‘More Accurate, Less Cringe’: OpenAI Rolls Out GPT-5.3 Instant in ChatGPT

OpenAI on Tuesday revealed the rollout of GPT-5.3 Immediate, an upgrade to ChatGPT’s default design focused on making discussions feel less uncomfortable and preachy, and “more straight” useful.

In an item post, OpenAI stated the brand-new variation minimizes excessively careful rejections, trims unneeded disclaimers, and provides more precise responses. The modifications show user grievances that earlier variations might sound stiff or self-important in daily interactions.

” More precise, less flinch,” OpenAI composed on X. “We heard your feedback loud and clear.”

Instead of presenting brand-new abilities, the upgrade targets regular interactions. OpenAI stated earlier variations in some cases decreased concerns they might securely address, or interrupted reactions with prolonged descriptions about security limitations.

” GPT‑5.2 Immediate ultimately responds to the concern, however in an effort to discuss its security limits, leads with a prolonged preamble about what it can not aid with,” OpenAI composed. “GPT‑5.3 Immediate, on the other hand, solves into the reaction.”

OpenAI reported enhancements in accurate dependability along with the tone modifications, declaring that internal assessments revealed hallucination rates stopped by almost 30%.

” On the higher-stakes examination, GPT‑5.3 Immediate minimizes hallucination rates by 26.8% when utilizing the web and 19.7% when relying just on its internal understanding, compared to previous designs,” OpenAI stated. “On the user-feedback examination, hallucinations reduce by 22.5% with web usage and 9.6% without web gain access to.”

OpenAI did not discuss what it specifies as “cringe,” however kept in mind that the brand-new design consists of more powerful composing capabilities, comparing GPT-5.2 and 5.3’s capability to compose poetry.

” 5.4 quicker than you believe,” the business stated in a different post, which drew speedy mockery from users on X, recommending that the tease was because of current reaction versus the company for its handle the Pentagon.

GPT-5.3 Immediate changes the default ChatGPT design beginning today, the business stated, while GPT-5.2 Immediate remains available under tradition alternatives for paid customers throughout a shift duration ending in early June.

The upgrade drew blended responses on social networks. Some users applauded the concentrate on more direct reactions without unneeded disclaimers, while others argued the genuine “cringe” at play was accepting an agreement with the U.S. Department of Defense when competing Anthropic decreased due to security issues.

Others grumbled that GPT-5.3 would never ever match the sense of intimacy lots of connected with the now-depreciated GPT-4o, and required the popular design’s return.

Last summer season, OpenAI dealt with a rise in reaction after the business suddenly changed the popular GPT-4o with GPT-5, triggering grievances that the brand-new design felt cooler and less helpful. Users flooded online forums with criticism, and some threatened to cancel memberships, leading OpenAI to bring back GPT-4o for paid users.

In January, OpenAI revealed that GPT-4o and its variations would be formally retired since February 13.

” I believe we have actually discovered a lesson about what it indicates to update an item for numerous countless individuals in one day,” OpenAI CEO Sam Altman stated at the time, calling the turnaround a wake-up call and “a lesson in updating an item utilized by numerous countless individuals” simultaneously.

OpenAI did not right away react to an ask for remark by Decrypt.