OpenAI says the new GPT-5.3 Instant model improves ChatGPT's conversational flow, reduces hallucinations, and cuts down on preachy disclaimers. (Express Photo)ChatGPT is useful for a wide range of tasks, but recently, many users have complained that conversations with the AI chatbot sometimes felt like therapy sessions, as if they were in some kind of crisis.Now, OpenAI says it is rolling out an update to GPT-5.3 Instant, ChatGPT’s most widely used model, which will make it less “cringe” and cut down on “preachy disclaimers.”In a blog post, OpenAI said that the GPT-5.3 update is aimed at improving accuracy and reducing “unnecessary dead ends, caveats, and overly declarative phrasing that can interrupt the flow of conversation.”The company also shared a post on X (formerly Twitter) comparing responses from GPT-5.2 Instant with those generated by the updated GPT-5.3 Instant model.GPT-5.3 Instant in ChatGPT is now rolling out to everyone.More accurate, less cringe.https://t.co/oJpXsp9TBc— OpenAI (@OpenAI) March 3, 2026Also Read | ChatGPT uninstalls spike 295% after OpenAI’s DoD deal; Claude climbs US App Store chartsAlso, the updated model will now focus more on things like tone, relevance, and conversational flow, things that do not show up in benchmarks but may make the AI chatbot frustrating for some. For example, OpenAI says the new model will cut back on phrases like “Stop. Take a breath.”GPT-5.3 Instant also reportedly delivers more factual responses than previous models, with OpenAI saying that it reduces hallucination rates by 26.8% when using the web and 19.7% when relying on its internal knowledge.OpenAI goes on to say that GPT-5.3 Instant also reduces unnecessary refusals, dials down on “overly defensive or moralizing preambles”, and answers questions without any unnecessary caveats. The updated model also improves on the “quality of answers” by balancing what it finds on the internet with what it knows about the topic. © IE Online Media Services Pvt Ltd