GPT-6 rolls out to free ChatGPT with interactive answers
Paid plans get GPT-6 Sol and Free and Go get GPT-6 Luna, and answers can now include charts, maps, forms and small tools built on the spot.
Updated
In short
- 01OpenAI began rolling out GPT-6 in ChatGPT on October 7, 2026, for paid plans, and from October 8 for Free and Go.
- 02OpenAI says GPT-6 Instant starts answering web-search questions 44% sooner on average than GPT-5.6 Instant, based on its own tests.
- 03OpenAI's system card says GPT-6 Sol and Luna did worse than GPT-5.6 on self-harm safety tests, in cases it calls borderline.
The story
OpenAI started rolling out GPT-6 in ChatGPT on October 7, 2026, and expanded the rollout to Free and Go users the next day. Its launch post says the new model is built for the more than 1.2 billion people who use ChatGPT each week.
The headline feature is Intelligent UI, which lets answers mix text with charts, maps, forms and buttons. On a press call, product manager Aarush Selvan said "the most helpful answers aren't just text," TechCrunch reported. TechCrunch's read is that the visuals could widen ChatGPT's audience, a goal OpenAI didn't state.
GPT-6 builds charts, maps and small tools into answers
GPT-6 decides how each answer should look. OpenAI's examples include a cooking timeline next to a recipe, road trip stops on a map, a bill splitter and a savings calculator. When plain text works best, it still answers in plain text.
OpenAI built a library of ready-made interface parts, plus a compiler that draws them while the model is still writing.
GPT-6 can also start answering before it finishes thinking. OpenAI says GPT-6 Instant starts answering web-search questions 44% sooner than GPT-5.6 Instant. That figure comes from OpenAI's own tests, and no one else has checked it yet.
OpenAI's help page on GPT-6 adds limits the post leaves out. Intelligent UI doesn't work at the Pro thinking level, and the older Mac and Windows desktop apps don't support it. A Layout and visuals switch in web settings cuts back the visuals but doesn't remove them all.
The system card shows GPT-6 slipped on some safety tests
The GPT-6 system card says the Sol and Luna models in ChatGPT are new October versions. Codex and ChatGPT Work still use the September versions. The card reports stronger resistance to jailbreaks than GPT-5.6.
It also shows statistically significant drops on OpenAI's self-harm tests for both models, and on gore and sexual content for Luna. OpenAI says it reviewed the failures and found them borderline but generally safe.
The Hacker News launch thread drew 458 comments. Recaply read the top 30 on October 11, a snapshot rather than a survey.
About a quarter welcomed the idea, and about one in five worried that visuals would bury simple answers. Six more doubted the tools' value or the accuracy of generated diagrams.
I want a quick answer, and I don't care for the boilerplate UI.
ddxv on Hacker News, October 7
GPT-6 with Intelligent UI is rolling out worldwide in ChatGPT's Chat tab, on the web and in updated mobile apps. Plus, Pro, Business and Enterprise run GPT-6 Sol, Free and Go run GPT-6 Luna, and Enterprise access depends on admin settings. When Recaply checked on October 11, the help page still listed GPT-5.6 Luna for Free and Go users.
Why it matters
Chatbot answers have mostly been walls of text, and Intelligent UI turns some into small apps built for one question. It's on by default for free users too, while OpenAI says the model's design judgment still needs work. Simple web calculators and bill splitters now compete with the answer box itself.
Share
