OpenAI's GPT-6 Gives ChatGPT an 'Intelligent UI' With Graphics, Buttons and Charts

Speaking at a media briefing, the company said the chatbot decides on its own when a response needs a more visual treatment and will integrate those elements directly into the answer. For simpler prompts ChatGPT will keep its existing text-only format, and users can request an interactive response at any time, after which the model will reformat the answer accordingly.

OpenAI demonstrated two examples. Asked how to assemble a 7-speed bicycle, ChatGPT generated a full design breakdown with visualizations of each major part and a clickable button for learning more about each one, rather than an extended text explanation. In a second case, a user asking for a lamb roast plan received images of each dish alongside a recipe that recalculated ingredient quantities when the number of guests was changed.

Product Manager Aarush Selvan said during the briefing that ChatGPT builds these visualizations from a library of components the GPT-6 models were pretrained on. The goal, he said, is to make learning complex topics easier. "ChatGPT has predominantly been a text-based interface," Selvan said, adding that the company wants to help people get things done and believes that "from the everyday things like planning a recipe or a trip or even trying to learn something, the most helpful answers aren't just text."

The Intelligent UI is not the first of its kind. It resembles Google LLC's "Generative Interface," introduced with the launch of Gemini 3 last year, though OpenAI's version appears more colorful and heavier on graphics. When Google introduced its feature there was considerable excitement around it, but it does not appear to have differentiated Gemini much as a chatbot.

OpenAI did not say whether the Intelligent UI will mean higher token consumption, though the extra generated elements suggest it may, something users might want to monitor. The company also did not address how susceptible the visual elements are to the accuracy problems that affect all AI chatbots, namely hallucinations.

The feature follows other moves by OpenAI to make its AI more visually appealing. Earlier this week the company said it would soon introduce the first image-based ads in ChatGPT's responses. Users may find it harder to tell which responses are ads and which are the chatbot's own output.

OpenAI said the Intelligent UI is available from today for ChatGPT Plus, Pro, Business and Enterprise subscribers, with workplace administrators needing to enable it for the Enterprise tier. Free users and Go subscribers will get access to it tomorrow.