GPT-6 Intelligent UI Features and Availability by Plan

ChatGPT’s GPT-6 intelligent UI combines charts and interactive tools with responses. Free·Go users are eligible for Luna, while users on the main paid plans are eligible for Sol. It is available in Chat, but not with the Pro reasoning option.

The rollout schedule and performance figures are based on OpenAI’s October 7, 2026 announcement.

What the intelligent UI does and how it works

The intelligent UI changes the layout of a response to suit the question. It can combine text with graphics, buttons, input forms, and charts. For simple explanations, it may show only text. It does not require a special prompt.

OpenAI designed the model to combine prebuilt screen components. A compiler processes the layout as it is generated and displays it on screen in sequence. This lets users see some content before the full response is complete. The approach is described in OpenAI’s launch announcement.

Models and rollout dates by plan

The launch announcement distinguishes between GPT-6 Sol and Luna by plan. Plus·Pro·Business·Enterprise were scheduled for a gradual rollout starting October 7, 2026. The rollout to Free·Go began the following day. A start date does not mean the rollout was complete for every account on that date.

This distinction appears in the October 7 entry of the ChatGPT release notes. However, official help articles do not agree on the Free·Go model name. The article dedicated to the intelligent UI says GPT-6 Luna. The general model help article also still describes it as GPT-5.6 Luna.

The announced eligibility and the model actually available in an account should therefore be treated separately. The difference between the articles alone does not establish that the gradual rollout is complete. Check the guidance shown in your account and the latest help articles to confirm which model you can use.

The Pro plan versus the Pro reasoning option

A Pro subscription itself does not exclude the intelligent UI. Support depends on the reasoning option selected. A reasoning option is a setting that determines how much in-depth reasoning to use for a response.

OpenAI’s help article “Intelligent UI in ChatGPT” states:

The Pro reasoning option uses GPT-6 Astra and does not support the intelligent UI.

Support up to Extra High does not mean every plan has access to that level. The model help article says Plus includes Medium and High. The same table excludes Extra High for Plus. That is why plan names and reasoning levels need to be read separately.

Enterprise admin settings and supported environments

In Enterprise, existing GPT-6 Sol access settings apply. Admins can turn off access. Workspace and role permissions also remain in place. This launch did not add a separate admin early access period. The source is the October 7 entry in the Enterprise and Edu release notes.

This change applies to Chat. The models powering Work and Codex do not change as part of this launch. That does not mean those products have no separate model updates.

Usage limits and what response speed means

The intelligent UI has no separate usage quota. Existing model, tool, and plan limits still apply. A calculation tool appearing on screen therefore does not guarantee unlimited use. Limits and reset information are available in the model and usage help article.

OpenAI compared how long it took responses to web search questions to begin. According to its October 7, 2026 announcement, GPT-6 Instant began responding 44% faster on average than GPT-5.6 Instant. This figure is not a reduction in the time needed to complete an entire task.

GPT-6 may continue reasoning or using tools after it starts responding. It may also add what it finds later without another request. The first content shown should therefore be distinguished from the completed response.

How to interpret situations where the intelligent UI does not appear

The absence of a chart alone does not mean the feature is unsupported. The model may choose text depending on the question. Account rollout status and the environment also need to be considered.

This table groups conditions from official articles by symptom. It is not a diagnosis of a particular account. In particular, a text-only response cannot distinguish a delayed rollout from normal behavior.

What to record when citing an interactive result

When recording an interactive result, it is better to capture the inputs along with the screen. This is a recordkeeping principle drawn from how results change when inputs change. It is not a separate citation standard set by OpenAI.

Some elements, including checklists, may retain their state after a refresh. Their state does not carry over between different conversations, however. Asking the same question in another conversation therefore will not, by itself, restore the earlier state.

• Input values and units: The conditions used for a calculation or chart.
• Time the result was checked: Information that shows whether the response was still being added to.
• Source data and reference date: Information needed to review the basis for the result.
• Text record of the result: A record of both the conditions selected on screen and the result.

For example, citing only the result of a cost-splitting tool leaves out the calculation conditions. The total amount, number of people, and splitting method must also be recorded to interpret the result. A visual layout is a way of explaining something, not a separate source of evidence in itself.

Frequently asked questions

Will every response become an interactive screen?

No. The format is selected to suit the question. A response that needs only text may be presented as text.

Can the intelligent UI be turned off completely?

The official help article describes settings for reducing visual elements. In the web settings, adjust the layout and visual elements under personalization. Some elements may remain even if you select a simple style.

Can I specify the response format I want?

You can state a format preference in custom instructions or an individual conversation. For example, you can ask for a table when comparing items. A specific screen layout is not guaranteed to be identical, however.

Can I keep using connected plugins?

Connected plugins remain available in ChatGPT. The intelligent UI help article gives Gmail and others as examples. Existing access permissions also apply to organization accounts.

Is it available to every account once the launch date has passed?

A rollout start date alone cannot confirm availability for an individual account. OpenAI described the rollout as gradual. Admin settings also affect organization accounts.