Ask ChatGPT to split a dinner bill, and you may now get a working bill splitter instead of a paragraph. On October 7, OpenAI began rolling out GPT-6 in ChatGPT together with a feature called Intelligent UI, which lets the chatbot build an interactive interface around each answer.
What is Intelligent UI?
Instead of replying only with text, ChatGPT can now include graphics, tappable buttons, forms, charts and other interactive elements in its answers. It can also build small tools on the spot, such as calculators, bill splitters and games. A comparison might appear side by side, and a recipe can adjust its quantities by guest count.
Text remains when text is the best answer. OpenAI product manager Aarush Selvan said text-only replies are often not the most helpful, and users can turn the visuals down if they prefer. Technically, OpenAI built a library of native, streamable components with a compiler that processes the interface as the model writes it, so elements appear progressively instead of all at once.
Who gets GPT-6 and when?
The rollout started on October 7 for Plus, Pro, Business and Enterprise users, who get GPT-6 Sol. Free and Go users follow on October 8 with GPT-6 Luna. For Enterprise accounts, availability depends on workplace admin settings. The models in ChatGPT’s Work tab and in Codex stay unchanged. OpenAI says more than 1.2 billion people use ChatGPT each week.
Is it faster?
GPT-6 can start answering while it is still thinking and add information as it goes. OpenAI says GPT-6 Instant begins responding 44% sooner on questions that need a web search than GPT-5.6 Instant. That is the company’s own measurement, and an independent usability test of Intelligent UI hasn’t appeared yet.
What OpenAI says about safety
The company says GPT-6 resists attempts to bypass its safety training better than GPT-5.6 Sol, especially attacks that adapt over several turns. It also says it updated protections against high-risk misuse involving cyberattacks, biological threats and violence, and that the model is better at admitting what it can’t do. The full results are in the system card, which one independent guide says also discloses regressions in some safety tests.
The bigger shift: the chat box becomes an app
Interactive outputs are not new, and other assistants, including Claude with its artifacts, already generate them. The difference is that in those cases the user usually has to ask. In GPT-6, the model decides on its own whether a question needs an interface, and it makes one by default for over a billion weekly users.
Two consequences follow. First, accuracy: a calculator built on the fly is only as correct as the model’s logic, so a number in a generated financial tool deserves the same check as any answer. Second, competition for attention. When ChatGPT answers a question with a working tool inside the conversation, there is less reason to click through to a website or open a separate app. For publishers and small tool makers, that is the part to watch.
What to watch next
Watch how Anthropic and Google respond, whether independent testers find Intelligent UI useful or distracting, and whether people keep it on once the novelty fades.