Chroma

Chroma is an open-source, Apache 2.0 licensed embedding database designed to serve as search infrastructure for AI applications. It provides fast, serverless, and scalable infrastructure supporting multiple search paradigms: vector search for semantic similarity, sparse vector search (BM25, SPLADE), full-text search with trigram and regex capabilities, and metadata filtering. Chroma is built on object storage and offers automatic data tiering, aiming to reduce operational overhead and costs. The platform is available as a cloud service with a free starter tier, or can be run locally via its open-source codebase. It is trusted by millions of developers and boasts over 15 million monthly downloads and 27,000 GitHub stars.

Chroma supports multiple programming languages including TypeScript, Python, and Rust, and provides a command-line interface (CLI) for development. Its features include dataset versioning through forking for A/B testing and rollouts, low-latency queries over billions of multi-tenant indexes, and SOC 2 Type II compliance. The platform is designed to scale with data and traffic, offering dedicated clusters for specific workloads. Chroma is positioned as a comprehensive search infrastructure for AI products, enabling developers to build applications that require efficient retrieval from large knowledge bases.

The tool is particularly suited for AI-powered search applications, knowledge base querying, and any scenario requiring fast, multi-modal search capabilities. Chroma's architecture emphasizes zero-ops infrastructure, with write throughput of 30 MB/s per collection and support for up to 1 million collections per database. It is used by a wide range of developers and organizations for building AI products that need robust, scalable search functionality.