Story
October 8, 2026
ChatGPT Is Trading Plain Answers for Interactive Tools
OpenAI’s GPT-6 rollout adds Intelligent UI, turning ChatGPT replies into interactive charts, diagrams and task-specific controls. The company says users can reduce the visuals, but its direction is clear: answers are becoming interfaces.
OpenAI’s latest ChatGPT overhaul begins with GPT-6 and a blunt redesign of what an answer can be. The company’s new Intelligent UI, rolling out globally on Wednesday, turns responses into interactive visual tools rather than walls of text: tappable buttons, editable graphs, tailored calculators and charts built for the question at hand.1
The shift is not merely cosmetic. In an early example, a request to explain how an airplane wing creates lift produced diagrams intended to unpack the physics step by step. OpenAI also demonstrated recipe visuals, bicycle-mechanics diagrams, hiking maps and personal-savings calculators — all designed to make complex information more usable in the moment.1
Aarush Selvan, an OpenAI product manager, framed the update as a response to the limits of the chatbot’s old format. “ChatGPT has predominantly been a text-based interface,” he said, but for planning a trip, preparing a recipe or learning something new, “the most helpful answers aren’t just text.”1
The companion framing is even more direct: Intelligent UI fills ChatGPT’s replies with “pictures, charts, and buttons,” recasting the product from a conversational assistant into something closer to an on-demand app builder.2 Sam Altman distilled that message further: “ChatGPT can now generate a custom UI for you!”
3
OpenAI is nevertheless leaving an escape hatch. Users can dial back the visuals, much as they can adjust other aspects of ChatGPT’s personality. The feature arrives first for Pro, Plus, Business and Enterprise customers, before reaching free and lower-cost Go users on Thursday.1
That sequencing captures the company’s wager: not everyone will want a richer interface every time, but OpenAI believes the future of AI answers is increasingly something people can manipulate, not just read.