OpenAI restores GPT-4o at the explicit request of users, after GPT-5 failed to make an impression at launch.
The long-awaited launch of GPT-5 last Thursday immediately caused a rush on ChatGPT. However, users who started working with the new model were left disappointed, partly because the option to switch back to older versions suddenly disappeared. At the request of users, OpenAI confirms that it is temporarily bringing back GPT-4o.
Sterile Chatbot
GPT-5 is supposed to be smarter and more efficient than previous GPT versions. OpenAI CEO Sam Altman compares the new model to a PhD student. ChatGPT users were not enthusiastic about the “sterile” and “dry” style of responses. Altman personally went to Reddit forums to ask for feedback.
OpenAI admits via social media that the rollout of GPT-5 did not go as planned. ChatGPT immediately faced a technical glitch after the launch, leading to an increased error margin in GPT-5’s responses. OpenAI says it is further tweaking the model to make the responses “warmer”.
No procrastination
Even with a brief return, the days of GPT-4o and other “legacy” models are numbered. GPT-5 will take over as soon as possible. OpenAI will monitor the use of older models in the coming days to determine how long they should be kept alive. The option to switch back to GPT-4o does not yet seem to be enabled on all accounts and will only be available for paying users.
GPT-5 has the capacity to switch between models based on a prompt. For simple questions, it uses a smaller and more efficient model that requires fewer resources, while more complex issues that require reasoning are passed on to the most comprehensive model.