Chroma is an open-source embedding database, maintained by the chroma-core project and licensed under Apache 2.0, that describes itself as “the open-source data infrastructure for AI.” It aims to be the simplest way for a developer to add vector search to an application: with a minimal API, a user can add documents and have them automatically embedded, then query for the most similar results, with metadata and full-text filtering supported.
Chroma is built to scale with a project’s needs, offering an in-memory mode for prototyping, persistent local storage, and Chroma Cloud, a hosted serverless service. The core has been rewritten substantially in Rust for performance, with client libraries for Python and JavaScript, and it supports vector, hybrid, and full-text search. By GitHub’s count the repository has tens of thousands of stars, reflecting its popularity as a starting point for retrieval-augmented generation projects.
Why business readers should care: Chroma sits at the easy-onramp end of the vector-database market, frequently the first store a team reaches for when prototyping a “chat with your documents” feature. Its open-source core lowers the cost of experimentation, while the hosted cloud option follows the now-familiar pattern of monetizing a free framework through a managed service.