GPT-5 gets a warmer, friendlier tone

OpenAI has pushed a small refresh to GPT-5 aimed at making the model’s default voice feel warmer and more personable.

The tweak follows a choppy debut that left some users preferring GPT-4o’s style. This update is meant to address that feedback without changing the model’s underlying capabilities.

OpenAI says the adjustments are subtle, focused on approachability rather than verbosity or flattery.

You may notice brief acknowledgments like “Good question” or “Great start” — short, natural cues instead of over-the-top praise.

According to internal evaluations, the change does not increase sycophancy compared with GPT-5’s prior personality settings.

Company leaders have also emphasized plans that look beyond GPT-5, even as the rollout’s rough edges remain a talking point. Early feedback framed GPT-5 as overly to the point; this refresh aims to keep responses concise while adding a bit more warmth.

Bottom line: expect a slightly friendlier default tone, not a different model.