Summary
OpenAI says GPT-6 brings interactive, visual responses to ChatGPT, including diagrams, forms and task-specific tools. The staged global rollout begins on October 7, 2026, with availability varying by plan.
OpenAI announced GPT-6 for ChatGPT on October 7, 2026, introducing a feature it calls Intelligent UI. The company says the model can respond with interactive interfaces—such as diagrams, charts, forms and buttons—instead of relying only on text. It can also create small tools, such as a bill splitter or calculator, within a conversation.
The rollout is global but staged by subscription tier. OpenAI says GPT-6 with Intelligent UI begins reaching Plus, Pro, Business and Enterprise users on October 7, with Free and Go tiers following from October 8. Enterprise access depends on workplace administrator settings.
Interfaces shaped around a question
OpenAI says GPT-6 chooses how to present an answer based on the request. A comparison could appear side by side; an explanation might use an interactive diagram; and a question best answered with prose can still receive a text response. The announcement illustrates the approach with examples including a visual breakdown of a bicycle’s systems, an explanation of the central limit theorem and a road-trip plan displayed on a map.
The idea is to let people explore an answer or use a purpose-built tool without leaving the conversation. In a learning interface, for example, a user might change an input and see how the result changes. These examples describe the kinds of experiences OpenAI says GPT-6 can create; the announcement does not characterize them as a separate software product.
How the responses are built—and what changes
OpenAI says it built a library of native interface components and a compiler that processes the interface as the model generates it. This allows parts of a response to appear progressively rather than waiting for the full interface to be finished. The company also says it trained GPT-6 to make choices about layout, visuals and interaction, evaluating generated interfaces for clarity, usefulness and completeness. OpenAI says improving the model’s design judgment and expanding what it can create remain ongoing work.
GPT-6 also changes how ChatGPT presents answers while a model is still working. OpenAI says it trained the model to provide useful partial responses and build them into a cohesive answer. For questions requiring web search, the company reports that GPT-6 Instant starts answering 44% sooner on average than GPT-5.6 Instant. OpenAI also reports better results in internal evaluations of difficult questions and everyday agentic tasks; these are company-reported evaluations, rather than results from an independent benchmark described in the announcement.
OpenAI says GPT-6 improves resistance to attempts to bypass its safety training, particularly attacks that adapt over multiple turns, and is better at communicating the limits of its capabilities. For Plus, Pro, Business and Enterprise, ChatGPT uses GPT-6 Sol; Free and Go use GPT-6 Luna. Both are described by OpenAI as tuned for everyday conversation. The release applies to ChatGPT’s Chat experience; the models powering Work and Codex are not changing as part of this update.