OpenAI announced on Wednesday an “Intelligent UI” upgrade for ChatGPT that augments text answers with custom visual components when they add value. The feature runs on the company’s latest GPT-6 model, which is being released to paying subscribers today and to free-tier accounts starting tomorrow. The rollout targets the platform’s more than a billion users, aiming to make interactions less reliant on long blocks of text.
During a pre-launch test, the system produced a detailed slug anatomy diagram that replaced a textual description. The image displayed the creature’s upper tentacles, muscular foot and other parts, each labeled with a button that could be pressed to reveal additional information. This interactive layout let the user explore the biology of the slug without scrolling through paragraphs.
The chatbot also generated a San Francisco apartment-affordability calculator that appeared as a compact widget with movable sliders for income, rent and other expenses. Adjusting the controls instantly updated a percentage figure showing how much of a user’s take-home pay would be required for various unit sizes. The visual tool turned a typical financial query into an on-screen experiment.
Perhaps the most striking demonstration was an airplane-seat explorer that let the user compare cabin layouts across four Alaska and Delta aircraft. The interface displayed seat maps where each seat could be tapped for details, and exit-row locations were highlighted in green. By interacting with the map, the user could quickly assess comfort and safety features without consulting a static diagram.
OpenAI’s product manager Aarush Selvan explained that a dedicated design team evaluates each response to decide whether a diagram, chart or button adds value or creates clutter. The system also accepts user instructions to reduce visual output, allowing those who prefer plain text to keep the interaction minimal. This balance aims to tailor the experience to individual preferences.
The visual-first approach mirrors Google’s earlier rollout of a “generative UI” for Search, which can render adjustable graphics such as a black-hole explainer with sliders. As chatbots become increasingly multimodal,blending text, images and voice,the line between AI-generated content and traditional UI design blurs, raising concerns that designers may lose control over software appearance. With AI capable of crafting bespoke interfaces on demand, future chatbots are expected to deliver highly personalized visual experiences.