Introduction

Welcome to Chapter 2: AI Inference on Hugging Face. In this part of the course, you’ll learn two production-focused ways to run models:

• Inference Providers: zero-setup, multi-provider APIs to run models immediately.
• Inference Endpoints: your own fully managed, scalable endpoint for a specific model.

By the end of this chapter, you’ll be able to pick the right option, call models from code, and ship a small app end to end.

What you’ll learn

• Understand the difference between Inference Providers and Inference Endpoints
• Call text, image, and audio models via simple APIs
• Stream responses and handle errors
• Deploy a dedicated endpoint and call it from your app

Prerequisites

• A Hugging Face account and Access Token
• Python 3.9+ (for code examples)
• pip install huggingface_hub

export HF_TOKEN=hf_xxx

$env:HF_TOKEN = "hf_xxx"

Providers vs Endpoints: when to use which?

• Use Inference Providers when you want the fastest path to results across many models, with automatic routing and a simple API.
• Use Inference Endpoints when you need a dedicated, configurable, production-ready API for one model, with strict performance, security, or compliance requirements.

We’ll start with Inference Providers, then move to Inference Endpoints.

1. We’ll learn about Inference Providers to get instant results with minimal setup. You’ll build a small app using Providers.
2. Then we’ll learn Inference Endpoints for dedicated, production-ready deployments. You’ll apply what you’ve learned to build a small app using an Endpoint.
3. Finally, you’ll review and take a short quiz.