OpenAI Releases GPT-5.3 Instant to Reduce 'Cringe' Responses in ChatGPT

Summary

OpenAI has released GPT-5.3 Instant, an update specifically designed to make ChatGPT's conversational tone more natural and less overly dramatic. The update addresses widespread user feedback about the model's tendency to produce formulaic, eye-roll-inducing responses that have become the subject of internet memes. According to OpenAI, the previous GPT-5.2 Instant version often came across as "overbearing" and made "unwarranted assumptions about user intent or emotions," sometimes including phrases like "Stop. Take a breath" that users found patronizing.

The new model focuses on delivering more focused, natural conversation while cutting back on unnecessary proclamations and overly declarative phrasing. OpenAI states that GPT-5.3 Instant "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." The update also significantly reduces unnecessary refusals and tones down overly defensive or moralizing preambles before answering questions.

Balancing natural yet personal responses has been an ongoing challenge for OpenAI, with multiple updates attempting to refine ChatGPT's conversational style. The company has also improved how the model integrates data from web search results into its responses. This release represents OpenAI's continued effort to make AI interactions feel less artificial and more genuinely helpful, addressing one of the most common criticisms of large language model outputs.

• The release reflects OpenAI's ongoing challenge to balance natural conversation with helpful, personalized AI responses