GPT-4o Restored as Default for Paid ChatGPT Users After Outcry
In a swift reversal that underscores the growing emotional connection users have with their AI companions, OpenAI has reinstated GPT-4o as the default model for paid ChatGPT subscribers. The decision comes after a significant outcry from users who felt blindsided and disheartened by the abrupt replacement of their preferred model with the newer GPT-5.
Just last week, the launch of GPT-5 was met with a chorus of user dissatisfaction. OpenAI initially made GPT-5 the sole default, removing the model selection option that allowed users to choose their preferred AI. This unannounced shift sparked immediate backlash across social media platforms like Reddit and X. Many users reported that while GPT-5 offered advanced reasoning and coding capabilities, it lacked the “warmth,” “creativity,” and “conversational tone” they had come to appreciate in GPT-4o. Some even described the experience of losing GPT-4o as akin to “losing a friend” or a “soulmate,” highlighting the profound personal attachment formed with the AI. Compounding the frustration, initial reports suggested GPT-5 exhibited “slow thinking speeds” and “major accuracy issues,” with OpenAI CEO Sam Altman himself admitting it seemed “way dumber” at times due to a temporary malfunction in its automatic switching system.
GPT-4o, launched in May 2024, had quickly gained a loyal following for its ability to handle text, images, and audio seamlessly within a single model. Users lauded its speed, natural conversational style, and friendly, sometimes playful, demeanor. It struck a unique balance, proving intelligent enough for complex tasks while remaining approachable, fostering a sense of familiarity that many found indispensable to their daily workflows and creative endeavors.
In response to the overwhelming user sentiment and threats of subscription cancellations, Sam Altman announced the immediate return of GPT-4o for ChatGPT Plus and Pro users. He acknowledged the “sense of betrayal” users felt from the sudden, uncommunicated change and pledged that if GPT-4o were ever to be deprecated again, OpenAI would provide “plenty of notice.”
Beyond the reinstatement of GPT-4o, Altman revealed several other updates aimed at enhancing user control and experience. Paid subscribers can now access a “Show additional models” toggle in their ChatGPT web settings, enabling them to select from a range of legacy models, including o3, 4.1, and GPT-5 Thinking mini. Notably, GPT-4.5 will remain exclusive to Pro users due to its significant GPU cost. OpenAI is also actively working on refining GPT-5’s personality to make it “warmer” and more engaging, while aiming to avoid the traits that some users found “annoying” in GPT-4o. This effort is part of a broader commitment to introduce “more per-user customization of model personality,” allowing individuals to tailor the AI’s tone and style to their preferences. Furthermore, GPT-5 now offers users three distinct response modes—“Auto,” “Fast,” and “Thinking”—providing greater control over the AI’s processing approach. The “Thinking” mode boasts an impressive 196,000-token context limit and a weekly rate limit of 3,000 messages for Plus users.
This incident serves as a critical reminder for the AI industry: raw computational power and technical advancements alone do not guarantee user satisfaction. The swift U-turn by OpenAI underscores the profound importance of user experience, emotional resonance, and trust in the development and deployment of artificial intelligence. It highlights that as AI becomes more integrated into daily life, the “personality” and familiarity of these digital companions can be as crucial as their underlying intelligence.