After the launch of the GPT-5 Openai, she promised to simplify the work with ChatGPT, using one “universal” model that would choose how best to respond to user requests. But users did not appreciate this approach. Now everything has returned to their own circles, but there is a nuance.
The head of OpenAi recently said that users can now choose between Auto, Fast and Thinking modes. Auto works as a GPT-5 router, but users can directly choose fast or “deliberate” answers. Paid users also gain access to old models, including GPT-4O, GPT-4.1 and O3. The GPT-4O is now available by default, while other models can be added through the settings. Difficult? Definitely. Even on the network, they noticed it.
Despite criticism of the public and the media, the head of the company noted that in the future Openai wants to make the personality of the models “more flexible” for each person.
Experts note that Openai still has a lot of work towards the “universality” of models.
