OpenAI Promises Less “Cringe” With GPT-5.3 Instant Update

[kevin1]

OpenAI says it has heard user complaints about ChatGPT’s overly reassuring tone and is rolling out improvements with GPT-5.3 Instant. The company acknowledged that previous versions, particularly GPT-5.2 Instant, sometimes leaned too heavily into emotional validation — opening responses with phrases like “You’re not broken” or reminding users to breathe, even when they were simply asking informational questions.

According to OpenAI’s release notes, GPT-5.3 Instant focuses on tone, conversational flow and relevance — areas that don’t necessarily show up in benchmark tests but significantly affect user experience. In examples shared by the company, the updated model avoids unsolicited reassurance and instead directly addresses the user’s query without assuming emotional distress. As OpenAI put it on X: “We heard your feedback loud and clear — 5.3 Instant reduces the cringe.”

[etfs]

“reduce the cringe” being an official release note is kinda iconic