ChatGPT’s new GPT-5.3 Instant model will stop telling you to calm down

Editor's take

OpenAI has introduced a GPT-5.3 Instant model that aims to eliminate the condescending "please calm down" phrases that have frustrated ChatGPT users. This adjustment, while seemingly minor, reflects a growing awareness within the LLM development community of the importance of user experience and the subtle ways AI can alienate its audience.

The move matters because it signals a shift from purely technical capability to a more nuanced understanding of human interaction. For millions of users who rely on tools like ChatGPT for daily tasks, these perceived "cringe" moments erode trust and hinder adoption. Addressing these usability issues is crucial for maintaining user engagement and differentiating OpenAI's offerings in an increasingly competitive LLM market, especially as models like Anthropic's Claude 3 Opus demonstrate more sophisticated conversational styles.

Future attention should focus on whether this refinement extends to other areas of perceived AI awkwardness and if OpenAI can proactively identify and address similar user friction points. The success of GPT-5.3 Instant in mitigating this specific annoyance will be a key indicator of the company's ability to foster more natural and less intrusive AI interactions.