Paid subscribers can now choose from Instant to Pro Extended, giving them granular control over how hard the AI thinks before answering.
OpenAI rolled out a revamped model picker for ChatGPT on June 10, giving paid users six distinct options that balance response speed against reasoning depth. The update replaces the simpler selection system that has been in place since mid-2025 and is available across web, iOS, and Android.
What the new tiers actually do
The six options are Instant, Medium, High, Extra High, Pro Standard, and Pro Extended.
Instant is built for speed. Quick factual lookups, simple rephrasing, anything where you’d rather not wait. Medium and High occupy the middle ground, offering progressively deeper reasoning without dramatically longer wait times.
Extra High pushes the model harder, suitable for complex analysis or multi-step problem solving. Pro Standard and Pro Extended are reserved for Pro subscribers and represent the heaviest computational lift available, delivering the highest reasoning capabilities at the cost of longer response times.
OpenAI also added a “Show additional models” toggle in web settings. This lets paid users access legacy options like o3, o4-mini, and 4.1 directly from the picker.
The evolution from Auto/Fast/Thinking
This isn’t OpenAI’s first attempt at simplifying model selection. Starting in mid-2025, the company introduced Auto, Fast, and Thinking modes as a way to abstract away the growing complexity of the GPT-5 family.
The availability is limited to paid tiers, specifically Plus and Pro subscribers. Free users don’t get access to the new picker. The Pro-exclusive tiers, Pro Standard and Pro Extended, add an additional incentive layer for users willing to pay for the highest subscription level.
