ChatGPT-5 will allow you to choose a communication model
OpenAI will return the choice of thinking model to ChatGPT-5, although it initially planned to replace it with an automatic algorithm. Sam Altman promises the following options: Auto (default), Fast (fast, but less in-depth), Thinking (slower, with deeper analysis), and Legacy Mode - GPT-4o and older models are now available to all paid users. The update is likely to be rolled out gradually, and some users have already seen the changes.
Why did this happen?
The model selection algorithm in GPT-5 did not meet expectations, although it was a reasonable approach to simplifying the interface.
Users (especially specialists) expressed dissatisfaction: they did not like the new models, GPT-4o was more comfortable and gave a different tone of communication. OpenAI actually listened to the feedback and promised not to remove the models without warning, as was planned when GPT-5 was launched.
Besides, the personalisation options, i.e. the settings of "personalities", such as cynic, robot, sensitive, and nerd, have not disappeared.
Bottom line.
Perhaps this is the most high-profile failure of OpenAI in recent years. The company faced significant criticism of almost all the innovations announced at the last presentation. Fortunately, the company responded to the criticism quickly and quite adequately and did not try to prove that it was just users who did not understand. Well, only those who do nothing make no mistakes.
