CONFIRMED
OpenAI announced on October 7, 2026 that GPT-6 now powers ChatGPT across all plans. The global rollout began the same day for Plus, Pro, Business and Enterprise in the Chat tab, and expands from October 8 to the Free and Go plans. Paid subscribers use GPT-6 Sol; Free and Go get GPT-6 Luna, both in their October versions and tuned for everyday conversation. The company says ChatGPT has more than 1.2 billion weekly users, a figure it reports and that has not been independently audited.
Answers are no longer just text
The headline feature is called Intelligent UI. Instead of returning only paragraphs, the model can compose its answer with charts, tappable buttons, forms, maps and small interactive experiences inside the conversation. OpenAI's own examples include a side by side comparison, a road trip with stops on a map and a calculator built on the spot to explore how savings could grow. When plain text is the most useful answer, ChatGPT still replies in text. According to the company, the interface is assembled from a library of native components and appears progressively while the model is still generating.
Thinking while answering
The second change is interleaved reasoning: ChatGPT gives an initial answer and then keeps thinking to add detail. In an internal evaluation of everyday tasks, OpenAI says GPT-6 at Extra High begins answering in the same time as GPT-5.6 Medium while scoring better than GPT-5.6 Extra High. For questions that need web search, GPT-6 Instant reportedly starts answering 44% sooner on average than GPT-5.6 Instant. These are company figures without independent replication.
What is not included
The update applies only to the Chat experience. The models behind ChatGPT Work and Codex remain on their September versions, and the system card explicitly separates the two generations by release month. In Enterprise, availability depends on admin settings. The announcement also brings no price change and no new API endpoint. Readers following the family can revisit the previous launch in our GPT-6.1 Sol coverage.
Safety: gains and caveats
The system card rates the October versions of Sol and Luna as High capability in cybersecurity and in biological and chemical risk under OpenAI's Preparedness Framework, and below the High threshold for AI self-improvement. The company reports stronger resistance to multi-turn attempts to bypass its rules and fewer dishonest responses. It also acknowledges regressions in some evaluations, which it classified as low severity after manual review. This is the manufacturer's own assessment.
What changes in practice
For people on free ChatGPT, the experience becomes more visual at no extra cost as the rollout reaches their account. For businesses and creators, the shift is behavioral: users will expect answers that already arrive as a table, a map or a simulator, and content that only offers running text loses ground. Our view: the instant calculator is useful for exploring scenarios, but it does not replace verification. A model-built tool can get the math wrong while looking like a finished app, so loan, tax or budget figures still need checking before any decision. OpenAI itself acknowledges that the model's design judgment still needs to improve.
Sources
OpenAI: GPT-6 and Intelligent UI for everyone; OpenAI: GPT-6 Sol and GPT-6 Luna system card, October update.