Published: October 7, 2026 · Last updated: October 7, 2026
TL;DR: OpenAI rolled out Intelligent UI today, a ChatGPT feature that swaps long text answers for actual working tools, buttons, calculators, forms, and charts generated on the spot for whatever you asked. It runs on GPT-6 and now reaches every ChatGPT account, including free users, not just paying subscribers.
Ask ChatGPT to help you compare two loan offers today, and you might not get a paragraph explaining the math. You might get an actual calculator, built on the spot, that you can type numbers into. OpenAI announced the change today in a post titled “GPT-6 and Intelligent UI for everyone,” and it’s a bigger shift than the name suggests.
Intelligent UI is OpenAI’s name for ChatGPT generating real interface elements inside a reply instead of describing things in text. A question about trip planning might come back with clickable itinerary cards. A request to compare products might produce a sortable table with buttons instead of a wall of bullet points. TechCrunch covered the launch under the headline “ChatGPT is getting a lot more visual, with the launch of a new interface,” and that’s really the point: the chat box stops being a text terminal and starts behaving more like an app that assembles itself around whatever you asked for.
The feature sits on top of GPT-6, specifically the Sol and Luna model pair OpenAI shipped a couple of weeks ago. Sol is the larger, more capable version; Luna is the smaller, cheaper one built for speed. Until today, that pair mostly lived behind a paywall. Sol and Luna rolled out in ChatGPT Work, Codex, and the API for Plus, Pro, Business, Enterprise, and Edu accounts, while free and Go users only got a limited trial of Luna in the desktop app. Today’s announcement closes that gap. Intelligent UI, and the GPT-6 models underneath it, is now available across every ChatGPT tier, including the free web and mobile apps that make up the bulk of OpenAI’s reported 1.2 billion users.
That scale is the part worth sitting with. OpenAI isn’t testing a new interface on a slice of power users first. It’s changing how answers render for essentially everyone who opens ChatGPT, on day one. That’s an aggressive rollout schedule for a feature that depends on the model correctly deciding, in real time, whether a given question deserves a chart, a form, or just a normal sentence. Get that call wrong often enough and the feature reads as a gimmick instead of a genuine upgrade.
The timing also lines up with a broader price and feature fight among the big model labs this week. Anthropic shipped Claude Haiku 5.5 in the same stretch, cutting its API pricing by 90% in a move multiple outlets described as matching GPT-6 Luna’s pricing directly. Reporting on Intelligent UI itself noted that OpenAI’s launch landed about six weeks after a research write-up on “generative UI” concepts started circulating, suggesting this wasn’t a surprise to anyone paying close attention inside the industry, just a race to actually ship it first at consumer scale.
For anyone building on top of ChatGPT, the practical question is what this does to how people consume answers. A generated chart or form inside a chat reply means less copy-pasting into a spreadsheet, but it also means OpenAI is quietly becoming the interface layer for tasks that used to require opening a separate app or website. That’s worth watching if your business depends on people clicking through to a tool instead of ChatGPT just building a smaller version of it for them on the spot.
Related: This isn’t the first time OpenAI has changed what a ChatGPT reply can contain. It already added a virtual try-on feature that generates images of you wearing clothes before you buy, and it’s been testing visual ads inside ChatGPT’s image generation results around the same time.
Bottom Line: Shipping a feature like this to every tier on day one, instead of gating it behind Plus or Pro first, tells you OpenAI thinks Intelligent UI is a retention play, not a premium perk. If it works as advertised, it’s a genuine improvement over scrolling through paragraphs for a number you could’ve just typed into a box. If the model misjudges what deserves a widget versus a sentence, this becomes the next thing people complain ChatGPT does too much of. Early days, but worth trying the next time you ask it something that has an obvious visual answer.
Follow Teck Hustlers: Facebook · X · Instagram · LinkedIn · TikTok · Pinterest
