OpenAI updates GPT-5.5 Instant with better context

OpenAI has updated its latest GPT-5.5 Instant model, the version users most commonly interact with when using ChatGPT. The model now has a better understanding of context and can adapt more effectively to user prompts by adding additional conditions or clarifications.

GPT-5.5 Instant was released publicly in May. At the time, OpenAI said that during testing, the model produced 52.5% fewer hallucinated statements and 37.3% fewer errors.

The model has now been improved to better recognize the user’s core goals or questions, while also maintaining context across multiple consecutive messages during a conversation. GPT-5.5 Instant is also better at handling complex questions and is more likely to answer multiple things the user wants to address. If a user further clarifies a question or challenges the model’s response, OpenAI says the model should adapt more efficiently and provide more relevant answers instead of repeating previous ones.

In addition, the updated GPT-5.5 Instant can better understand location-based context and show genuinely relevant nearby options if, for example, a user asks for restaurant recommendations while traveling. It can also provide product recommendations and company information, as well as display photos when the chatbot determines they could be useful.

Since GPT-5.5 Instant is OpenAI’s default model, these improvements should be noticeable to users.