LambdaDB overview

LambdaDB is a serverless search database with built-in data versioning. Use vector, full-text, and hybrid search to find relevant data, and manage the data versions your applications read.

​

Store and search your data

Organize documents, embeddings, and metadata in collections, and choose index types for the fields you want to search or filter. Bring your own vectors or use native embeddings to generate document and query vectors from text and images. Search with full-text, dense vector, or sparse vector queries. Combine retrieval paths with hybrid search, narrow matches with filters, and optionally use native reranking to reorder retrieved results. See the search overview for query options, including keyword facets.

​

Test and version your data

Use branches, tags, and aliases to manage changes to a collection:

• Branch to isolate changes and test them without changing what applications read from another branch.
• Tag a committed snapshot to pin a data version for evaluation or later use.
• Use an alias to select a branch or tag for reads and switch the target when ready.

You can also create a point-in-time branch from retained committed history to inspect or recover an earlier state. Snapshot retention defaults to 30 days and is configurable; a tag preserves its pinned snapshot while the tag exists.

​

Run without managing servers

Documents, vectors, and indexes live in durable object storage, while compute scales independently for indexing and search. LambdaDB meters storage, reads, writes, and managed inference, with no compute charge between requests. See usage and pricing for plan allowances and charges.

​

Get started

Connect your coding agent to LambdaDB and run your first search, or build directly with an SDK. For direct HTTP requests, see the REST API reference.