Chroma

What it does

Chroma is an open-source search database for AI. It takes text documents, embeddings, and metadata as input, indexes them, and returns ranked results from vector, full-text, regex, and metadata queries. The core Python and JavaScript clients let developers add documents to named collections and query them with a few function calls.

Developers use Chroma to give AI agents and applications retrieval over their own data, such as documentation or repositories. The open-source project is distributed under Apache 2.0 and includes a server mode started with a local command. Chroma Cloud, the hosted service, is being wound down, so new work should target the open-source database or a self-hosted deployment.

Chroma suits developers building retrieval features who want a locally runnable open-source database. It is an infrastructure library, not an end-user chat app. Hosted Cloud is no longer open to new customers: writes end November 22, 2026 and reads end December 22, 2026, so migration is required before then.

What it can do

• Vector similarity search over stored embeddings
• Full-text and BM25/SPLADE sparse vector search
• Regex and trigram text search
• Metadata filtering and faceted search
• Collection forking for dataset versioning
• Built-in tokenization and embedding of added documents

Example workflows

Local document search prototype

Input: text documents with string IDs and metadata. Steps: create a local client, create a collection, add documents, run a query with n_results. Output: the nearest matching documents with IDs and distances.

Metadata-filtered retrieval

Input: documents tagged with source metadata. Steps: add documents with metadatas, query with a where filter on a metadata field. Output: ranked results restricted to the matching source.

Before you choose

• Chroma Cloud writes are disabled on November 22, 2026 and reads on December 22, 2026; migrate data and remove Cloud credentials before then.
• The open-source license is free, but running self-hosted infrastructure carries your own hosting and embedding-model costs.

Listed under

• Coding
• Sandboxes and infrastructure
• AI developers
• Developer tools
• Infrastructure
• Open source
• RAG
• Search

Get started

Run pip install chromadb or npm install chromadb, then create a client and collection.

1. Install the client with pip install chromadb or npm install chromadb.
2. Create a client and a named collection in your code.
3. Add documents with unique string IDs and optional metadata, then query the collection with query_texts and n_results.

ChromaDB is available under the Apache 2.0 license. You supply and pay for your hosting infrastructure. The existing Chroma Cloud service is winding down.

Chroma is free to use. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Chroma is joining Replit | Chroma ↗www.trychroma.com
• Chroma Cloud wind-down timeline | Chroma ↗www.trychroma.com
• GitHub - chroma-core/chroma: Search infrastructure for AI · GitHub ↗github.com
• Getting Started - Chroma Docs ↗docs.trychroma.com
• Chroma - open-source search infrastructure for AI ↗www.trychroma.com

Company website

trychroma.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.