LanceDB

Founded by Lei Xu

An open-source, serverless vector database for AI that allows you to store, manage, and query embeddings, metadata, and full-text search directly on disk or object storage.

Cost

Free Tier

Rating

★ People love it

Time to value

Quick Setup (< 1 hour)

You can use LanceDB to store raw data, embeddings, and features in a single table that supports vector search, full-text search, and SQL queries. It handles text, images, audio, and video data together while providing fast retrieval for AI applications. The database works as an embedded library or cloud service, storing data in Lance format for efficient compression and random access. You can build RAG systems, recommendation engines, and search applications with built-in versioning and S3-compatible storage.

What LanceDB does

Import datasets from HuggingFace Hub directlyQuery data using SQL and vector search togetherStore embeddings alongside original images and textSet up distributed vector indexing for large datasetsCreate full-text search indexes on document collectionsVersion control machine learning datasets automaticallyDeploy embedded database in Python applicationsStream data processing pipelines with Ray integrationStores raw data and embeddings in the same tableSupports text, images, audio and video in one databaseBuilt-in vector and full-text search capabilitiesWorks as embedded library or cloud serviceNative versioning and data evolution supportS3-compatible object storage integrationLance format with 50% better compression than ParquetScales to 10 billion vectors with distributed indexing