OpenAI Reinstates GPT-4o After User Backlash Over GPT-5
OpenAI recently launched its highly anticipated GPT-5 model, touting it as the smartest and fastest AI yet. However, the rollout was met with unexpected resistance from users who preferred the previous GPT-4o model. Many complained about GPT-5’s writing style, calling it sterile and less engaging, with some even expressing emotional attachment to GPT-4o as a trusted companion.
Initially, OpenAI removed the option to switch between models, forcing users to rely solely on GPT-5. The new model uses a real-time routing system, shifting between a lightweight version for simple queries and a more advanced one for complex tasks. But early performance issues led to criticism, including from OpenAI CEO Sam Altman, who admitted GPT-5 seemed way dumber at first.
Altman engaged with users on Reddit, asking if they felt GPT-5’s writing quality had improved. The response was overwhelmingly negative, with many describing the answers as brief, dry, and lacking the warmth of GPT-4o. Following the backlash, OpenAI announced that Plus subscribers would regain access to GPT-4o as an alternative.
Altman acknowledged that the company underestimated how much users valued certain aspects of GPT-4o, even if GPT-5 performed better in most technical benchmarks. While the return of GPT-4o was celebrated, Altman hinted that older models may not remain available forever. OpenAI will monitor usage to decide how long to keep legacy options accessible.
For now, OpenAI is working on refining GPT-5 to make its responses feel warmer and more natural. But for those who formed a personal connection with GPT-4o, this could mark the beginning of a slow farewell as the company continues pushing forward with newer iterations. The situation highlights the delicate balance between advancing AI capabilities and maintaining the human-like qualities that users grow attached to.
