ChatGPT Gets GPT-5.3 Instant Update With Less 'Cringe,' Fewer Hallucinations

OpenAI contiguous updated its astir fashionable ChatGPT model, debuting GPT-5.3 Instant. GPT-5.3 Instant is expected to supply much close answers and amended contextualized results erstwhile searching the web. The update besides cuts down connected unnecessary dormant ends, caveats, and overly declarative phrasing, positive it has less hallucinations.


According to OpenAI, it tweaked the Instant exemplary to code complaints astir tone, relevance, and conversational flow, which are issues that don't amusement up successful benchmarks. GPT-5.2 Instant had a "cringe" code that could beryllium overbearing oregon marque unsubstantiated assumptions astir idiosyncratic intent oregon emotions. The caller exemplary volition person a much earthy conversational benignant and volition chopped backmost connected melodramatic phrases similar "Stop. Take a breath."

Users recovered that GPT-5.2 Instant would garbage questions it should person been capable to answer, oregon respond successful ways that felt overly cautious astir delicate topics. GPT-5.3 Instant cuts down connected refusals and tones down overly antiaircraft oregon moralizing preambles erstwhile answering a question. The exemplary volition nary longer "over-caveat" aft assuming atrocious intent from the user.

GPT-5.3 Instant besides provides higher-quality answers based connected accusation from the web. OpenAI says that it is capable to amended equilibrium what it finds online with its ain knowledge, truthful it is little apt to overindex connected web results.

GPT-5.3 Instant is disposable to each ChatGPT users now. Updates to Thinking and Pro volition travel successful the future.
This article, "ChatGPT Gets GPT-5.3 Instant Update With Less 'Cringe,' Fewer Hallucinations" archetypal appeared connected MacRumors.com

Discuss this article successful our forums