On October 7, 2026, OpenAI brought GPT-6 to everyone in ChatGPT, free plan included. The big change is not a benchmark. It is the shape of the answer. ChatGPT now often replies with a small interactive interface instead of a long block of text. OpenAI calls it Intelligent UI.

What Intelligent UI is

OpenAI trained GPT-6 to compose responses from text, visuals and interactive elements, and to choose the mix based on your question. A response can include graphics, tappable buttons, forms, charts and interactive experiences you use right in the conversation.

The format changes with the question. A comparison might come side by side, an explanation might become an interactive diagram, and when plain text is the most useful answer, you still get plain text.

OpenAI's demos

OpenAI's demos show three kinds of tools built on request: a savings calculator, a bill splitter, and a game you can play in the chat.

The bill splitter is the one that sold me. The prompt is a photo of a receipt plus "Bill splitter for a party of five based on what we ate, can you make it look like a clipboard." GPT-6 returns a working splitter: a name for each guest, each dish with checkboxes for who shared it, tax, and a tip selector, with a running total at the bottom.

OpenAI also runs the same prompt on both models: a plan for a Sunday lamb roast with friends, guest count still unknown.

  • GPT-5.6 Instant gives a long, good text answer with a table of lamb weights by guest count.
  • GPT-6 Instant gives a menu, a guest counter with plus and minus buttons that recalculates the shopping list, and a step by step cooking checklist.

Same information, but one of them you can actually use while you cook.

How it works

OpenAI built a library of native, streamable components plus a compiler that processes the interface while the model is still generating it. So the interface appears piece by piece instead of after the full answer is done.

GPT-6 can also begin answering while it keeps thinking. For questions that need web search, OpenAI says GPT-6 Instant starts answering 44% sooner, on average, than GPT-5.6 Instant. In OpenAI's side by side timing clips, at the 6 second mark GPT-5.6 is still searching while GPT-6 already shows a table. These are OpenAI's own internal measurements.

Who gets which model

  • Plus, Pro, Business and Enterprise: GPT-6 Sol. Rollout started October 7.
  • Free and Go: GPT-6 Luna. Rollout started the next day.
  • Enterprise availability depends on workplace admin settings.
  • This update covers the Chat experience only. The models behind Work and Codex do not change.

OpenAI says more than 1.2 billion people use ChatGPT each week, and the free tier is included, so most of them will see this.

The catch

OpenAI says itself that there is still work to do on the model's design judgment and on what it can create. All the examples above are OpenAI's own curated demos. I would test it on your real daily questions before you decide it beats a text answer. If you are on the API, note that this launch is about ChatGPT; for the API side of the last generation, see my post on GPT 5.6 Luna's price cut.

My take

This is the right direction. Most "answers" I ask ChatGPT for are really small tools I want to use once: split this bill, scale this recipe, compare these options. Getting the tool instead of instructions saves real time. The open question is how often the model picks the right format on messy, real prompts, and that only shows up after a few weeks of use.

Sources: OpenAI: GPT-6 and Intelligent UI for everyone and the GPT-6 system card