GPT-6 Comes to All of ChatGPT, With Interactive Answers Built In
OpenAI is rolling out GPT-6 to ChatGPT worldwide, paired with a new system called Intelligent UI.
The rollout began on October 7, 2026, and continued into October 8 across all ChatGPT tiers. Paid plans including Plus, Pro, Business and Enterprise were first globally, with Free and Go accounts scheduled to follow on rollout day Engadget. OpenAI described the release as GPT-6 for all ChatGPT users, with faster responses and visual, interactive experiences OpenAI.
The plan split comes down to two versions, or weights, of the model. Plus, Pro, Business and Enterprise run on GPT-6 Sol. Free and Go run on GPT-6 Luna. Luna is less capable than Sol. API documentation describes Sol as the balance of intelligence and cost, and Luna for cost-sensitive, high-volume work, which matches how ChatGPT routes them.
Intelligent UI is the main functional change. ChatGPT can now answer with fully interactive interfaces instead of static text. GPT-6 was trained to make more visual responses, including graphics, tappable buttons, forms and charts. It can combine text, visuals and interactive parts to fit a question, such as side-by-side comparisons, diagrams and tools. Text is still available, and the model can choose plain text when that is clearer.
Latency and reasoning control are the other changes. GPT-6 can start answering while it is still thinking, a streaming approach meant to cut the wait on complex tasks. In ChatGPT, users can choose how much the model thinks OpenAI Help Center. In practice, that is control over the tradeoff between reasoning effort and how fast the first words appear, with more detail filled in as the answer continues.
OpenAI said it would roll out Sol and Luna gradually through the day to keep service stable. That is standard capacity management for a change this large. Replacing inference capacity in stages, meaning the servers that run the model, avoids sudden load spikes when many users try the new output types at once.
The release follows a canceled nearby launch. OpenAI shelved GPT-6.1 Astra, which had been planned for October, after internal safety tests raised concerns Reuters. Astra is not in the ChatGPT rollout. Sol and Luna are the shipping models.
The broader context here is a move of interface work into the model itself. Chatbots have long produced markdown, code snippets and images. What changes with Intelligent UI is that layout, choice of controls and interaction state become direct model outputs, not fixed frontend templates around an answer. That moves prompt design and testing closer to product design. It also raises practical engineering questions around determinism, or getting the same result twice, accessibility, checking input in generated forms, and how to test a response that can render differently on each run.
In my view, worth flagging is how quickly this will feel normal. My children never learned to fear a blank command line because they never met one. They expect software to show, not tell. A model that offers buttons, charts and tools when useful fits that habit and removes the copy-paste step between answer and action. The long-term gain is less time translating machine output into human work, which is what each earlier platform shift quietly delivered.
