OpenAI put GPT-6 in front of every ChatGPT tier at once. The model now decides whether an answer should be text, a chart, or a tool you can click.

OpenAI began rolling out GPT-6 inside ChatGPT’s Chat tab on 7 October 2026, starting with Plus, Pro, Business and Enterprise accounts, with Free and Go tiers following on 8 October. The rollout does not touch ChatGPT’s separate Work or Codex modes, which keep their existing models. OpenAI says the change reaches more than 1.2 billion people who use ChatGPT weekly.

Paid tiers get GPT-6 Sol; Free and Go tiers get GPT-6 Luna, which 18BYTE has previously profiled the Sol/Terra/Luna naming scheme for under the GPT-5.6 generation. Alongside the model swap, OpenAI introduced a feature it calls Intelligent UI, which lets ChatGPT respond with graphics, tappable buttons, forms and charts, or small interactive tools built on the fly, rather than defaulting to plain text.

Why it matters

This is the broadest simultaneous rollout OpenAI has described for a flagship model change, covering free and paid users at once rather than staging access by tier over weeks. For a product used by over a billion people weekly, even a modest shift in default response format changes how ChatGPT is used for everyday tasks like comparisons, planning and calculations.

Intelligent UI also pushes ChatGPT further from a pure chat interface toward a tool-generation surface, which matters for any product that competes with ChatGPT on task completion rather than conversation alone.

What Intelligent UI actually changes

According to OpenAI’s own announcement, reproduced consistently across reporting, GPT-6 was trained to compose each response from text, visuals and interactive elements, choosing the format automatically based on the question. A comparison can render side by side, an explanation can become an interactive diagram the user can adjust, and a one-off request can produce a small tool, such as a savings calculator, a bill splitter, or a simple game played inside the conversation. ChatGPT can still answer in plain text when OpenAI judges that to be the better fit.

OpenAI also says GPT-6 can begin answering before it finishes reasoning, with the interface filling in progressively, and that GPT-6 Instant starts answering questions that need a web search 44 percent sooner on average than GPT-5.6 Instant. That figure is OpenAI’s own comparison against its prior model and has not been independently measured.

The primary announcement sits behind a fetch block

OpenAI’s own release page for this announcement, and its help-centre articles describing Intelligent UI and the Sol/Luna split, returned access errors on every attempt to fetch them directly. The description above is reconstructed from several independent outlets with named reporting — Engadget, MacRumors, The Decoder and others — whose accounts of the rollout dates, tier mapping and feature description all agree with each other and with text quoted from OpenAI’s announcement on OpenAI’s own developer community forum, posted by forum members rather than an OpenAI staff account.

That convergence across multiple independently reporting outlets is reasonable grounds to report the story, but it falls short of this publication’s bar for direct primary confirmation, since none of us could independently load openai.com or help.openai.com to verify the wording first-hand.

What this means for buyers

Anyone evaluating ChatGPT against competing assistants should test Intelligent UI directly rather than assume it behaves like a chat response: a tool that renders an interactive calculator or chart may not export, copy or embed the way plain text does, which matters for workflows that pipe ChatGPT output elsewhere. Teams building on the API separately from ChatGPT are unaffected for now; this change is scoped to the ChatGPT consumer product’s Chat tab, not the Work or Codex modes or the developer API.

Buyers comparing small-model cost and capability should note that Free and Go users are on GPT-6 Luna, not the same model paid tiers get, so experience reports from free users do not describe what Plus and above actually run.

What would change our reading

Direct access to OpenAI’s own announcement and help-centre pages would let us confirm the exact wording OpenAI uses for Intelligent UI and the 44 percent web-search figure’s methodology, rather than relying on consistent secondary reporting; we will re-check this once those pages are reachable. Independent testing of Intelligent UI across both Sol and Luna variants over the coming weeks, particularly whether interactive elements degrade gracefully when copied or exported, would also sharpen the buyer guidance above.