
OpenAI has released an update to ChatGPT that it says should make its most commonly used model less “cringe” and more natural. Users should see fewer overly dramatic, jarring responses as a result.
Instead, OpenAI says the new version “delivers more accurate answers, richer and better-contextualized results when searching the web, and reduces unnecessary dead ends, caveats, and overly declarative phrasing that can interrupt the flow of conversation.”
GPT‑5.2 Instant’s tone could sometimes feel “cringe,” coming








