OpenAI has rolled out GPT-5.5 Instant as the new default model for ChatGPT. The update brings noticeable improvements in accuracy, response quality, and user experience. Users can now enjoy more reliable answers with fewer errors and less unnecessary flair.
Better Accuracy and Cleaner Responses
According to OpenAI, the new model produces 52.5% fewer hallucinated claims in internal tests. It also cuts inaccurate responses by 37.3% in conversations that users previously flagged for factual mistakes.
Transitioning to response style, GPT-5.5 Instant delivers more direct and to-the-point answers while keeping the friendly conversational tone that people love. The model now asks fewer unnecessary follow-up questions and significantly reduces emoji spam, resulting in cleaner and more professional outputs.
Stronger Personalization with Memory Sources
The updated model searches uploaded files, connected Gmail accounts, and past conversations much faster. This helps it generate highly personalized replies and reduces the need for users to repeat the same information in different chats.
Alongside the model upgrade, OpenAI introduced a new Memory Sources feature. This tool shows users some of the context the AI used to personalize responses. People can easily delete or update stored information if it becomes outdated. Importantly, memory sources do not appear when users share their chats with others. Users stay fully in control through options to delete memories, modify saved data, or use temporary chats that leave no trace.
OpenAI noted that the current version does not display every detail yet, but the company plans to expand the feature soon. Personalization improvements started first for Plus and Pro subscribers, while free and other users will get access in the coming weeks. The rollout of GPT-5.5 Instant has already begun worldwide.
