Frustration with GPT-4: Users Complain About Declining Performance of OpenAI's ChatGPT-4

Concerns have been rising among users about a perceived decline in performance with OpenAI's ChatGPT-4. Many have described the model as "lazier" and "dumber," particularly in comparison to its earlier reasoning capabilities. Industry experts believe this shift may be linked to a significant redesign, as OpenAI has introduced a mixture of experts (MoE) model aimed at enhancing output quality, reducing costs, and improving response times.

One user involved in website development expressed frustration, stating, "The current GPT-4 is incredibly disappointing. It's like driving a Ferrari for a month only to find it has turned into a rundown pickup truck. I’m not sure if I'm willing to continue paying for this." Another user noted, “GPT-4 keeps repeating code and other information. It feels like it has suffered a cognitive decline. If you weren’t familiar with its previous capabilities, you might not notice the difference, but for those who relied on GPT-4, its performance drop is evident.”

Since its launch at the end of last year, ChatGPT quickly gained traction, initially based on GPT-3 and GPT-3.5. In March, GPT-4 debuted, becoming the preferred model for developers and tech professionals. Renowned for its multimodal capabilities, GPT-4 can handle both images and text inputs. According to Sharon Zhou, CEO of Lamini, while GPT-4 may perform slower than before, it excels in accuracy. User feedback indicates that the current version of GPT-4 is faster yet less accurate compared to its predecessors, which, despite being slower and more costly, provided more precise information.