OpenAI Rolls GPT-6 and an Interactive UI to Every ChatGPT Tier
OpenAI has begun rolling GPT-6 to every ChatGPT tier alongside Intelligent UI, an interface that mixes text with interactive charts, forms, diagrams, and side-by-side layouts. Sol and Luna reach paid and limited free access, while GPT-6 Astra stays reserved for Pro, Enterprise, and Business Premium users and the API. The rollout started on 7 October.
On this page
GPT-6 everywhere, Astra behind the rope
OpenAI announced that GPT-6 is rolling out globally to every ChatGPT tier, alongside a new interface system called Intelligent UI, in a post titled GPT-6 and Intelligent UI for everyone published this week. The model lineup underneath stays tiered: GPT-6 Sol and GPT-6 Luna reach Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex, free and Go users get limited or expanded access, and GPT-6 Astra, billed as the company's most intelligent and aligned model, remains reserved for Pro, Enterprise, and Business Premium subscribers and the API.
The tiering means the week's two OpenAI stories connect: the same company publishing 372 mathematics results from an unreleased internal model is simultaneously moving its released flagship from a premium gate to the default answer engine for hundreds of millions of accounts.
Intelligent UI: answers as interfaces
The interface change is the more consequential half of the announcement for daily use. Intelligent UI lets ChatGPT combine text, visuals, and interactive elements in a single response: charts that render instead of being described, forms that collect structured input in place, diagrams that update, and side-by-side layouts for comparisons. In effect, the answer stops being a message and becomes a small application assembled per question, with OpenAI also promising faster responses in the new format.
There is a practical consequence for how people use the product, and it cuts both ways. Interfaces make verification easier, because a chart with the wrong numbers is checkable in a way a confident paragraph is not. They also make errors more persuasive, because an interactive, well-laid-out answer carries authority that plain text does not. EU users should note that the text portions of these responses now carry textGrain watermarking under the AI Act's provenance rules, which arrived this month.
What free and Go users actually get
The rollout language matters at the edges. Free and Go users receive limited or expanded access rather than the full frontier tier, and the announcement pairs the rollout with improved GPT-5.6 Sol and unlimited everyday use for free accounts, which keeps a capable, older model as the unlimited workhorse while GPT-6 sits behind whatever quota OpenAI sets.
For developers, the same models are available through the API, where Astra's premium positioning translates directly into per-token pricing. For everyone else, the honest summary is that the default ChatGPT answer is now generated by the company's flagship model and rendered as a small application, with the previous frontier model kept as the free tier's floor. Watch what that does to verification habits, because interfaces ask to be trusted more than paragraphs do.