Chroma DB
An open-source vector database for AI apps. Run it locally or self-host under Apache 2.0, or use managed search through Chroma Cloud.
Chroma DB is an open-source search database for developers building AI apps and agents that need to retrieve information from their own data. You can run it locally or host it on your own infrastructure under the Apache 2.0 license. Chroma Cloud is a separate hosted service for managed, serverless search.
Its search capabilities cover several ways of finding relevant records. Vector search matches semantic similarity, while sparse vector search supports lexical retrieval with BM25 and SPLADE. Full-text search includes trigram matching and regular expressions. Metadata filters and faceted search let applications narrow results by stored attributes, and grouping can aggregate results by metadata keys.
Collections can hold documents, embeddings and metadata, including arrays of strings, numbers and booleans. Collection forking creates copies through copy-on-write for dataset versioning, A/B tests and rollouts. Developers can access Chroma through Python and JavaScript or TypeScript clients; the project uses Rust as its main implementation language.
For teams choosing hosted infrastructure, Chroma Cloud supports vector, hybrid and full-text search. Its object-storage architecture uses automatic data tiering and caching, and the service scales with data and traffic. Cloud Sync handles ingestion into Chroma Cloud, with capabilities to crawl and index web pages or chunk, embed and index GitHub repositories.
