This model is designed to provide more accurate answers and search results on the web are placed in a more appropriate context.
The update also helps reduce answers falling into "cut-offs", limit unnecessary warnings or overly assertive expressions, and reduce the phenomenon of "information illusions".
According to OpenAI, the company has tweaked the Instant model to handle user complaints about tone, level of relevance and coherence in conversations. Previously, GPT-5. 2 Instant was said to have a "unpleasant" tone, sometimes imposing or making unfounded assumptions about user intentions or emotions.
Users also reflect that GPT-5.2 Instant sometimes refuses to answer completely answerable questions, or responds too cautiously to sensitive topics.
GPT-5.3 Instant is adjusted to reduce unnecessary rejection, while limiting defensive openings when responding. The model will also no longer give too many warnings after self-suggesting that the user has bad intentions.
In addition, GPT-5.3 Instant also provides higher quality answers based on information from the web. OpenAI said the new model has better balance between online data and available knowledge, thereby reducing the risk of over-reliance on search results.
Currently, GPT-5.3 Instant has been provided to all ChatGPT users. Updates for Thinking and Pro versions are expected to be deployed in the near future.
