Cognite Python SDK Documentation

This is the Cognite Python SDK for developers and data scientists working with Cognite Data Fusion (CDF). The package is tightly integrated with pandas, and helps you work easily and efficiently with data in Cognite Data Fusion (CDF).

Installation

To install or upgrade this package:

pip:

pip install --upgrade cognite-sdk

poetry:

poetry add cognite-sdk

uv:

uv add cognite-sdk

To install with optional dependencies:

pip install "cognite-sdk[pandas, geo]"

Available extras:

  • numpy: numpy

  • pandas: pandas

  • geo: geopandas, shapely

  • sympy: sympy

  • all: all of the above

What’s new in v8

The SDK v8 introduces full async support with the new AsyncCogniteClient. This enables:

  • Native async/await patterns for all API operations

  • Non-blocking concurrent operations directly in Notebooks (including browser-based via Pyodide) and UI frameworks like Streamlit

  • Significantly faster file uploads on Windows (new underlying HTTP client, httpx)

  • Async variants for all helper/utility methods on data classes (e.g., Function now has call_async)

Other improvements:

  • Simpler client instantiation: pass either cluster or base_url

  • Typed instance apply classes (e.g., CogniteAssetApply) now work correctly with patch updates (replace=False)

  • Better pandas DataFrame columns for datapoints: now uses MultiIndex for identifiers, aggregates, units, etc.

  • Read data classes now have correct types (no longer Optional on required fields)

  • More specific exceptions are now always raised when appropriate (CogniteNotFoundError, CogniteDuplicatedError)

For a complete list of changes, see the Migration Guide.

The synchronous CogniteClient remains fully supported and now wraps the async client internally.

# Async client (new in v8!)
from cognite.client import AsyncCogniteClient

async def main():
    client = AsyncCogniteClient()
    tss = await client.time_series.list()

# Sync client (still supported)
from cognite.client import CogniteClient

client = CogniteClient()
tss = client.time_series.list()

See the Quickstart for configuration examples.

Contents

Getting Started

Core Resources

Data Processing

Specialized Features

Reference