---
id: ngcsdk
version: "4.34.10"
license: Apache-2.0
license_treatment: permissive
maintenance: active
---

# ngcsdk — NVIDIA GPU Cloud SDK

License: permissive · Maintenance: active · Downloads: 270.9K/mo

## What it is and what it does

NGC-SDK is NVIDIA's official Python client for the NVIDIA GPU Cloud platform. It provides authenticated access to NGC's model registry, allowing you to query, list, and retrieve metadata about models and other resources hosted in the cloud. The SDK handles API authentication via API keys, manages token caching, and returns results as both Python objects and JSON, making it straightforward to integrate NGC resources into Python applications and workflows.

The package is built on a foundation of common HTTP and cloud libraries (aiohttp, requests, boto3, docker) to support both synchronous and asynchronous operations, as well as container and storage interactions. It requires Python 3.10 or later and is actively maintained, with recent releases indicating ongoing support for current Python versions.

Use it for:

- Query and retrieve model metadata from the NGC registry in Python applications.
- Automate access to NVIDIA's pre-trained models and resources for ML workflows.
- Integrate NGC resources into CI/CD pipelines or batch processing jobs.
- Build internal tools that need authenticated programmatic access to NGC.
- Debug NGC API interactions using built-in logging and configuration inspection.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

SDK for interacting with NVIDIA GPU Cloud, providing Python utilities to access NGC registry, models, and cloud resources via authenticated API calls.

Yes. The package is actively maintained, has low install friction, carries a permissive license, and is purpose-built for NGC integration. Install it if you need programmatic access to NVIDIA GPU Cloud resources from Python. Verify that your use case aligns with the 19 runtime dependencies (particularly docker and boto3) before committing to production.

## Install

pip install ngcsdk
uv add ngcsdk
poetry add ngcsdk

## Installing ngcsdk

Before you install: Low friction installation with a pure-wheel distribution. Actively maintained with a recent release. Requires Python 3.10 or later and pulls in 19 runtime dependencies including aiohttp, boto3, docker, and requests for cloud and container operations.

License in practice: Licensed under Apache-2.0 (permissive), allowing commercial use, modification, and distribution with minimal restrictions—suitable for most integration scenarios.

Quickstart:

pip install ngcsdk

from ngcsdk import Client

clt = Client()
clt.configure(api_key='your-api-key', org_name='nvidia', team_name='no-team')
model = clt.registry.model.info('nvidia/nemotron')

Requires Python >= 3.10 and a valid NGC API key (obtainable from https://ngc.nvidia.com/setup).

Verify before relying:

- Whether the 19 runtime dependencies (including docker, boto3, aiohttp) are all required for basic use or if many are optional.
- Performance characteristics and rate limits when querying the NGC registry.
- Whether telemetry extras are commonly needed or only for advanced monitoring scenarios.

## Package facts

- License: Apache-2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 270.9K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

nvidia gpu cloud sdk, ngc registry client, nvidia model registry access, gpu cloud python api, nvidia cloud integration, nvidia-cloud, model-registry, gpu-cloud

[View on SkillFed](https://skillfed.io/packages/ngcsdk) · [View on PyPI](https://pypi.org/project/ngcsdk/)