OpenAI is bringing back its older GPT-4o model to ChatGPT after widespread user dissatisfaction with the recent GPT-5 rollout. CEO Sam Altman admitted that the launch was "bumpy" and acknowledged the need for improvement. According to Altman, the GPT-5 rollout suffered from unexpected technical disruptions, delays, and a user experience that many described as "horrible".
The technical issues with GPT-5's real-time router, designed to choose the best model for a given prompt, were a major contributor to the problems. The router was malfunctioning, leading to subpar performance. Users also complained about GPT-5's responses being short, insufficient, and lacking personality compared to GPT-4o. Furthermore, the removal of older models like GPT-4o meant users lost access to specific features and personalities they had grown accustomed to.
In response to the backlash, OpenAI is reinstating GPT-4o for ChatGPT Plus users, allowing them to choose between the older model and GPT-5. The company is also doubling rate limits for GPT-5 users to alleviate concerns about usage limits. Additionally, OpenAI is working on making the model-switching process more transparent and efficient.
By bringing back GPT-4o and addressing the technical issues, OpenAI aims to provide a better user experience and improve the overall performance of its AI models. The company's willingness to listen to user feedback and make adjustments demonstrates its commitment to delivering high-quality AI solutions.
