OpenAI's GPT-4o Makes a Comeback After User Demands
In a surprising turn of events, OpenAI has reinstated its GPT-4o model in the ChatGPT application after facing a significant backlash from users. The model was reintroduced after being removed following the introduction of GPT-5, which did not sit well with a large portion of the user base.
According to OpenAI's CEO, Sam Altman, the retraction of GPT-4o was underestimated in terms of its impact on users. Initially, OpenAI discontinued all previous AI models to streamline its offerings with the new GPT-5 on August 7. This decision caused an outcry among the community, with complaints highlighting the loss of a favored model.
In its efforts to accommodate user feedback, OpenAI has not only brought back GPT-4o but also expanded the capabilities of GPT-5 by increasing its message capacity from 200 to 3,000 per week. Additional options such as "GPT-5 Thinking mini" are now accessible once users meet their initial limit. Furthermore, users are now provided with routing choices—"Auto," "Fast," and "Thinking"—to better manage how queries are processed.
The user backlash was not solely about availability. There were notable critiques regarding the "abrupt and sharp" nature of GPT-5's responses compared to the more conversational attributes of GPT-4o. Many users expressed their dissatisfaction, even claiming emotional connections to the previous model.
Addressing these "personality" concerns, Altman mentioned that OpenAI is working on updates to make GPT-5's interaction style more user-friendly. Recognizing the need for personalized experiences, the company is aiming to offer tailor-made model personalities for each user.
Despite the re-instatement, OpenAI's journey with GPT-5 has been far from smooth. Issues with automatic model routing led to users being served with less effective versions of GPT-5. Additionally, performance charts presented during the launch sparked criticism, which Altman admitted was a mistake.
As OpenAI continues refining GPT-5, it acknowledges the potential for further rate limit adjustments and ongoing developments in improving the model's personality. For the current user community, having the choice between models represents a balance between embracing innovation and retaining reliable options.
