Ask ChatGPT how a seven-speed bicycle mechanism works and, instead of a long text, it can return a diagram with buttons that highlight each part. This capability, dubbed Intelligent UI, arrives with GPT-6, the new model that OpenAI is beginning to roll out in ChatGPT starting today.
The change is significant because it modifies the basic way a chatbot communicates. For years, these tools have responded almost exclusively with blocks of text. Now, the response can become something the user interacts with directly within the conversation.
What is Intelligent UI and how does it change ChatGPT responses?
As OpenAI explained in its announcement, GPT-6 combines text with diagrams, charts, forms, and clickable buttons, all integrated into the same chat. The company states that it specifically trained the model to decide when a visual response is more useful than a purely textual one, and to automatically provide the appropriate format.
The fundamental difference lies in the type of experience. Until now, the user read an explanation. With Intelligent UI, they can manipulate that explanation: click on a part of a diagram, move a tile, or enter data into a form generated on the fly.
Examples shown by OpenAI: from bicycles to mahjong
To illustrate the feature, OpenAI shared two specific cases. In the first, when asked about the design of a seven-speed bicycle, ChatGPT generates an interactive diagram with buttons that highlight each component of the gear-shifting mechanism.
In the second example, the chatbot explains the rules of mahjong. To do this, it creates images of each game tile and organizes them into navigable categories, in a format that feels more like an educational card than a traditional paragraph of text.
Calculators, games, and applications generated within the chat
Beyond explanatory diagrams, Intelligent UI allows you to ask ChatGPT to build small functional tools to solve specific tasks. OpenAI mentions as examples a retirement savings calculator, a retro-style game, and an app to split a bill among several people.
These tools appear integrated directly into the response, without the user having to leave the conversation. A possible scenario would be organizing a dinner for several people and asking ChatGPT to generate a bill splitter on the spot based on what each person consumed, without needing to copy figures into an external spreadsheet.
GPT-6 Sol and GPT-6 Luna: how access is distributed
OpenAI introduced the first models of the GPT-6 family last month. Now it is integrating them into ChatGPT under two variants: Sol and Luna. The rollout is not the same for all users and does not start on the same day.
Subscribers to the Plus, Pro, Business, and Enterprise plans receive global access starting today with the mid-range GPT-6 Sol model. Users of the Go and free plans will have to wait until Thursday and will receive GPT-6 Luna, a more efficient but different variant in terms of capacity.
The interactive Intelligent UI will reach all subscription levels, although the engine that generates the responses will not be the same depending on the plan subscribed to.
Improvements in web searches and security, according to OpenAI
The company also claims that GPT-6 allows ChatGPT to perform more accurate web searches and offer faster partial responses while it continues to gather information. In practice, this should mean that the user sees a preview of the response before the full query finishes processing.
OpenAI also claims that GPT-6 showed greater resistance to attempts to bypass its safety training. This is a statement included in the company's official press release, which does not detail the methodology used to measure that improvement.
What this change means for applications and developers
If a chat is capable of instantly generating a calculator or a small game, the line separating a conversational assistant from an independent application becomes more blurred. Simple utilities that are currently sought in app stores or specific websites could begin to be generated on demand within the conversation with ChatGPT itself.
For those who develop simple tools, this move introduces a new form of direct competition with OpenAI's assistant. For the company, the obvious goal is to keep users on its own platform for longer. It is worth remembering that this is a corporate announcement whose real impact will depend on user behavior once the feature is available on a massive scale.
What changes for those who use ChatGPT daily
Those who use ChatGPT frequently will notice more visual responses when explaining mechanisms, learning game rules, or performing calculations. It will also be possible to explicitly request a calculator, a form, or a diagram, instead of waiting for the model to decide the response format on its own.
Those with a Plus plan or higher can already try GPT-6 Sol starting today. Users of the Go and free plans will have to wait until Thursday to access GPT-6 Luna. In any case, if a calculator generated by the chat is used for relevant decisions, such as retirement savings, it is prudent to verify the results with another source before making decisions.
Frequently Asked Questions
What is Intelligent UI in ChatGPT?
It is a GPT-6 feature that allows ChatGPT to respond with interactive diagrams, charts, forms, and buttons, in addition to text, within the same conversation.
Who can use GPT-6 in ChatGPT and when?
Subscribers to the Plus, Pro, Business, and Enterprise plans get access to GPT-6 Sol starting today. Users of the Go and free plans will receive GPT-6 Luna starting Thursday.
What is the difference between GPT-6 Sol and GPT-6 Luna?
GPT-6 Sol is the mid-range model intended for higher-tier paid plans, while GPT-6 Luna is a more efficient variant reserved for Go and free plans.
What kind of tools can ChatGPT generate with this feature?
Examples shown by OpenAI include interactive diagrams, a retirement savings calculator, a retro-style game, and an app for splitting bills among several people.
Does GPT-6 also improve ChatGPT’s security?
OpenAI claims the model shows greater resistance to attempts to bypass its safety training, although the company does not publicly detail how it measured that improvement.
Is it reliable to use a calculator generated by ChatGPT for financial decisions?
It can be useful as an initial reference, but it is advisable to verify the results with another source before making important decisions, such as those related to retirement savings.
