LanceDB is a vector database for multimodal data, built on the Lance columnar format. It stores, indexes, and searches vectors together with their metadata and the underlying text, images, videos, and point clouds, and it combines vector similarity search, full-text search, and SQL over the same data. Versioning is automatic and zero-copy, so a dataset's history is kept without separate infrastructure, and index building can use a GPU.
It runs embedded and locally or in your own cloud, with a managed cloud and enterprise offering for production-scale search without servers to manage. Client libraries cover Python, TypeScript or JavaScript, and Rust, with a REST API as well, and it plugs into LangChain, LlamaIndex, Apache Arrow, pandas, Polars, and DuckDB.
Features
- Vector search: approximate nearest-neighbour search over billions of vectors, with GPU-assisted index building
- Combined retrieval: vector similarity, full-text search, and SQL filtering against the same tables
- Multimodal storage: vectors, metadata, and multimodal payloads such as text, images, videos, and point clouds stored together
- Columnar format: built on Lance for storage and analytics, with zero-copy automatic versioning of datasets
- Deployment options: embedded and local, self-hosted in your own cloud, or a managed cloud and enterprise service
- SDKs: Python, TypeScript or JavaScript, and Rust libraries plus a REST API
- Ecosystem integrations: LangChain, LlamaIndex, Apache Arrow, pandas, Polars, and DuckDB
Integrates with
CrewAI
Python multi-agent framework pairing autonomous role-based Crews with event-driven Flows for precise orchestration
LangChain
Framework for building LLM applications and agents from interoperable components and a large integrations library
LlamaIndex
Open-source data framework for building LLM and agentic apps over your own data, with 300+ integrations
