← All lakebases
vector
LanceDB
A vector database whose storage format is open and readable by other tools.
Kind
vector
Vendor
LanceDB
Workload
Vector and multimodal search
Storage backend
Object storage (S3, GCS, Azure)
Separates storage and compute
Yes
Object storage is source of truth
Yes
Open source
Yes
Access paths
Open source, LanceDB Cloud
LanceDB is an open-source database for vectors and other AI data. It stores tables and vector indexes in Lance, an open columnar format built for random access and machine learning, and it reads and writes that data directly on object storage.
Why it qualifies
- Object storage is the source of truth. Tables and indexes live in object storage (S3, GCS, or Azure Blob), not on a single node's disk.
- Storage and compute are separate. LanceDB runs embedded or as a managed service, with stateless query processes reading from the shared store.
- Operational workload. It serves live vector and full-text search for retrieval and AI applications.
- Storage is in an open format. Lance is Apache-2.0 and readable by tools like pandas, Polars, DuckDB, and Spark, so the same data is available outside LanceDB with no export step.
Where it sits in the category
LanceDB is one of the clearest examples of the open-format criterion. Because the data is Lance on object storage, another service can read it directly, which is the property that removes batch exports and copy pipelines.