Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/openai/openai-cookbook/llms.txt

Use this file to discover all available pages before exploring further.

The OpenAI Cookbook is an open-source, community-maintained library of practical examples and reference guides for the OpenAI API. Every entry is a runnable Jupyter notebook or a focused article written by developers who have solved a real problem — and want to share how. If you’re looking for copy-paste-ready code rather than conceptual docs, you’re in the right place.

What’s inside

The cookbook organizes its 289+ examples into topic areas that map to the most common API use cases. Each category contains both introductory guides and advanced patterns you can adapt to your own projects.

Models & Prompting

Understand how large language models work and write more effective prompts for GPT-5, o-series reasoning models, and open-weight alternatives.

Agents & Automation

Build multi-step agents using the OpenAI Agents SDK — with tools, handoffs, tracing, and guardrails for production systems.

Embeddings & Search

Turn text into dense vector representations and use them for semantic search, clustering, classification, and recommendation.

Fine-tuning & Evals

Customize models with your own data through supervised fine-tuning, DPO, or Reinforcement Fine-Tuning, and evaluate quality systematically.

Audio & Speech

Transcribe audio with Whisper, stream real-time conversations with the Realtime API, and synthesize natural-sounding speech with TTS.

Vision & Images

Analyze images, documents, and video frames with GPT-4o and GPT-5.4, and generate or edit images with DALL-E and GPT Image models.

How the cookbook works

Most examples are Jupyter notebooks stored under examples/<topic>/ in the GitHub repository. Longer narrative guides live in articles/. Every piece of content is registered in registry.yaml, which is what drives the cookbook website.
Most code is written in Python, but the underlying API concepts apply to any language. Official SDKs exist for Python and Node.js, and the REST API is callable from any HTTP client.
To run any notebook, you need an OpenAI API key and Python 3.9 or later. See the Quickstart to get set up in a few minutes.

Who contributes

The cookbook is community-driven. Examples come from OpenAI engineers, partner companies, and independent developers. You’ll find entries from enterprise integration patterns to experimental research techniques — all reviewed before publication. If you’ve built something useful with the OpenAI API, the cookbook welcomes your contribution. See the contributing guide for step-by-step instructions.

Explore by section

Responses API

Stateful, tool-augmented completions with built-in file search and web browsing.

Function Calling

Give models structured access to external APIs and business logic.

Open Models

Run and fine-tune open-weight models via the OpenAI-compatible API.

Build docs developers (and LLMs) love