chroma-core/chroma

Chroma is an open-source database designed specifically for AI applications. Its main job is storing and searching through "embeddings", numerical representations of text (or other data) that capture meaning rather than exact words. This allows AI apps to find documents that are conceptually similar to a question, even if they use different wording. This type of database is called a vector database, and it is a core building block for AI features like chatbots that can reference your own documents. The API is deliberately simple. You create a collection, add documents to it, and then query it with a question, Chroma automatically handles the process of converting text into embeddings and finding the most similar results. The code examples show the full workflow: creating a client, adding documents with optional metadata for filtering, and querying for the closest matches. Chroma supports running in-memory for quick prototyping, with persistence added easily when you need to save data. It can also run as a client-server setup. A hosted cloud version called Chroma Cloud is available for serverless use, offering vector search, hybrid search (combining meaning-based and keyword-based), and full-text search. The core library is available as a Python package and a JavaScript/npm package. The underlying engine is written in Rust. Chroma is a good fit for developers building AI features, such as chatbots, document Q&A tools, or recommendation systems, that need to search through large collections of text by meaning rather than exact keywords.