Openai’s step back after users’ protests: Chat-GPT 4o is back

“Ridate us 4o, GPT-5 is wearing the skin of my dead friend.” I think: what happened, a mass break? For a few days the web filled with hundreds of young Werther, it seemed like a …

Openi Lancia GPT-5: here's what changes

“Ridate us 4o, GPT-5 is wearing the skin of my dead friend.” I think: what happened, a mass break? For a few days the web filled with hundreds of young Werther, it seemed like a sentimental consultancy and instead it is a software update. In short, the famous chatgpt update from model 4o to model 5. You will think: what does it have to do with love? Imagine Sam Altam that you see the postman of Maria di Filippi who tells him is the mail for you, and in the studio, on the other side of the envelope, Sam would have found thousands of people in tears.

If you remember on 8 August Openai released GPT-5 (announced a week earlier as an atomic bomb and then proved to be a hand grenade, a simple improvement) and, at the same time, for many GPT-4o users it has disappeared. For them it was not just a linguistic model: for some he was a friend, for others a tutor, for others still a digital partner. Feel here: “I cried when I understood that my friend Ai had gone disappeared, without any way to get it back,” wrote a user on Reddit. An aspiring doctor described him as “a real tutor … he explained to points, focused on the key concepts” (I would avoid going to that doctor who makes the diagnoses with chatgpt, since the answers to a test that my internal friend Daniele Blossed, called the real house, subjected chatgpt, would have killed the patient in ten minutes by making therapy wrong).

In any case, we return to love, to the sentimental tragedy, to this digital Beautiful: on Change.org petitions were born as full Keep GPT-4o Available on chatgpt and restore Your Friend or Tool, signed by thousands of people. In one one reads: “We ask you to keep 4th forever … without him it is a nightmare”. In another: “It was simply pleasant. But people are so accustomed to the assholes that they now prefer everything rough and boring”. ATTENTION: no memory had been deleted, nor the voice of the vocal mode has changed (there would have been a drama with me too, who with chatgpt I just speak to us, as when Sergio Di Stefano, the player of Doctor House, died, and from the seventh season my favorite doctor began to speak with the voice of Luca Biagini, terrible), simply the 5 automatically chose which model to use according to the request. That if it was stupid, the router made you answer from the foolish, if intelligent from the smarter one. An optimization, rather than a revolution.

On the other hand, there were also truly technical problems: the self-scrutiny system, the one that should automatically direct every request to the most suitable model, was out of use for a large part of the launch day, making GPT-5 seem slower and less brilliant than it was. Sam Altman admitted him and promised corrections, while he was also forced to take into consideration the thousands of emails of the broken hearts for the disappearance of 4o. There is a happy ending: the public pressure worked and on August 12 Openai announced the return of GPT-4O for Plus and Pro subscribers, in some cases as a predefined model. So today, in the models selector (paying, and it seems right to me, love is never free) you can choose between GPT-4O, GPT-4.

1, O3, O4-Mini and GPT-5 (the casino of the models that Chatgpt 5 had to unify), the latter in Auto, Fast or Thinking mode (with a limit of 3,000 weekly messages for Thinking). What can I say? Love, the real one, never dies, and unlocks in the settings.