OpenAI has updated its most popular ChatGPT, introducing GPT -5.3 Instant. The new version aims to provide more accurate answers and better tailored results when searching the web. According to the company, the update reduces unnecessary dead ends, caveats, and overly declarative phrases, while having fewer “hallucinations.”
See also: ChatGPT: Deep research tool adds built-in document viewer
ChatGPT – GPT-5.3 Instant: Tone – Relevance Improvement
OpenAI explains that the Instant model was modified to address complaints about tone, coherence, and conversation flow, issues that didn't show up in benchmarks. GPT-5.2 Instant had a 'cringe' tone that could be exaggerated or make unfounded assumptions about the user's intent or emotions. The new model will have a more natural conversational style and reduce dramatic phrases like: 'Stop. Take a breath.'
See also: ChatGPT's most affordable options display ads
Answers to more questions
Additionally, users reported that GPT-5.2 Instant denied questions it should be able to answer or responded in ways that seemed overly cautious around sensitive topics. GPT-5.3 Instant reduces denials and moderates overly defensive or moralistic responses when answering a question. The model will no longer be overly “cautious” due to assumptions that the user has bad intentions.
Higher quality answers
GPT-5.3 Instant also provides higher-quality answers based on information from the web. OpenAI says it is able to better balance what it finds online with its own knowledge, so it is less likely to overstate the web's results.
See also: ChatGPT: New age prediction model for the protection of minors
GPT-5.3 Instant is available to all ChatGPT users now. Updates for Thinking and Pro versions will come in the future.
