Particle.news

Download on the App Store

OpenAI Rolls Out Warmer GPT-5 Personality, Adds Selectable Personas

OpenAI targets complaints about a terse tone by tuning GPT-5's default behavior in its web and mobile apps.

Overview

  • ChatGPT’s new friendlier tone is rolling out for the standard GPT-5 variant in web and mobile, while the higher‑capability Thinking model remains unchanged.
  • OpenAI is introducing four preset text personas—Zyniker, Roboter, Zuhörer and Nerd—to let users pick different conversational styles.
  • CEO Sam Altman says the goal is a warmer feel without GPT-4o’s perceived fussiness and signals a push toward stronger per‑user personality controls.
  • OpenAI describes the tweaks as subtle and aimed at greater accessibility, citing internal tests that show no increase in submissiveness compared with the prior GPT-5 personality.
  • The shift follows user frustration over GPT-5’s brief, impersonal replies and reinstatement of GPT-4o, with reporting also noting research that warmer styles can raise error risks.