lancedb/lancedb

An open-source, embedded retrieval library and multimodal AI lakehouse designed for fast, scalable vector search and data management in AI/ML applications.

Core Features

Fast Vector Search: Billions of vectors searchable in milliseconds with state-of-the-art indexing.

Comprehensive Search: Supports vector similarity, full-text search, and SQL queries.

Multimodal Data Support: Stores, queries, and filters vectors, metadata, and multimodal data (text, images, videos, point clouds).

Advanced Data Management: Features zero-copy operations, automatic versioning, and GPU support for index building.

Flexible Deployment: Available as 100% open-source for local/cloud deployment and as a managed cloud/enterprise solution.

GitHub Repo Full Preview Documentation

Detailed Introduction

LanceDB is a multimodal AI lakehouse built on the efficient Lance columnar format, providing a central platform for developing, training, and analyzing AI workloads. It offers production-ready vector search capabilities, allowing developers to store, index, and query petabytes of multimodal data and vectors with ease. Designed for scalability and performance, LanceDB simplifies data management for AI/ML applications, enabling rapid retrieval and analysis without complex infrastructure.