On February 13, OpenAI will retire several models from the ChatGPT model picker, including GPT-4o, due to low daily usage of only 0.1% among its users. This decision follows a previous backlash when GPT-4o was removed last August after the launch of GPT-5, prompting OpenAI to reinstate it amid discussions on user attachment to different model personalities.
OpenAI's CEO, Sam Altman, had previously assured users that any future deprecations would come with adequate notice. Currently, users can select from seven distinct personalities and set custom instructions to personalize their experience. Additionally, a “Show additional models” toggle has been introduced for paid users, allowing access to models such as o3 and GPT-5 Thinking mini.
Despite the retirement of these models from ChatGPT, OpenAI confirmed that there will be no immediate changes to the API, ensuring that developers can still access these models. The company is also working on updates to enhance GPT-5’s personality to make it feel warmer while avoiding previous criticisms associated with GPT-4o.