NVIDIA AI Model Center

A Python skill that provides a unified interface to 42+ NVIDIA NIM API models — LLM chat, vision analysis, text embeddings, image generation, and more.

Quick Start

from model_center import ModelCenter

center = ModelCenter()

# List all models
models = center.list_models()
print("Available categories:", list(models.keys()))

# Get model info
info = center.get_model_info('nemotron-3-super-8b')
print(f"Model: {info['name']}, Provider: {info['provider']}")

# Compare pricing
comparisons = center.compare_pricing(['nemotron-3-super-8b', 'llama-3.1-70b-instruct'])
for c in comparisons:
    print(f"{c['name']}: ${c['input_price']}/M in, ${c['output_price']}/M out")

# Get recommendations
rec = center.recommend_model('code generation', 'low', False)
print(f"Recommended: {rec}")

# Estimate cost
cost = center.estimate_cost('nemotron-3-super-8b', 1000, 500)
print(f"Estimated cost: ${cost['total_cost']}")

# Chat with a model (requires NVIDIA_API_KEY env var)
# response = center.chat_completion(
#     model='nemotron-3-super-8b',
#     messages=[{'role': 'user', 'content': 'Hello'}],
#     temperature=0.7,
#     max_tokens=100
# )

Setup

1. Get an API key from build.nvidia.com
2. Set the NVIDIA_API_KEY environment variable:

   $env:NVIDIA_API_KEY = "nvapi-..."
   

3. Install dependency: pip install requests
4. Import and use ModelCenter or NVIDIAAPIClient from model_center.py

API

ModelCenter

Categories

• llm: Chat completion models (Nemotron, Llama, Mixtral, etc.)
• vision: Image understanding models
• embedding: Text embedding models
• image: Image generation models
• moderation: Content moderation models

Source

The implementation lives in model_center.py (548 lines) in this skill directory.