Greg Brockman

FrontierScience is a new benchmark for evaluating AI’s expert-level scientific reasoning across physics, chemistry, and biology. It measures both Olympiad-style problem solving and real research tasks, helping track how well advanced models can support and accelerate scientific work.

ChatGPT Images is powered by the new GPT Image 1.5 model. It delivers precise edits that preserve lighting and facial likeness, stronger instruction following, and 4x faster generation. Available now in ChatGPT and via API.

Group chats are now in ChatGPT. You can add friends, family, or coworkers to a shared conversation to collaborate, plan, and create together with AI. It's separate from your private chats.

GPT-5.1 Instant and Thinking are trained using the same stack as OpenAI's reasoning models, so they score higher on factuality and complex problem-solving than GPT-5, while also introducing a more natural, conversational tone. We think many people will find that GPT-5.1 does a better job of bringing IQ and EQ together, but one default clearly can’t meet everyone’s needs.

gpt-oss-safeguard is a new family of open-source safety models (120b & 20b) from OpenAI. They use reasoning to classify content based on a custom, developer-provided policy at inference time, providing an explainable chain-of-thought for each decision.

Apps are now in ChatGPT. Chat directly with interactive apps from Spotify, Canva, Zillow, and more, right in your conversation. Built on the new open Apps SDK, it’s a new generation of apps that turns ChatGPT into the next major app platform.

ChatGPT now lets you buy products directly in chat. Powered by the new open Agentic Commerce Protocol (ACP) built with Stripe, Instant Checkout connects you with merchants like Etsy (and soon Shopify) to turn product discovery into a seamless purchase.

Codex just got faster, more reliable, and better at real-time collaboration and tackling tasks independently anywhere you develop—whether via the terminal, IDE, web, or even your phone.

gpt-realtime is OpenAI's new speech-to-speech model for production voice agents, delivering low latency and natural, expressive speech. The Realtime API is now GA, adding key features for developers like remote MCP support, image input, and SIP phone calling.

AGENTS.md is a simple, open format for guiding coding agents, used by over 20k open-source projects. Think of it as a README for agents.

With major improvements in reasoning, code quality, and UX. It handles complex coding tasks with minimal prompting, provides clear explanations, and introduces enhanced agentic capabilities, making it a powerful collaborator and intelligent assistant for all.

ChatGPT's new "study mode" is a learning experience that guides you to answers with hints and questions, instead of just giving you the solution. It offers personalized, step-by-step support to help you build a deeper understanding of any topic.

ChatGPT now thinks and acts, proactively choosing from a toolbox of agentic skills to complete tasks for you using its own computer.

OpenAI Image API now features gpt-image-1! Get GPT-4o level quality for image generation & advanced editing (multi-ref images, inpainting). Better instruction following & text rendering.

OpenAI o3 and o4-mini are reasoning models that think with images & agentically use tools (Search, Code, DALL-E). SOTA multimodal performance. Available in ChatGPT & API.

A new series of GPT models featuring major improvements on coding, instruction following, and long context—plus our first-ever nano model.

Learn AI with OpenAI Academy, a free online resource hub. Offers workshops, videos, and content for all backgrounds, from basics to advanced topics.

Unlocking useful and valuable image generation with a natively multimodal model capable of precise, accurate, photorealistic outputs.

GPT-4.5 is our most advanced model yet, scaling up unsupervised learning for better pattern recognition, deeper knowledge, and fewer hallucinations. It feels more natural, understands intent better, and excels at writing, programming, and problem-solving.

SWE-Lancer is an open-source benchmark from OpenAI, featuring 1,400+ real-world software engineering tasks sourced from Upwork. Test your AI's coding and managerial skills.