Vqlite: document-oriented vector retrieval

Understand the project

Practical overview

Use Vqlite when an application needs self-hosted vector retrieval and its documents naturally correspond to multiple vectors and shared metadata.

Developers adding vector retrieval to an application through a REST API.
Teams building knowledge-search or retrieval-augmented generation components.
Operators who want a self-hosted search service organized around documents with multiple embeddings.

Provide vector similarity retrieval to an application through a RESTful interface.
Associate multiple vectors with one document instead of managing every vector as a separate record.
Store document metadata alongside vectors without adding MySQL or Redis solely for that storage.

Before you choose

The README documents self-hosted Docker and source-build options. It lists Linux or macOS and Go as requirements and recommends Docker.

The document-first model supports multiple vectors per document and shared metadata.
Google ScaNN is the search core.
The project describes retrieval-speed trade-offs involving segment sizing and count, and reorder settings that affect both speed and recall.
The README directs users to a Python SDK directory for sample code and recommends reading Design.md before adapting the project.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up the documented Docker-based Vqlite service with a mounted configuration file and persistent data directory.

Copy vqlite.yaml to the machine that will run Vqlite and make the appropriate configuration changes.
Pull the image ghcr.io/vqlite/vqlite:latest.
Run the container using the documented configuration-file and data-directory mounts, publishing port 8880.

A Vqlite container is configured to serve on the published port and use the mounted data directory.

The README does not explain the configuration fields or show a request for adding or searching vectors.
The documented Docker command uses local paths and settings that may need adjustment for a particular host.

Evidence and transparency

Source record details

Source availability: Active
Sources checked: 18 October 2026
Original website title: VQLite/VQLite · Official homepage title · retained in its source language