What it is

Chroma is an open-source vector database designed to store, index, and retrieve embeddings for AI applications. It runs as a local Python or JavaScript client or in client-server mode, providing a lightweight infrastructure layer for AI search and retrieval workloads. Built in Rust and licensed under Apache 2.0, it targets developers building RAG, agents, and other embedding-dependent systems.

It solves the problem of managing high-dimensional vector data without relying on proprietary or complex enterprise databases. Chroma eliminates boilerplate around tokenization, embedding, and indexing by handling those steps automatically during add() and query(), while still allowing custom embeddings when needed.

Key capabilities

• Automatic tokenization, embedding, and indexing during document ingestion via collection.add()
• Query by text, with optional metadata and document-content filters (where, where_document)
• In-memory and persistent storage modes via client configuration
• Full CRUD operations on collections: create_collection, get_or_create_collection, delete_collection
• Client-server deployment mode via chroma run --path /chroma_db_path
• Support for Python (pip install chromadb) and JavaScript (npm install chromadb) clients
• Hosted service (Chroma Cloud) offering serverless vector, hybrid, and full-text search

Who uses it and how

• AI/ML engineers prototyping RAG applications using Colab notebooks or local scripts with minimal setup
• Agent frameworks integrating Chroma as short-term memory or knowledge store via the 4-function API
• Developers deploying local vector search in desktop or edge tools using persistent client mode
• Teams evaluating Chroma Cloud for production workloads before committing to self-hosting

Getting started

Install the Python client with pip install chromadb and run chromadb.Client() for in-memory use, or chroma run --path /chroma_db_path for server mode. JavaScript users run npm install chromadb. Chroma Cloud offers a hosted alternative with $5 free credits.

When to use it — and when not to

Chroma replaces paid vector databases for prototyping and small-scale deployments, especially where simplicity and low friction matter more than enterprise features. Self-hoster must manage storage, persistence, and optional SMTP for alerts, but no external database or embedding service is required. It is not ideal for large-scale production workloads needing advanced security, multi-tenancy, or strict SLAs—those cases may prefer Chroma Cloud or enterprise-grade alternatives.

pip install chromadb # python client
# for javascript, npm install chromadb!
# for client-server mode, chroma run --path /chroma_db_path

import chromadb
# setup Chroma in-memory, for easy prototyping. Can add persistence easily!
client = chromadb.Client()

# Create collection. get_collection, get_or_create_collection, delete_collection also available!
collection = client.create_collection("all-my-documents")

# Add docs to the collection. Can also update and delete. Row-based API coming soon!
collection.add(
    documents=["This is document1", "This is document2"], # we handle tokenization, embedding, and indexing automatically. You can skip that and add your own embeddings as well
    metadatas=[{"source": "notion"}, {"source": "google-docs"}], # filter on these!
    ids=["doc1", "doc2"], # unique for each doc
)

# Query/search 2 most similar results. You can also .get by id
results = collection.query(
    query_texts=["This is a query document"],
    n_results=2,
    # where={"metadata_field": "is_equal_to_this"}, # optional filter
    # where_document={"$contains":"search_string"}  # optional filter
)