Data Lakehouse Format
6.3k 2026-04-13

lance-format/lance

An open lakehouse format designed for multimodal AI, offering high-performance vector search, lightning-fast random access, and robust data versioning capabilities.

Core Features

Expressive hybrid search (vector, full-text, SQL)
Lightning-fast random access (100x faster than Parquet/Iceberg)
Native support for multimodal data (images, video, audio, text, embeddings)
Efficient data evolution and schema changes
Zero-copy versioning with ACID transactions and time travel

Quick Start

pip install pylance

Detailed Introduction

Lance is an innovative open lakehouse format specifically engineered for multimodal AI workloads. It provides a unified file format, table format, and catalog specification, enabling the construction of complete lakehouses on object storage. This project addresses critical needs in AI workflows by facilitating high-performance I/O, random access, and advanced search capabilities for diverse data types, including embeddings. Its design prioritizes efficiency, scalability, and seamless integration with popular data science tools, making it ideal for building search engines, feature stores, and large-scale ML training pipelines.

OSS Alternative

Explore the best open source alternatives to commercial software.

© 2026 OSS Alternative. hotgithub.com - All rights reserved.