The rollout is staged to manage demand and ensure stability. Paid ChatGPT tiers, including Plus, Pro, Business, and Enterprise, are receiving the update first in the Chat tab. Free and Go users will gain access to the new features the following day. Enterprise access remains subject to existing workplace administrative settings. For the paid tiers, the underlying model is GPT-6 Sol, while the Free and Go tiers utilize GPT-6 Luna. Both variants are tuned for everyday conversation. Notably, the models powering the Work and Codex products remain unchanged by this specific update.
Intelligent UI allows ChatGPT to build the format of an answer around the specific task instead of forcing every request into a text block. The system can combine prose with graphics, tappable buttons, forms, charts, and other interactive elements. It chooses between side-by-side comparisons, diagrams, maps, or a plain textual answer depending on what is most useful. For example, the system can place a cooking timeline beside a recipe, plot road-trip stops on a map, or create on-demand tools such as savings calculators, bill splitters, and in-chat games. Learners can also adjust inputs to complex topics to better grasp the material.
Technically, the system relies on a library of native, streamable components and a compiler that processes the interface while the model generates it. This architecture allows the response to appear progressively on web and mobile devices without waiting for the entire answer to be generated. Training for this capability focused on the model’s decision-making regarding content, layout, visuals, and when interactive elements are warranted. OpenAI acknowledges that the model’s design judgment and the range of experiences it can create still require further development.
Performance improvements extend beyond visual presentation. GPT-6 can begin answering while it continues to think, interleaving reasoning with a series of useful partial responses intended to remain cohesive. In OpenAI’s internal evaluations of high-value everyday agentic tasks, the GPT-6 Extra High setting began answering as quickly as GPT-5.6 Medium while scoring above GPT-5.6 Extra High overall. For web-search questions, GPT-6 Instant started responding 44% sooner on average than GPT-5.6 Instant. The company states that the new generation more often identifies when to search and finds support for its answers.
Safety measures for GPT-6 build on previous advances, with stronger resistance to multi-turn attempts to bypass safeguards. Updated protections address cyberattacks, biological threats, and violence. The model is trained to use conversational context when risk is not visible from a single prompt, avoid needless refusals, and state more clearly when it lacks the information or tools required to complete a request.
This launch expands a model family that was introduced to paying customers last month. OpenAI frames Intelligent UI as a step toward a longer-term aim: allowing people to describe what they want to accomplish and work inside an experience created specifically around that task. The company cites this as part of its mission to bring the benefits of artificial general intelligence safely to humanity. While the update is now available to the broader user base, the specific capabilities of the Work and Codex suites remain separate from this ChatGPT rollout.
Model Variants and Availability
The distinction between model variants is clear in this release. GPT-6 Sol serves the paid user base, offering the full suite of capabilities associated with the higher-tier subscription. GPT-6 Luna is deployed for Free and Go users, ensuring that the core conversational benefits of the new generation are accessible to the widest possible audience. The separation allows OpenAI to manage computational resources while maintaining a consistent user experience across different service levels. The exclusion of Work and Codex models from this specific update suggests that those products follow a separate development and deployment timeline, likely focusing on specialized professional and coding tasks rather than general consumer interaction.
The introduction of streamable components addresses a common limitation in large language model interfaces, where users often wait for a complete block of text before any information is visible. By compiling the interface in parallel with the model’s generation process, OpenAI reduces perceived latency. This is particularly relevant for complex queries that require multiple steps of reasoning or data retrieval. The ability to render interactive tools like calculators and charts directly within the chat window reduces the need for users to copy and paste data into external applications, streamlining workflows for both professional and personal use.
As the rollout progresses to free users, the stability of the Intelligent UI system will be a key factor in its adoption. OpenAI’s admission that design judgment needs work suggests that users may encounter suboptimal layouts or interactive elements that do not perfectly match their intent. However, the core technical shift—from static text to dynamic, task-specific interfaces—positions ChatGPT as a more versatile tool for a wide range of digital tasks. The next confirmed step in the deployment timeline is the extension of these features to Free and Go tiers, which will determine the true scale of the new interface’s impact on daily user interactions.
