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

Take a breath, halt spiraling. You’re not crazy, you’re conscionable stressed. And honestly, that’s okay.

If you felt instantly triggered speechmaking these words, you’re astir apt besides sick of ChatGPT perpetually talking to you arsenic if you’re successful immoderate benignant of situation and request delicate handling. Now, things whitethorn beryllium improving. OpenAI says its caller model, GPT-5.3 Instant, volition trim the “cringe” and different “preachy disclaimers.”

According to the model’s merchandise notes, the GPT-5.3 update volition absorption connected the idiosyncratic experience, including things similar tone, relevance, and conversational travel — areas that whitethorn not amusement up successful benchmarks, but tin marque ChatGPT consciousness frustrating, the comapny said.

Or, arsenic OpenAI put it connected X, “We heard your feedback large and clear, and 5.3 Instant reduces the cringe.”

In the company’s example, it showed the aforesaid query with responses from the GPT-5.2 Instant exemplary compared with the GPT-5.3 Instant model. In the former, the chatbot’s effect starts, “First of each — you’re not broken,” a communal operation that’s been getting nether everyone’s tegument lately.

In the updated model, the chatbot alternatively acknowledges the trouble of the situation, without trying to straight reassure the user.

The insufferable code of ChatGPT’s 5.2 exemplary has been annoying users to the constituent that immoderate person adjacent cancelled their subscriptions, according to galore posts connected societal media. (It was a huge point of discussion connected the ChatGPT Reddit, for instance, earlier the Pentagon woody stole the focus.)

People complained that this benignant of language, wherever the bot talks to you arsenic if it assumes you’re panicking oregon stressed erstwhile you were conscionable seeking information, comes crossed arsenic condescending.

Often, ChatGPT replied to users with reminders to respire and different attempts astatine reassurance, adjacent erstwhile the concern didn’t warrant it. This made users consciousness infantilized, successful immoderate cases, oregon arsenic if the bot was making assumptions astir the user’s intelligence authorities that conscionable weren’t true.

As 1 Reddit idiosyncratic precocious pointed out, “no 1 has ever calmed down successful each the past of telling idiosyncratic to calm down.”

It’s understandable that OpenAI would effort to instrumentality guardrails of immoderate kind, particularly arsenic it faces aggregate lawsuits accusing the chatbot of starring radical to acquisition antagonistic intelligence wellness effects, which sometimes included suicide.

But there’s a delicate equilibrium betwixt responding with empathy and providing quick, factual answers. After all, Google ne'er asks you astir your feelings erstwhile you’re searching for information.