Tibo's promise to "release or reset for 28 consecutive days" comes on the third day, and today's benefit is a banked reset for paid users.
The real protagonist is GPT-6.
Last month, OpenAI launched its first GPT-6 models for paid users, and now it has expanded its coverage further.
Just recently, OpenAI announced the launch of GPT-6 for all ChatGPT users worldwide (including free and paid users).
These models will replace the GPT-5.6 Sol and GPT-5.6 Luna in ChatGPT.
Access to GPT-6 Sol and GPT-6 Luna via Codex and ChatGPT Work will continue to use previously released versions.
In this upgrade, the most noteworthy new capability is the Intelligent UI (smart interface).
It can quickly generate interactive answers, make everyday problems more intuitive, help users understand complex topics more easily, and create interactive tools that can be directly used for current tasks.
And just hours before the release of GPT-6, Sam Altman made a meaningful statement: "Tonight, as I look up at the stars, I feel a tremendous awe. Your sea is so rough and my ship is so small".
Putting GPT-6 at an all-encompassing node, the phrase has some resemblance to the increasingly powerful modeling capabilities and the increasing number of things AI can do, but in the face of greater unknowns, humans are still just getting started.
Intelligent UI
Unlike previously primarily text-based responses, GPT-6 can now independently decide how to combine elements such as text, images, graphs, buttons, forms, and directly generate interactive response interfaces based on user questions.
Specific presentation varies with the task. For example, in product or program comparisons, ChatGPT can be used for parallel presentation; when interpreting complex concepts, interactive graphics can be generated; and if simple text is sufficient, it will continue to be used in traditional text form.
OpenAI hopes to make everyday questions more intuitive. For example, when designing a plan for a roast on a Sunday, recipes and schedules can be displayed simultaneously; when planning a self-driving route, stops along the route can be directly displayed on a map and marked where it is worth going.
It is easier to understand complex problems. With visualization and interactive interpretation, users can more intuitively explore how things work, adjust inputs and see how the results change.
Users can also directly have ChatGPT create a tool for their current task. For example, they can generate a calculator to see how their savings grow, a bill-sharing tool to use at a friend's dinner party, or even generate a mini-game that can be played directly in a conversation.
How the Intelligent UI works
From a technical point of view, the core of the Intelligent UI is to enable GPT-6 to select the most appropriate interface format for different problems, while maintaining a consistent experience on the web and mobile.
For this purpose, OpenAI has built a native library of components that supports flow-generated and supports the development of an interface compiler. The model can call on these components when generating responses and decide on its own how to combine elements such as text, graphs, buttons, tables; the compiler receives the model output on the one hand, and the interface is progressively rendered on the other, so that the user does not have to wait for the entire response to be generated, and the interactive interface can appear continuously.
In order to give GPT-6 better design judgment, OpenAI also expanded the training methodology to help models better judge content, layout, visual presentation, and interaction. During the training, OpenAI will evaluate the interface generated by the model in terms of clarity, usability, and completeness.
Through these trainings, GPT-6 learns how to use component libraries and make more rational design decisions, including how to organize information clearly, when to join interactions, and when simple text responses are enough.
Of course, there is still room for improvement in the scope of model design judgments and generated content. However, the Intelligent UI has already marked an important change in the way people interact with ChatGPT.
In the past, different issues were often placed in the same dialogue framework; now ChatGPT can combine interfaces, data, and operations to create completely new ways for users to explore, understand, and interact with information.
With GPT-6, ChatGPT can now start generating answers while continuing to think.
With the advent of rational models, users can make the problem more difficult to solve, but often have to wait for the model to finish thinking before seeing the answer.
Now, ChatGPT can think and answer erroneously. OpenAI has trained GPT-6 to take into account the user's waiting time and gradually organize and output responses based on the knowledge and phased conclusions currently in its possession.
GPT-6 also learns to continuously supplement effective information through multi-phase responses, avoiding meaningless filling as much as possible, while maintaining overall consistency in the final answer and actually approaching a one-time complete written answer.
In an internal evaluation of high-value, daily Agent tasks, the GPT-6 Extra High's response time was equivalent to GPT-5.6 Medium, but its overall score was higher than GPT-5.6 Extra High.
For questions that require a web search, the average response time for GPT-6 Instant is 44% earlier than for GPT-5.6 Instant.
GPT-6 also improves the quality of responses when searching for web pages. It can more accurately determine when it is necessary to search for external information and can more reliably find relevant content that supports the answers.
In an internal assessment of high-difficulty problems, GPT-6 was more frequently accurate in capturing and returning key points in user problems than GPT-5.6.
Available scope
Starting today, the GPT-6 with Intelligent UI is being rolled out to global Plus, Pro, Business and Enterprise users via ChatGPT's Chat page.
Starting tomorrow, the coverage will be further extended to Free and Go users.
In ChatGPT:
Plus、Pro、Business and Enterprise users are supported mainly by GPT-6 Sol;
Free and Go users are mainly supported by GPT-6 Luna.
Both models were specifically optimized for everyday conversational scenarios.
This update applies only to ChatGPT's Chat experience. The models currently used in ChatGPT Work and Codex will not change with this release.
What happens next?
OpenAI says its mission is to safely deliver the benefits of general AI to all humanity. GPT-6 brings the latest generation of smart capabilities to more than 1.2 billion users who use ChatGPT every week.
Intelligent UI is seen as an important step towards the next phase of ChatGPT: In the future, ChatGPT will be able to dynamically shape matching interfaces based on the tasks the user is completing.
In the long run, OpenAI hopes that ChatGPT will be able to instantly create the user interface needed for more everyday tasks. Users only need to describe what they want to accomplish to complete tasks directly in the interactive experience generated around that goal.
Over the past few decades, people have had to learn how to use software to look for functionality in fixed interfaces pre-designed for specific tasks.
OpenAI believes that the future will be different: people will no longer need to constantly adapt to software, and software will begin to actively adapt to people.
The link is as follows: