You Can Pick Chatgpt's Older Ai Models Again

ChatGPT will now let you to take betwixt respective GPT-5 variants and previous OpenAI models. In a post connected X, OpenAI CEO Sam Altman has revealed that nan chatbot's exemplary picker now comes pinch 3 GPT-5 choices: Auto, Fast and Thinking. He said astir "will want Auto," which is astir apt nan modular type that's already a reasoning model. But users will beryllium capable to take nan fast-responding type aliases nan Thinking type that implies it delivers longer, much broad answers if they want to. GPT-4o is now besides backmost successful nan exemplary picker for each paying users by default.

OpenAI removed GPT-4o pinch nan motorboat of GPT-5: The institution importantly simplified ChatGPT because GPT-5 was expected to connection a unified acquisition for users. However, its determination was met by criticism from group who preferred GPT-4o's personality better, particularly since it didn't springiness them announcement that it was deprecating older models. In his post, Altman promised users that if nan institution ever does deprecate GPT-4o, OpenAI "will springiness plentifulness of notice." In summation to nan GPT-5 and GPT-4o models, users tin toggle to "Show further models" successful ChatGPT web settings if they want to entree o3, 4.1, and GPT-5 Thinking mini. GPT-4.5 is only disposable to subscribers paying for nan $200-a-month Pro tier, though, because it "costs a batch of GPUs."

Altman said that GPT-5 Thinking now has complaint limits of 3,000 messages a week, aft which users will only beryllium capable to usage GPT-5 Thinking mini. He has besides revealed that OpenAI is moving to update GPT-5's characteristic to make it warmer, but "not arsenic annoying (to astir users) arsenic GPT-4o."