OpenAI is rolling out GPT-6 with a new "Intelligent UI" feature that turns answers into interactive interfaces with charts, buttons, and forms. The model can respond while still thinking, cutting wait times by 44 percent. Paying customers get GPT-6 Sol, while free users get GPT-6 Luna.
SynthePulse Insight · AI deep reading
Version 1 · 1 source
OpenAI is rolling out GPT-6 to all ChatGPT users, replacing text-heavy answers with an interactive interface called "Intelligent UI," and claiming it can begin answering while still in a "thinking" state, cutting wait time by 44%. But the 44% and the internal test conclusions both come from OpenAI's own account, and the report provides no third-party verification.
OpenAI is rolling out GPT-6 to all ChatGPT users. According to the report, the most visible change in this upgrade is not the model itself but how answers are presented: GPT-6 carries a new feature called "Intelligent UI," presenting answers as an interactive interface with graphics, buttons, charts, and forms rather than plain text.
The second selling point, alongside it, is speed. The report says GPT-6 can begin answering while still in a "thinking" state, and OpenAI says this approach can cut wait time by 44% and may produce better answers. Taken together, this redesign tries to change two things at once: what users "see" and how long they "wait"—the former is form, the latter is feel.
The mechanism given in the report is that the interface format automatically adapts to the question. Comparison-type content is shown side by side, and explanatory content can be turned into interactive charts. This means the form of the answer is no longer manually chosen by the user but determined by the type of question.
Going further, users can also generate mini-tools directly in the chat, such as a savings calculator or a game. In other words, ChatGPT's output is no longer limited to content you "read" but can become something you "use."
It should be noted that the report does not say which specific question types trigger the interactive interface in Intelligent UI, nor whether it is enabled by default for all users and all questions. These boundaries remain unclear for now.
On speed, the only specific number in the report is "wait time cut by 44%," and the source is OpenAI's own claim. The report also mentions that in internal tests, GPT-6 using this method scored higher than its predecessor GPT-5.6 on difficult web search tasks.
Both of these need to be read as "self-reported." The report provides no third-party verification and does not disclose the measurement methodology or baseline for the 44%, so whether real-world experience matches remains unknown; "may produce better answers" is likewise OpenAI's expectation, with no independent evidence yet.
The internal test item is also limited in information: the report gives no specific scores for GPT-5.6 and GPT-6 on the difficult web search test, nor does it explain the test set or the margin of lead. So the conclusion that it "scored higher" can currently only remain a directional description.
The tiering given in the report is: paid users get GPT-6 Sol, free users get GPT-6 Luna. On rollout pace, this rollout starts globally the same day for Plus, Pro, Business, and Enterprise users, with free users and Go users following a day later.
Two things are not made clear here. First, the specific capability differences between GPT-6 Sol and GPT-6 Luna are not explained; second, after free users and Go users follow a day later, whether their features match the paid tiers is also unknown.
In addition, the report gives no specific parameters, context length, pricing, or available-region details for GPT-6. For a rollout aimed at "all users," the absence of this information means it is currently impossible to judge whether users across different tiers actually get the same set of capabilities.
The report mentions that Google rolled out a similar feature called "Neural Expressive" to its Gemini chatbot back in May. This provides a temporal reference point for Intelligent UI: the interactive answer format is not an OpenAI first.
But from the report, the two belong to the same category of interactive answer format, and the report does not explain the specific differences between them in implementation or capability, nor is there comparable public data. So "who does it better" cannot currently be judged; only the similarity in direction and the order in time can be confirmed.
The report's information sources are labeled as OpenAI and its System Card. This means the descriptions in the article about feature form, tiers, and rollout timeline originate from OpenAI's side; and the 44% wait-time reduction and internal test conclusions likewise come from that side.
Based only on the one report provided, it is impossible to independently verify OpenAI's 44% wait-time reduction and internal test conclusions. The report also provides no third-party verification.
Therefore, what can be treated as confirmed in this article is: GPT-6 is rolling out to all ChatGPT users, carries Intelligent UI, the interface automatically adapts to questions, mini-tools can be generated in chat, it can answer while "thinking," it comes in two tiers, Sol and Luna, and the rollout pace is paid users the same day and free users a day later. The 44%, the internal test scores, and "better answers" should all be placed in the category of OpenAI's own account.
This article is based solely on the one report provided (THE DECODER, October 7, 2026) and its labeled sources OpenAI and the System Card. The report provides no third-party verification; the 44% wait-time reduction and internal test conclusions are both OpenAI's own account; the report also gives no GPT-6 parameters, context length, pricing, available regions, capability differences between Sol and Luna, trigger conditions for Intelligent UI, or specific scores for GPT-5.6 and GPT-6 on the difficult web search test. Wherever these dimensions are involved, this article treats them as "unknown" and does not fill them in.
This GPT-6 redesign puts its emphasis on answer form and the feel of waiting: using Intelligent UI to replace text with an interactive interface, and using "answering while thinking" to compress the wait. But the key numbers supporting these two points—44% and the internal test scores—currently come only from OpenAI's side, and the measurement methodology, specific scores, and tier differences are all undisclosed. Until third-party verification appears, this upgrade is better understood as a clearly directional shift in form rather than a proven leap in performance.
Primary report
Primary source