Why ChatGPTs GPT-6 Ditched Plain Text for Tappable Interfaces

ChatGPT has spent four years answering almost everything with a wall of text. That changes with GPT-6, which OpenAI began rolling out October 7 alongside a feature called Intelligent UI — letting the chatbot respond with diagrams, tappable buttons, charts, forms and small embedded tools instead of paragraphs alone.

OpenAI product manager Aarush Selvan framed the shift around a real limitation, telling reporters "ChatGPT has predominantly been a text-based interface," but that the most helpful answers for tasks like planning a trip or learning something new "aren't just text." A few details define how the system actually decides what to build:

• Intelligent UI runs on native, streamable components rendered through a compiler that builds the interface as GPT-6 generates its response, according to OpenAI's Help Center
• Sample use cases include an interactive recipe widget that adjusts ingredient quantities by guest count, comparison tables shown side by side, and small on-demand tools like a bill splitter or calculator
• GPT-6 can also begin responding while still "thinking," a change OpenAI says cuts wait times by 44% and scored higher on difficult web searches in internal testing

The rollout splits across two model variants — paid Plus, Pro, Business and Enterprise accounts get the mid-range GPT-6 Sol immediately, while Free and Go users receive the lighter GPT-6 Luna starting October 8.

Both are tuned for everyday conversation rather than specialized reasoning tasks. OpenAI says users who prefer plain text can adjust the feature's output, and plans to keep expanding which interface types GPT-6 can generate.

With more than 1.2 billion people using ChatGPT weekly, the bet is straightforward: if an AI chatbot can build the right interactive tool on the fly, it may compete less with other chatbots and more with the purpose-built apps people currently open to do the same task.