OpenAI ORG

Robust Speech Recognition via Large-Scale Weak Supervision

Whisper is OpenAI's open-source speech recognition model that converts audio files to text, supports 99+ languages, and can translate spoken words directly into English, installed with a single pip command.

Lightweight coding agent that runs in your terminal

Codex CLI is an AI coding agent from OpenAI that runs in your terminal, letting you ask questions about your code and have the assistant take actions on local files without leaving the command line.

Examples and guides for using the OpenAI API

A collection of practical, tested code examples and step-by-step guides for using the OpenAI API, covering common tasks in Python via Jupyter Notebooks.

A toolkit for developing and comparing reinforcement learning algorithms.

A standard toolkit of simulated game and physics environments for teaching AI agents through trial and reward. Now superseded by Gymnasium, use this repo only for legacy projects.

CLIP (Contrastive Language-Image Pretraining), Predict the most relevant text snippet given an image

CLIP is an AI model from OpenAI that matches images to text descriptions without labeled training data, you describe what you want and it finds or classifies matching images with zero examples.

The official Python library for the OpenAI API

openai-python is OpenAI's official Python library that lets you call GPT and other OpenAI models from Python in a few lines of code, handling authentication, streaming, vision inputs, and both sync and async clients automatically.

A lightweight, powerful framework for multi-agent workflows

OpenAI's official Python toolkit for building AI products where multiple AI helpers work together, hand off tasks, use tools like web search or code execution, and remember conversations.

Symphony turns project work into isolated, autonomous implementation runs, allowing teams to manage work instead of supervising coding agents.

Symphony is an AI engineering workflow tool that lets your team delegate coding tasks to autonomous agents, you manage the work queue, agents implement and verify the code with proof-of-work outputs, and Symphony handles safe merging.

Code for the paper "Language Models are Unsupervised Multitask Learners"

OpenAI's original code and model weights for GPT-2, the 2019 language model that could generate fluent text, answer questions, and summarize without task-specific training, released as an archived research artifact.

Skills Catalog for Codex

A catalog of reusable skills (plugins) that teach OpenAI Codex agents how to handle specific tasks like code review or project planning without re-explaining each time.

Educational framework exploring ergonomic, lightweight multi-agent orchestration. Managed by OpenAI Solution team.

Educational Python framework for building systems where multiple AI agents coordinate and hand off tasks to each other based on what users need.

Use Codex from Claude Code to review code or delegate tasks.

A Claude Code plugin that lets you call OpenAI's Codex AI agent for code reviews, bug fixes, and investigative tasks without leaving your editor.

The ChatGPT Retrieval Plugin lets you easily find personal or work documents by asking questions in natural language.

A self-hosted FastAPI service that gives ChatGPT or any AI assistant access to your own documents by storing and searching them with vector embeddings, you control where the data lives.

gpt-oss-120b and gpt-oss-20b are two open-weight language models by OpenAI

Two open-weight AI models from OpenAI, a 120B and a 20B parameter model, downloadable and runnable on your own GPU without API access, licensed Apache 2.0.

Evals is a framework for evaluating LLMs and LLM systems, and an open-source registry of benchmarks.

A framework and benchmark library for testing how well AI language models perform, run existing tests or write your own to measure accuracy on tasks specific to your app.

tiktoken is a fast BPE tokeniser for use with OpenAI's models.

Tiktoken is a fast Python tokenizer for OpenAI language models that converts text to token numbers and back, 3-6x faster than comparable tools, useful for counting tokens before sending API requests.

OpenAI Baselines: high-quality implementations of reinforcement learning algorithms

Reference Python implementations of classic reinforcement learning algorithms (DQN, PPO2, A2C, DDPG) from OpenAI, used as a baseline for RL research.

GPT-3: Language Models are Few-Shot Learners

Companion repository for the 2020 GPT-3 paper. Holds sample outputs, synthetic test datasets, training data statistics, and a model card. No code or model weights.

Generate 3D objects conditioned on text or images

Shap-E is an OpenAI research model that generates 3D objects from text prompts or images, outputting shapes as implicit functions that can be previewed as animated 3D models.

An educational resource to help anyone learn deep reinforcement learning.

An OpenAI learning package that teaches deep reinforcement learning through written guides, readable Python implementations of key algorithms, and practice exercises, aimed at people new to the field.

Official JavaScript / TypeScript library for the OpenAI API

The official JavaScript and TypeScript library for connecting Node.js or browser apps to OpenAI's AI models including GPT. Install with one command, then send prompts and receive responses, including streamed, piece-by-piece output.

PyTorch package for the discrete VAE used for DALL·E.

Official code release of one component from OpenAI's DALL-E image model: the discrete VAE that compresses and reconstructs images. The full text-to-image model is not included, intended for researchers studying the architecture.

Code for the paper "Jukebox: A Generative Model for Music"

Universe: a software platform for measuring and training an AI's general intelligence across the world's supply of games, websites and other applications.

No description.

OpenAI's 2021 research code for generating high-quality images using diffusion models with classifier guidance, includes pre-trained models at multiple resolutions and scripts to run or train your own.

This is a simple demonstration of more advanced, agentic patterns built on top of the Realtime API.

Point cloud diffusion for 3D model synthesis

Official repo for consistency models.

Demo of a customer service use case implemented with the OpenAI Agents SDK

macOS and Linux VMs on Apple Silicon to use in CI and other automations

Train the smallest LM you can that fits in 16MB. Best model wins!

No description.

Renderer for the harmony response format to be used with gpt-oss

No description.

Get a ChatGPT plugin up and running in under 5 minutes!

No description.

Release for Improved Denoising Diffusion Probabilistic Models

GLIDE: a diffusion-based text-conditional image synthesis model

React app for inspecting, building and debugging with the Realtime API

Retro Games in Gym

The official Go library for the OpenAI API

OpenAI Plugins

Code for the paper "Evaluating Large Language Models Trained on Code"

A lightweight, powerful framework for multi-agent workflows and voice agents

Code for reproducing results in "Glow: Generative Flow with Invertible 1x1 Convolutions"

MuJoCo is a physics engine for detailed, efficient rigid body simulations with contacts. mujoco-py allows using MuJoCo from Python 3.

Code for openai.fm, a demo for the OpenAI Speech API

Code for a multi-agent particle environment used in the paper "Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments"

Node.js example app from the OpenAI API quickstart tutorial

The official .NET library for the OpenAI API

No description.

OpenAI Privacy Filter

OpenAPI specification for the OpenAI API

Code for the paper "Improved Techniques for Training GANs"

Code and model for the paper "Improving Language Understanding by Generative Pre-Training"

Example apps for the Apps SDK

Consistency Distilled Diff VAE

800,000 step-level correctness labels on LLM solutions to MATH problems

No description.

Dataset of GPT-2 outputs for research in detection, biases, and more

Code for the MADDPG algorithm from the paper "Multi-Agent Actor-Critic for Mixed Cooperative-Competitive Environments"

OpenAI Assistants API quickstart with Next.js.

Code for the paper "PixelCNN++: A PixelCNN Implementation with Discretized Logistic Mixture Likelihood and Other Modifications"

No description.

GPT-5 coding examples

Example Discord bot written in Python that uses the completions API to have conversations with the `text-davinci-003` model, and the moderations API to filter the messages.

Python example app from the OpenAI API quickstart tutorial

Environment generation code for the paper "Emergent Tool Use From Multi-Agent Autocurricula"

Learn how to use CUA (our Computer Using Agent) via the API on multiple computer environments.

Video PreTraining (VPT): Learning to Act by Watching Unlabeled Online Videos

Code for the paper "Neural MMO: A Massively Multiagent Game Environment for Training and Evaluating Intelligent Agents"

Code for the paper "Evolution Strategies as a Scalable Alternative to Reinforcement Learning"

Examples of using sparse attention, as in "Generating Long Sequences with Sparse Transformers"

MLE-bench is a benchmark for measuring how well AI agents perform at machine learning engineering

Instructions on how to use the Realtime API on Microcontrollers and Embedded Platforms

The official Java library for the OpenAI API

No description.

This repo contains the dataset and code for the paper "SWE-Lancer: Can Frontier LLMs Earn $1 Million from Real-World Freelance Software Engineering?"

Code for the paper Fine-Tuning Language Models from Human Preferences

No description.

OpenAI Frontier Evals

Procgen Benchmark: Procedurally-Generated Game-Like Gym-Environments

No description.

Code for reproducing key results in the paper "InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets"

Efficient GPU kernels for block-sparse matrix multiplication and convolution

Code for "Learning to summarize from human feedback"

No description.

Code for the paper "On First-Order Meta-Learning Algorithms"

Node.js + JavaScript reference client for the Realtime API (beta)

Base docker image used in Codex environments

Code for the paper "Exploration by Random Network Distillation"

No description.

No description.

Starter app to build with OpenAI ChatKit + Agent Builder

Starter app to build with the OpenAI Responses API

Code for the paper "Emergent Complexity via Multi-agent Competition"

Code for the paper "Large-Scale Study of Curiosity-Driven Learning"

Build hours code to share.

The OpenAI Model Spec

Demo of a UI testing agent using the OpenAI CUA model and the Responses API.

Code for the paper "DeepType: Multilingual Entity Linking by Neural Type System Evolution"

Starter app to build with OpenAI ChatKit SDK

Official CLI for the OpenAI API

Code for the paper "Meta-Learning Shared Hierarchies"

Tools for accelerating safe exploration research.

No description.

No description.

Open-source release accompanying Gao et al. 2025

Demo showing how to use the OpenAI Realtime API to navigate a 3D scene via tool calling

Basic constrained RL agents used in experiments for the "Benchmarking Safe Exploration in Deep Reinforcement Learning" paper.

Repository for the paper "Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images"

Orchestrator for running Tart Virtual Machines on a cluster of Apple Silicon devices

Official Ruby SDK for the OpenAI API

Formal to Formal Mathematics Benchmark

Robotics Gym Environments

An official OpenAI toolkit for social scientists and data scientists to measure quantitative attributes in text, images, or audio using the GPT API.

Visualize harmony chat data and codex sessions in your browser

No description.

No description.

A digital camera you can build yourself with Codex.

Code for Implicit Generation and Generalization with Energy Based Models

Learn how to crawl your website and build a Q/A bot with the OpenAI API

Role-specific Codex plugin templates

code for reproducing some of the diagrams in the paper "Multimodal Neurons in Artificial Neural Networks"

API to access OpenAI Gym from other languages via HTTP

Code for the paper "Phasic Policy Gradient"

Create, publish and virtualize ephemeral Linux VMs with ease

Sample application to add voice capabilities to the Agents SDK

Code for the paper "Evolved Policy Gradients"

Example of using Realtime API as a meeting assistant to manage a Kanban Board

Code for the paper "OpenAI Remote Rendering Backend"

No description.

OpenAI Guardrails - Python

Code for the blog post "Learning Montezuma’s Revenge from a Single Demonstration"

Code and example data for the paper: Rule Based Rewards for Language Model Safety

For educational materials related to the spinning up workshops.

No description.

Demo of a customer support agent interface using NextJS and the OpenAI Responses API with File Search

No description.

Consistency models trained on CIFAR-10, in JAX.

Example code for using OpenAI’s NodeJS SDK with discord.js SDK to create a Discord Bot that uses Slash Commands.

Vectorized interface for reinforcement learning environments

Deno build of the official Typescript library for the OpenAI API.

Javascript BPE Encoder Decoder for GPT-2 / GPT-3

Publicly releasable baselines for the Retro contest

public release of Excel / OpenAI API integration

No description.

No description.

Unity Machine Learning Agents Toolkit

Open-sourced evaluation suite from the Monitoring Monitorability paper

Lightweight Android capture developer tools for macOS

No description.

OS repo for Knowledge Retrieval starter kit

Code for the paper "Improving GANs Using Optimal Transport"

Production-Grade Container Scheduling and Management

Sample app to get started using the Video API with Sora

Text-to-image samples collected for the evaluation of DALL-E 3 in the whitepaper.

Example MCP server compatible with Deep Research

OpenAI Retro Contest

OpenAI 2025 ICPC Submissions

Training Sonic with RLlib

No description.

Submissions for AI and Efficiency SOTA's

No description.

Software networking with isolation for Tart

Interactive Textbook Demo

A reverse proxy that provides authentication with Google, Github or other provider

No description.

No description.

A helpful gpt-based triage tool for BugCrowd bugbounty programs.

Classifiers for "Investigating Affective Use and Emotional Well-being in ChatGPT"

Next.js Photobooth app built with the OpenAI Image API

Code for the paper "Understanding RL Vision"

Tensors and Dynamic neural networks in Python with strong GPU acceleration

No description.

The Prometheus monitoring system and time series database.

Test Please ignore

No description.

No description.

No description.

Service for quickly aliasing and redirecting to long URLs

No description.

Guest agent for Tart VMs

Azure Command-Line Interface

No description.

A public-domain dataset of prompts and scenarios for evaluating compliance with the OpenAI Model Spec.

Experiments for paper

A GitHub Action for checking the status of a workflow on the main branch as a status on the PR

Development repository for the etcd cookbook

Installs the LogDNA host agent. Currently only supports apt systems.

Homebrew for the OpenAI CLI.

Run separate pipelines for each folder in your monorepo

Enrich your fluentd events with Kubernetes metadata

Evaluation harness and tooling for measuring model compliance with the OpenAI Model Spec.

Python Pushshift.io API Wrapper (for comment/submission search)

No description.

No description.

Chef cookbook to set node's hostname and FQDN.

Home to TypeScript packages for ChatGPT Sites

No description.