OpenAI said on Oct. 7 that GPT-6 is starting a worldwide rollout in ChatGPT’s conversation tab. The big addition is something called Intelligent UI, which lets replies blend text, charts, buttons, forms, and other interactive components.
Access for Plus, Pro, Business, and Enterprise users begins that same day. Then, from the following day, OpenAI expands availability to the free tier and the Go plan. Paid plans use GPT-6 Sol. Free and Go run on GPT-6 Luna. OpenAI said both versions are tuned for everyday conversations.
OpenAI released the first batch of GPT-6 models to paying customers last month. Now that new model generation is coming to ChatGPT, which the company said reaches more than 1.2 billion users every week. One limit, though: the change applies only to the conversation tab. The models powering Work and Codex are staying the same.
Intelligent UI picks charts, maps, or tools based on the question
OpenAI said it trained GPT-6 to mix text, visual elements, and interactive components according to the prompt. Comparison-style questions can appear side by side. Explanations can show up as interactive diagrams. And if plain text does the job, the model can just answer in text.
The company’s examples are pretty direct: putting a Sunday barbecue schedule next to a recipe, or pinning road-trip stops right onto a map. Users can also ask ChatGPT to make a small tool inside the conversation itself. Think a calculator for deposit growth. Or a bill-splitting tool for group meals. Even a game you can play in-chat.
Streaming component library lets the interface show up progressively
On the technical side, OpenAI said it built a streamable native component library, plus a compiler that processes the interface while the model is generating its reply. So the interface does not have to sit there waiting for the full answer to finish before appearing. It can arrive piece by piece as the response is generated.
OpenAI also said the model’s sense of design still has room to get better.
GPT-6 can answer even while it is still reasoning
There is another shift here. ChatGPT can now reason and answer at the same time. OpenAI said earlier reasoning models usually waited until the thinking process was done before returning anything. GPT-6 does it differently: it factors in user wait time and builds the response in segments using information it has already gathered.
By OpenAI’s internal evaluations, GPT-6 Extra High starts responding in about the same time as GPT-5.6 Medium, while earning a higher overall score than GPT-5.6 Extra High. For questions that need web search, GPT-6 Instant began answering 44% earlier on average than GPT-5.6 Instant.
Safety work carries over from Astra
On safety, OpenAI said GPT-6 includes several improvements from Astra. In adversarial testing, the model did a better job resisting attempts to get around safety training, especially attacks that shift tactics across multiple turns of conversation. More detailed results are in the system card.
OpenAI ended its announcement with this line: "The future will be different: instead of people adapting to software, software will adapt to people."