LanceDB

Store, search, and prepare multimodal data for AI applications

Visit website

About LanceDB

LanceDB is a multimodal data platform for teams building AI applications and training models. Built on the open-source Lance format, it keeps raw data, metadata, embeddings, and derived features together in versioned tables. Developers can combine vector, full-text, and hybrid search to retrieve relevant information, then filter, sample, or deduplicate records for downstream workflows. The embedded edition runs inside an application without a separate database server, while the enterprise service adds distributed infrastructure for larger workloads. Teams can use the same data layer for retrieval, feature engineering, and loading training examples into PyTorch, reducing the need to move datasets between separate systems.

Discussion

Tried it? Share a useful thought or ask a question.