On May 7, 2026, OpenAI unveiled its latest foundational model, GPT-5.5 Instant, which will succeed GPT-5.3 Instant as the primary model for its ChatGPT chatbot. The new model is designed to significantly reduce the occurrence of "hallucinations," particularly in sensitive fields such as law, medicine, and finance, while maintaining the low latency characteristic of its predecessor.
Earlier last month, OpenAI had already introduced the GPT-5.5 model, highlighting enhancements in areas such as code generation and knowledge management. Notably, GPT-5.5 Instant achieved a score of 81.2 on the AIME 2025 mathematics test, a marked improvement over the previous version's 65.4. Additionally, it excelled in the multimodal MMMU-Pro test, scoring 76 points compared to 69.2 for its forerunner.
A key feature of the new version is its improved context management, allowing GPT-5.5 Instant to utilize a search function to reference past dialogues, files, and emails from Gmail, thereby providing more personalized responses. This feature will initially be available to Plus and Pro subscription users on the web, with plans to extend it to mobile users soon. OpenAI also intends to roll out access to this feature for free users as well as Go, Business, and enterprise clients in the coming weeks.
Moreover, ChatGPT will now display memory sources for all models, enabling users to understand the origins of responses and allowing them to delete outdated sources or correct inaccuracies. However, if a chat is shared with others, those individuals will not have access to the memory sources.
For developers, the GPT-5.5 model will be accessible via an API titled "chat-latest," while the previous version will remain available as an option for paying users for three months.
The introduction of GPT-5.5 Instant marks a significant step forward in AI chatbot technology, enhancing reliability and user experience, which could shift the competitive landscape as other companies strive to match or exceed these advancements.
Informational material. 18+.
