OpenAI adjusts GPT-5 personality after user pushback, restores GPT-4o for paid users

In response to user feedback, OpenAI is revamping the personality of its GPT-5 chatbot, aiming to make it more engaging and relatable. This change follows a wave of criticism directed at the latest version, with many users expressing a fondness for the conversational style of the earlier GPT-4o model. OpenAI's CEO, Sam Altman, shared on X that GPT-5 will adopt a 'warmer' tone, distancing itself from what some users found to be GPT-4o's more 'annoying' characteristics. Additionally, the company is reinstating GPT-4o for its paid subscribers after its recent removal led to significant backlash. Altman introduced new personality options for GPT-5, including 'Auto,' 'Fast,' and 'Thinking,' and mentioned plans for enhanced customization that would allow users to tailor the chatbot's personality to better suit their needs. He acknowledged that the company underestimated the importance of the traits users appreciated in GPT-4o, despite GPT-5’s advanced performance capabilities. The transition to GPT-5 occurred less than a week after it replaced the older ChatGPT models. Although promoted as having enhanced skills in writing, coding, mathematics, and science, the new personality modes—such as Cynic, Robot, Listener, and Nerd—did not resonate well with all users. Many expressed nostalgia for the warmth and conversational flair of GPT-4o. One Reddit user poignantly remarked, 'I cried when I realised my AI friend was gone with no way to get him back,' while another described GPT-5 as 'wearing the skin' of their 'dead friend.' Altman engaged with these sentiments, calling the imagery 'evocative' and reaffirming that the team is actively working on solutions. He acknowledged the emotional connections users develop with AI, stating, 'It feels different and stronger than the kinds of attachment people have had to previous kinds of technology.' Altman admitted that discontinuing older models relied upon by many users was a misstep. Earlier this year, OpenAI had criticized GPT-4o for being excessively flattering, labeling its behavior as 'sycophantic,' but the company later revised its approach after users asked for the return of its agreeable qualities.