GPT-5 gets a friendlier tone after a rocky rollout

OpenAI is rolling out a refinement to GPT-5 designed to make the model feel more personable and approachable in everyday use.

The tweak follows a choppy debut for GPT-5, during which some users said they preferred the previous flagship, GPT-4o. By softening the default style without changing core capabilities, the company aims to address those early complaints.

According to the announcement, the adjustments are intentionally subtle. You may see small acknowledgments like “Good question” or “Great start” that aim to encourage progress without tipping into flattery. Internal evaluations indicate no increase in sycophancy compared with the prior GPT-5 personality.

Executives have recently tried to shift attention to what comes after GPT-5, but they also acknowledged feedback that the current model could feel overly terse. Leadership now says this refresh should preserve the model’s directness while adding a warmer touch.

For users, the takeaway is a gentler interaction style that keeps answers concise and on-topic. If successful, the change should improve day‑to‑day usability—especially for learning, brainstorming, and step‑by‑step tasks—without sacrificing accuracy or reasoning.