On Friday evening, OpenAI announced they are enhancing their latest model with a more "warm and friendly" approach.

The company recently launched the highly anticipated GPT-5. During this process, CEO Sam Altman admitted the rollout has been "more challenging than expected," with some users expressing preference for the earlier GPT-4o model.

OpenAI is now addressing these concerns through the update, implementing changes described as "subtle" but making GPT-5 "more approachable."

"You'll notice small, genuine details like 'good question' or 'great start' rather than flattery," the company stated in a social media post. "Internal testing shows flattery levels haven't increased compared to previous GPT-5 iterations."

During an executive dinner with journalists this week, OpenAI leadership attempted to shift focus toward projects beyond GPT-5, yet as Max Zeff reported, the rocky launch remained an unspoken issue. Regarding model friendliness, VP Nick Turley noted GPT-5 was "very straightforward," but the newly announced update will make it feel warmer now.