OpenAI has launched GPT-5.3 Instant, the newest version of the model that powers daily conversations on ChatGPT. The update is already live for all users. According to the company, this upgrade focuses on improved accuracy, smoother replies, and fewer unnecessary refusals. It comes after user feedback pointed out that earlier versions were sometimes too cautious or inconsistent in tone.
OpenAI says GPT-5.3 Instant is better at telling the difference between harmful requests and normal questions. This helps reduce abrupt refusals while still keeping safety rules in place. Users can expect clearer and more direct answers in many cases.
The model also handles web-based information more carefully. When browsing is enabled, it combines fresh online data with its internal knowledge to give more complete responses. The goal is to present important details early and reduce confusion in longer replies.
The company tested the model in sensitive areas such as medicine, law, and finance. OpenAI reported that hallucination rates dropped by up to 26.8% when web browsing was used. When relying only on internal knowledge, the drop was around 19.7%. In tests based on user-reported factual errors, hallucinations decreased by 22.5% with web access and 9.6% without it. These results suggest the model is now more reliable, especially when it checks facts online before responding.
GPT-5.3 Instant also aims to improve how it sounds. OpenAI says the model now avoids dramatic or overly reassuring language. Replies are designed to feel more natural and balanced. Users can also adjust tone preferences in Settings, giving more control over how responses appear.
The update is available to all ChatGPT users and developers through the API under ‘gpt-5.3-chat-latest.’ OpenAI confirmed that GPT-5.2 will remain available as a legacy option for paid users for three months and will be retired on June 3, 2026. With this rollout, the company is clearly focusing on making ChatGPT more accurate, useful, and user-friendly for everyday tasks.
Also Read: 4 Things ChatGPT Is Absolutely Terrible At (And You Should Know)