Chroma
ActiveDescription
Chroma is an open-source AI-native embedding database designed for building LLM applications. It provides simple APIs to store embeddings and perform similarity search, making it ideal for RAG applications.
Key Features
- Minimal core API — only 4 functions: create_collection, add, query, get — up and running in 5 minutes
- Auto embedding — automatically handles tokenization, embedding, and indexing on add, no manual processing
- Metadata filtering — precise filtering by metadata fields and full-text document search
- Hybrid search — supports combined vector similarity search and full-text retrieval modes
- Multi-language clients — Python and JavaScript/TypeScript clients, pip/npm one-click install
- Persistent storage — in-memory mode for prototyping, persistent mode for production, chroma run for server
Use Cases
💡 RAG application backend — provide vector retrieval for LLMs, enable Q&A over private data
💡 Semantic search — similarity retrieval for unstructured data like documents and image descriptions
💡 Recommendation systems — personalized recommendations based on user behavior embeddings
💡 Knowledge graph supplement — store entity embeddings for semantic-level knowledge association queries
💡 Rapid prototyping — quickly validate retrieval effectiveness of AI apps using in-memory mode
Strengths & Limitations
✅ Strengths
- • Actively maintained, recent updates
- • High community interest (29.1k stars)
- • Permissive open-source license (Apache-2.0)
- • Established track record (3 years in production)
Categories
Quick Start
pip install chromadb
import chromadb
# Start in-memory for quick prototyping
client = chromadb.Client()
collection = client.create_collection('my-docs')
# Add documents (auto-embedding)
collection.add(
documents=['This is document 1', 'This is document 2'],
metadatas=[{'source': 'notion'}, {'source': 'google-docs'}],
ids=['doc1', 'doc2']
)
# Query top 2 most similar results
results = collection.query(
query_texts=['query document'],
n_results=2
)
print(results)