← Back to feed

OpenAI is completing its GPT-6 rollout with the broadest expansion yet. As of October 8, ChatGPT’s Free and Go tiers are receiving access to GPT-6, following yesterday’s rollout to Plus, Pro, Business, and Enterprise. The update pairs model access with a new interface called Intelligent UI that replaces flat text responses with embedded interactive elements.

Intelligent UI activates automatically when ChatGPT determines a visual or interactive format would be more useful. No special prompt is required. Responses can include widgets, step-by-step interactive tools, data visualisations, and dynamic layouts. Users can reduce the feature’s output under Layout and visuals settings, which dials back visual elements but does not switch the interface to a fully text-only mode.

What Intelligent UI Actually Is

The framing from OpenAI is functional: “fast, interactive answers that make everyday questions more visual, complex topics easier to grasp, and interactive tools for your task available on the spot.” This is distinct from the canvas or code interpreter surfaces that already existed. Intelligent UI operates inside the main chat interface and applies to ordinary conversational queries.

The design shift moves ChatGPT closer to a mini-application layer than a text terminal. A recipe query could render with a step counter. A travel question could embed a map widget. A calculation could show workings in a structured table. None of this requires the user to change how they type questions.

Model Tier Structure

The GPT-6 family available after this rollout:

• GPT-6 Astra: Pro $100, Pro $200, Business, Enterprise. Requires Codex CLI 0.153.0 or newer for Work and Codex access.
• GPT-6 (default): Plus, Business, Enterprise, and now Free and Go users.
• GPT-5.6 Luna: Remains the default for Free and Go users where GPT-6 has not yet propagated; powers the Think mode on those tiers.

Pro subscriptions have weekly usage limits on GPT-6 Pro. After hitting the limit, ChatGPT falls back to GPT-5.6 Thinking at Medium effort.

Safety Assessment

OpenAI’s Deployment Safety Hub published the accompanying system card for the October 2026 update to GPT-6 Sol and GPT-6 Luna. Both models are classified as High capability in cybersecurity and biological and chemical domains under the Preparedness Framework. Neither model crossed the High threshold in AI Self-Improvement. OpenAI says the safeguards implemented match those documented in the GPT-5.6 system card for Sol and Luna variants.

The public framing of a mass consumer rollout alongside a safety card rating the model as High in cyber and bio is increasingly standard practice for OpenAI, but the combination is worth registering. The October update is the largest single-day expansion of GPT-6 access since the model launched in September.

Distribution

GPT-6 appears in the ChatGPT model menu without an “Astra” label for non-Pro plans. Users not yet seeing it are told the rollout is ongoing and account eligibility may vary. Managed workspace users are subject to their organisation’s model-access permissions.

Intelligent UI is rolling out concurrently with GPT-6 access and is not tied to a specific tier.