Code beginner · 3 min read

How to create a Chroma collection in python

Q: How to create a Chroma collection in python

Use the chromadb Python client to create a Chroma collection by initializing a client and calling client.create_collection(name="your_collection_name").

Direct answer

Use the chromadb Python client to create a Chroma collection by initializing a client and calling client.create_collection(name="your_collection_name").

Setup

Install

bash

pip install chromadb

Imports

python

import chromadb
from chromadb.config import Settings

Examples

inCreate a collection named 'my_docs'

outCollection 'my_docs' created successfully.

inCreate a collection named 'research_papers'

outCollection 'research_papers' created successfully.

inCreate a collection with an existing name 'my_docs'

outCollection 'my_docs' already exists or retrieved existing collection.

Integration steps

Install the chromadb Python package.
Import chromadb and configure the client.
Initialize the Chroma client with optional settings.
Call create_collection with your desired collection name.
Use the returned collection object to add or query documents.

Full code

python

import chromadb
from chromadb.config import Settings

# Initialize Chroma client with default settings
client = chromadb.Client(Settings())

# Create a collection named 'my_collection'
collection = client.create_collection(name="my_collection")

print(f"Collection '{collection.name}' created successfully.")

output

Collection 'my_collection' created successfully.

API trace

Request

json

{"method": "POST", "endpoint": "/collections", "body": {"name": "my_collection"}}

Response

json

{"id": "collection_id", "name": "my_collection", "status": "created"}

Extractcollection = client.create_collection(name="my_collection")

Variants

Retrieve existing collection if it exists ›

Use when you want to avoid errors if the collection already exists and want to reuse it.

python

import chromadb
from chromadb.config import Settings

client = chromadb.Client(Settings())

# This will create or return existing collection
collection = client.get_or_create_collection(name="my_collection")

print(f"Using collection: {collection.name}")

Initialize client with persistent local storage ›

Use when you want your Chroma database to persist data locally across sessions.

python

import chromadb
from chromadb.config import Settings

settings = Settings(chroma_db_impl="duckdb+parquet", persist_directory="./chroma_db")
client = chromadb.Client(settings)

collection = client.create_collection(name="persistent_collection")
print(f"Persistent collection '{collection.name}' created.")

Performance

Latency~100-300ms for collection creation depending on storage backend

CostFree for local and open-source usage; cloud-hosted options may vary

Rate limitsNo enforced rate limits on local usage; cloud services may impose limits

Keep collection names short and descriptive to reduce metadata overhead.
Batch inserts into collections to improve throughput.
Use persistent storage settings to avoid recreating collections repeatedly.

Approach	Latency	Cost/call	Best for
Default in-memory client	~100ms	Free	Quick prototyping
Persistent local storage	~200-300ms	Free	Production with data persistence
Cloud-hosted Chroma	Varies	Depends on provider	Scalable multi-user deployments

✓

Quick tip

Always initialize the Chroma client with appropriate settings to control persistence and performance.

⚠

Common mistake

Trying to create a collection with a name that already exists without handling the exception or retrieving the existing collection.

Verified 2026-04

Verify ↗