admin

Quick Links

…​ Graph Objects …​ Vertex Objects …​ Memory Objects …​ Vector Objects

…​ Queries …​ Vertex Condition …​ Constants …​ Limits

…​ VGX Expression Language …​ Select Statement

Other Useful Information

VGX Transaction Protocol

VGX Transaction Interconnect protocol definition

VGX Core Concepts

VGX Core Concepts

PYVGX