GPT-6 transforms ChatGPT's responses into interactive interfaces

GPT-6 brings Intelligent UI to ChatGPT, combining text, visuals, tools and interactive interfaces inside everyday conversations.

OpenAI is expanding GPT-6 across ChatGPT and using the new generation to rethink how the product delivers answers. The main addition is Intelligent UI: instead of systematically returning a stream of text, ChatGPT can now compose responses using graphics, buttons, forms, maps, controls and interactive experiences that work directly inside the conversation.

The goal is not simply to make ChatGPT more visual. GPT-6 can choose the format based on the question itself. A comparison might become a side-by-side interface, a complex explanation can turn into an interactive diagram, while a straightforward request can still receive a conventional text response. The presentation itself effectively becomes part of the generated answer.

In OpenAI's demonstrations, a dinner plan becomes an interface where the guest count can be changed and ingredient quantities adjust accordingly. A scientific concept can be explored by changing inputs and observing the result, while a simple request can generate a calculator, a bill splitter or even a playable game directly inside ChatGPT.

OpenAI says Intelligent UI relies on a library of native components combined with a compiler that assembles the interface progressively while GPT-6 generates its response. The company also expanded its training methods so the model can make decisions about layout, interaction and when a simple text response remains the better option.

GPT-6 also changes when answers begin appearing. ChatGPT can start responding while the model is still reasoning or using tools, allowing users to begin reading before the entire task has finished. Additional findings can then appear progressively without requiring another prompt. OpenAI confirms the behavior in its ChatGPT release notes.

OpenAI is also reporting improvements for questions that require web search. In internal testing, GPT-6 Instant started responding 44% sooner on average than GPT-5.6 Instant. The company says the model is also better at deciding when to search and at finding information that supports its answer. These figures come from OpenAI's own internal evaluations.

The rollout uses different models depending on the subscription tier. GPT-6 Sol powers ChatGPT for Plus, Pro, Business and Enterprise customers, while GPT-6 Luna is used for Free and Go. Paid-tier deployment began on October 7, with Free and Go rollout starting on October 8. Both support Intelligent UI. The Pro reasoning option powered by GPT-6 Astra does not currently support the new interface system.

The update applies specifically to the Chat experience. OpenAI says the models powering Work and Codex are not changing as part of this release. Intelligent UI is rolling out on the web and supported updated applications.

Alongside the product release, OpenAI published updated safety evaluations for GPT-6 Sol and Luna. The company reports stronger resistance to attempts to bypass safeguards, including adaptive attacks spanning several turns, along with updated protections for higher-risk misuse involving cybersecurity, biology and violence. More detail is available in OpenAI's October GPT-6 Sol and Luna safety update.

Intelligent UI ultimately represents a broader shift in what ChatGPT can become. Until now, users asked questions inside a mostly fixed interface. With GPT-6, parts of that interface can be assembled dynamically around the task itself. OpenAI's longer-term direction is clear: instead of requiring people to adapt to predetermined software interfaces, software could increasingly adapt itself to what people are trying to accomplish.