DeepInfra Cookbooks

A collection of cookbooks, tutorials, and examples for using AI models on DeepInfra. This repository provides practical guides, performance benchmarks, and production-ready code examples to help developers build with AI models efficiently.

What's Included

Models

• Nemotron 3 Nano - NVIDIA's open reasoning model with hybrid MoE-Mamba architecture

Agents

• Deep Agent on Nemotron 3 Ultra - Build a LangChain Deep Agent on NVIDIA Nemotron 3 Ultra via DeepInfra's OpenAI-compatible API, using the Deep Agents harness profile tuned for Nemotron 3 Ultra

Structure

Each model cookbook includes:

• Tutorial Notebook - Step-by-step Jupyter notebook guide
• Examples - Ready-to-run use case examples
• Benchmarks - Performance and cost analysis tools
• Documentation - Model-specific README with setup instructions

Quick Start

1. Navigate to a model directory (e.g., models/nemotron-3-nano/)
2. Install dependencies: pip install -r requirements.txt
3. Set your DeepInfra API key: export DEEPINFRA_API_KEY="your-api-key"
4. Follow the model-specific README for detailed instructions

Resources

• DeepInfra Documentation
• DeepInfra Dashboard