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

OpenAI Updates ChatGPT with GPT-5.3 Instant for Smoother Conversations

OpenAI has launched GPT-5.3 Instant, an updated default model for ChatGPT designed to enhance conversational flow and direct helpfulness. The company stated that this new version aims to reduce unnecessary refusals and improve factual accuracy, addressing user feedback about previous models sounding stiff or overly cautious.

Key Improvements in GPT-5.3 Instant

The GPT-5.3 Instant update focuses on refining routine interactions rather than introducing new capabilities. OpenAI noted that earlier versions sometimes declined safe questions or interrupted responses with lengthy disclaimers. The new model is designed to provide more direct answers, getting straight to the response without lengthy preambles about safety boundaries.

Internal evaluations indicate significant improvements in factual reliability. OpenAI reported a 26.8% reduction in hallucination rates when GPT-5.3 Instant uses the web and a 19.7% decrease when relying solely on its internal knowledge, compared to prior models. User-feedback evaluations showed a 22.5% decrease in hallucinations with web access and a 9.6% decrease without it.

While the term "cringe" was not explicitly defined by OpenAI, the company highlighted stronger writing abilities, citing poetry generation as an example of improvement over GPT-5.2.

Transition and User Reactions

GPT-5.3 Instant replaces the default ChatGPT model immediately. GPT-5.2 Instant will remain accessible for paid subscribers under legacy options until June 3, following a three-month transition period. OpenAI also teased "5.4 sooner than you think," a statement that drew user speculation on X regarding recent controversies.

Social media reactions to the update were mixed. Some users welcomed the more direct responses, while others criticized OpenAI's contract with the U.S. Department of Defense, especially after rival Anthropic declined a similar deal due to safety concerns. Some users expressed a desire for the return of GPT-4o, which they felt offered a more intimate conversational experience.

Lessons from Past Updates

This rollout follows a similar incident last summer when OpenAI replaced GPT-4o with GPT-5, leading to user backlash over the new model feeling colder. This resulted in OpenAI restoring GPT-4o for paid users. In January, OpenAI announced the retirement of GPT-4o and older models as of February 13. OpenAI CEO Sam Altman acknowledged the previous rollout as a lesson in upgrading products used by hundreds of millions of people.