OpenAI announced an updated version of the GPT-5.5 Instant model. The new version significantly improves its ability to understand user intent, capturing the underlying meaning behind questions and performing more reliably and stably when handling complex tasks.

GPT-5.5 Instant was officially launched on May 5 this year. In high-risk areas such as medicine, law, and finance, the hallucination rate has decreased by 52.5% compared to the previous version. It also shows significant improvements in mathematical, scientific, and visual reasoning capabilities. An upgrade on May 30 made the model's responses more natural, readable, and structured, reducing the appearance of long lists. This latest upgrade further enhances the model's insight into user intent, making the output more aligned with actual user needs.

More reliable for complex tasks, more practical for shopping recommendations

The upgraded model performs more stably when handling complex constraints, accurately meeting the needs of multi-condition tasks. Especially in scenarios such as shopping and local recommendations, the model's output becomes more practical and coherent. When users ask about product choices, consumption advice, or nearby services, the recommendations better connect with the context and are more in line with real-world usage scenarios, no longer being rigid listings but truly "understanding what you want" responses.

Paid users can start using the new GPT-5.5 Instant today, while free users will be able to access it tomorrow. From reducing hallucinations to optimizing expressions, and deepening the understanding of user intent, OpenAI is continuously raising the baseline of the smart assistant experience through frequent iterations.