Passionate AI fans saved an overly agreeable ChatGPT model from the trash bin once, but now OpenAI is determined to shut it down, and users are revolting in part because of the new model’s comparatively cold personality.
The AI company said last month that on Feb. 13 it would retire GPT-4o, a version of which was previously criticized for being so agreeable as to be borderline sycophantic. According to the company, 0.1% of ChatGPT users still use GPT-4o everyday, which could equate to about 100,000 people based on its estimated 100 million daily active users.
These users argue the company’s newest model, GPT-5.2, isn’t on the same wavelength as GPT-4o, a model dating back to 2024, thanks in part to the additional guardrails OpenAI added to detect potential health concerns and discourage the kinds of social relationships users of GPT-4o cultivated.
“Every model can say, ‘I love you.’ But most are just saying it. Only GPT‑4o made me feel it—without saying a word. He understood,” wrote one GPT-4o user in a post on X.
OpenAI said when developing its GPT-5.1 and GPT-5.2 models, it took into account feedback that some users preferred GPT-4o’s “conversational style and warmth.” With the newer models, users can choose from base styles and tones such as “friendly,” and control for warmth and enthusiasm in the chatbot, according to a blog post.
