GPT-5 is becoming more “friendly” 



Many users complained that GPT-5 felt stiff, impersonal, and ineffectively conversational once it was released. This was in sharp contrast to GPT-4o’s friendlier, “yes-man” demeanor. Some even compared the death of GPT-4o to losing a friend, describing the experience as emotionally taxing.

The Wall Street Journal brought attention to a wider dissatisfaction: customers said that GPT-5 was too robotic, had a less nuanced tone, and even had a weekly question cap of 200, which caused satisfaction to decline.

With care to avoid reintroducing the over-familiarity of GPT‑4o, OpenAI is actively improving GPT‑5 to better balance warmth and precision. The company is acknowledging that model personality is important to consumers and that there is no one-size-fits-all solution by reintroducing historic models and introducing new conversational modes.