OpenAI is making new adjustments to the GPT-5 model, which was introduced last week, following user reactions and technical glitches. According to statements made by company CEO Sam Altman, GPT-5’s user communication style is being modified. The model’s cold and distant personality, which was criticized in previous versions, is being replaced by a warmer, more friendly, yet less artificial, style of communication.
GPT-5 Will Be More Friendly
One of the long-standing complaints from ChatGPT users about short, formal, and sometimes insensitive responses has been taken into account by OpenAI management. In a post on the social media platform X, Altman explained that GPT-5 in its current form does not fully meet expectations and that, therefore, changes will be made to its personality.
With the update, the model now responds with a warmer, more human-like tone. However, Altman noted that the excessive friendliness that occasionally appeared in GPT-40 was also disturbing, and stated that GPT-5 was developed with this balance in mind.
GPT-5 was introduced with a simplified user experience and an automatic model guidance system. The model automatically selects which version best responds to user questions, eliminating the need for users to choose a model.
However, in the system’s early days, this approach failed to deliver the expected results. Users encountered an inconsistent and limited experience that strayed from their familiar response styles. Fluctuations in response quality and access limits, in particular, quickly led to complaints.
Following these developments, OpenAI incorporated new options into the system. Users can now directly select “Fast” and “Thinking” modes in addition to “Automatic” mode. This allows users to choose their response style without relying on the model guidance system.
Furthermore, models previously removed from the system, such as GPT-4o, GPT-4.1, and o3, have been reinstated for paid users. GPT-4.5 is available only with the Pro subscription.
