[DigitalToday reporter Hwang Chi-gyu (황치규)] OpenAI has replaced ChatGPT’s default model from GPT-5.3 Instant to the newly unveiled GPT-5.5 Instant, TechCrunch reported on May 5 (local time).

OpenAI said the new model reduces hallucinations in sensitive areas such as law, medicine and finance, while maintaining the low latency provided by the previous model.

In the AIME 2025 math test, GPT-5.5 Instant scored 81.2. That is higher than the previous model’s 65.4. It also scored 76.0 on the multimodal reasoning benchmark MMMU-Pro, beating the previous model’s 69.2.

Context management has also been strengthened. GPT-5.5 Instant can use a search tool to reference past chats, files and Gmail to provide more personalised answers. It will be offered first to Plus and Pro users on the web, with mobile support to be added soon. It is expected to be available to free, Go Business and enterprise users within weeks.

With this update, ChatGPT now displays memory sources across all models, allowing users to check where an answer was generated. Users can delete old sources or correct wrong answers. Memory sources are not shared even if a conversation is shared with others.