When OpenAI unveiled its much-hyped GPT-5 earlier this week, the company expected applause. Instead, it got a wave of frustration from a vocal section of its user base, many of whom demanded the return of the beloved GPT-4o. Some users even described GPT-4o as a “friend” and “trusted companion,” mourning its removal like the loss of a long-time ally.
The controversy began when OpenAI replaced the model selection option with a default GPT-5 experience. The new model runs on a “real-time router” system, switching between an efficiency-focused engine for simple queries and a deeper reasoning model for complex tasks. On paper, this was meant to make GPT-5 the “smartest, fastest, and most useful” version yet. But in practice, early adopters claimed it felt “colder,” “less human,” and, in some cases, “dumber.”
OpenAI CEO Sam Altman addressed the criticism head-on, admitting that the team had underestimated how much users valued GPT-4o’s unique personality and style. While insisting GPT-5 outperforms older models in most areas, Altman acknowledged that some of its “warmth” had been lost. In a gesture to calm the uproar, he announced on X that Plus subscribers could now choose between GPT-5 and GPT-4o, at least for the time being.
The return of GPT-4o was celebrated across social media, but there’s no guarantee it will remain available forever. Altman has hinted that OpenAI will “watch usage” before deciding how long to keep legacy models online. Meanwhile, the company is working on fine-tuning GPT-5 to feel more “human” while maintaining its advanced capabilities. For those who have formed a bond with GPT-4o, the clock may already be ticking.
Read more: GPT 5 Is Here! Smarter, Faster, and Free for Everyone
