Chroma is open-source search infrastructure for AI, released under Apache 2.0. It offers fast, serverless and scalable vector, sparse vector, full-text, regex and metadata search, built on object storage with automatic data tiering between memory, SSD and cold storage. Clients exist for TypeScript, Python and Rust, along with a command-line tool and forking for dataset versioning. Chroma Cloud pricing is usage-based: the Starter plan costs $0 per month with $5 in free credits, and the Team plan is $250 per month plus usage, with custom Enterprise pricing.
Developers building AI applications and agents are the core audience, particularly those who want search without tuning servers. Startups can begin on the free Starter plan or run Chroma locally, growing teams can move to Team with more databases and Slack support, and enterprises can choose BYOC in their own VPC, multi-region replication and point-in-time recovery. The project reports more than 15 million monthly downloads and 27 thousand GitHub stars, and the cloud holds SOC 2 Type II certification.
Typical use cases include knowledge bases for retrieval-augmented generation, agent memory and search, AI app backends, and multi-tenant products that need many separate indexes. Developers can combine lexical search such as BM25 with semantic vector search, filter by metadata, and use forking to test dataset changes or roll them out safely. Published latency figures for 100k vectors at 384 dimensions list a warm p50 of 20ms, and per-collection limits include 5 million records. Teams can also self-host when they prefer to keep everything local. Because storage and queries are billed separately, teams can estimate monthly cost with the pricing calculator by entering vector size, document size and query volume before they commit.
