ChatGPT now answers with interfaces, not paragraphs. That changes where a link can go.

Since October 7, 2026, ChatGPT no longer answers only in paragraphs: OpenAI has shipped GPT-6 along with a feature called Intelligent UI, which lets the model build its answer out of illustrations, buttons, forms, charts and interactive components, and pick that format itself based on the question. For a company's visibility, the format matters, because the format of an answer decides how much room a link has in it.

What exactly did OpenAI announce on October 7, 2026?

OpenAI published an announcement titled "GPT-6 and Intelligent UI for everyone" on October 7, 2026, opening its latest model generation to every ChatGPT user. The global rollout began on October 7 in the Chat tab for the Plus, Pro, Business and Enterprise plans, which run on a model called GPT-6 Sol, then extended on October 8 to the Free and Go plans, which run on GPT-6 Luna. OpenAI states in that announcement that more than 1.2 billion people use ChatGPT every week. The models behind Work and Codex are unchanged by this release.

Intelligent UI is ChatGPT's ability to answer with an interface instead of a block of text. OpenAI writes that it trained GPT-6 to compose answers using text, visuals and interactive elements, choosing how to combine them according to the question. The example OpenAI leads with in its announcement is a meal plan: the answer renders a menu, a cooking timeline and a guest counter that recalculates shopping quantities as you change it. That answer is not an article any more. It is a small tool.

Does an interface leave room for a link?

A paragraph can carry a citation at the end of a sentence. An interface made of buttons, a form and a calculator offers no equivalent place to hang one. That is the real question Intelligent UI raises for a company website: text with a link attached is no longer the default shape of a ChatGPT answer, it is one option among several, and the model is the one choosing.

OpenAI's announcement of October 7, 2026 lists the components these interfaces are built from, illustrations, buttons, forms, charts and interactive experiences, and says nothing about how sources appear inside them. So nobody can claim today that these answers cite fewer sites: that would be an extrapolation, and this blog does not make them. What is documented is that the shape of the answer is now the model's call, and that this call happens before any question of citation.

What does "better at deciding when it needs to search" actually mean?

OpenAI writes that GPT-6 is better at judging when a web search is needed, and that it more reliably finds information which actually supports its answer. Translated for a business owner: before the question of which site ChatGPT cites, there is an earlier decision, made by the model, about whether to consult the web at all or answer without it. That decision cannot be steered from your website, and it is not visible in the answer on screen.

OpenAI puts a number on the speed gain: for questions that require a web search, GPT-6 Instant starts answering 44% sooner on average than GPT-5.6 Instant, measured by OpenAI on its own production search traffic. OpenAI presents these changes as quality improvements and nowhere states that the model searches the web less often. The judgement call exists, it moved on October 7, 2026, and its outcome stays invisible from the outside.

What does this change for a small or mid-sized business?

For a small or mid-sized business, OpenAI's October 7, 2026 announcement calls for no action this week, but it does confirm where effort pays off. Search engines still carry the majority of commercial discovery, and they are shedding clicks rather than disappearing: Ahrefs measured across 300,000 keywords that the click-through rate of the number one organic result is 58% lower when an AI overview is displayed, comparing December 2023 with December 2025. Set against the 1.2 billion weekly ChatGPT users OpenAI reports, the practical consequences come down to three.

• A customer who gets a complete, costed, interactive plan inside ChatGPT has no reason left to open three pages and assemble it themselves.
• Search still sends the larger share of traffic, which is exactly why a page has to be worth citing in both places rather than tuned for one.
• Your own website stays the only surface where you control the content, the public files and the structured data, whichever model comes to read them.

The third point is the one that survives version changes. This blog documented it in "ChatGPT is citing YouTube and Facebook far less. The lesson is about what you do not own.": surfaces you do not own can lose their standing overnight, with nobody telling you. A listing on someone else's platform depends on a judgement call you do not make. A page on your own domain depends on you.

Should you react to GPT-6 at all?

No, and that is the honest answer. Changing your website because OpenAI changed model is chasing a target that moves faster than your pages: a model version is replaced within months, while a well-structured page stays readable by whichever model comes next. The signals that make a site legible to an answer engine, clean HTML, accurate structured data, reachable public files, were the same on October 6, 2026 and on October 8, 2026.

The reflex to avoid is testing the new model by asking whether it knows your company. This blog explained why in "Asking ChatGPT whether it knows you measures nothing": an answer you get on one evening measures the model's variability, not your visibility. And if your site has nothing worth citing, meaning no page that genuinely answers a question a customer is asking, no version of ChatGPT will change that. In that specific case the budget belongs in content, not in measurement.