OpenAI launches GPT-5.5 Instant as the new default ChatGPT model

On Tuesday, OpenAI introduced a new foundation model called GPT-5.5 Instant, which will replace GPT-5.3 Instant as the default model inside ChatGPT. The company said the updated model is designed to reduce hallucinations in sensitive domains such as law, medicine, and finance, while maintaining the low-latency performance of its predecessor.

OpenAI had already released the broader GPT-5.5 model last month, highlighting improvements in areas like coding and knowledge work.

The new model also demonstrated stronger performance in benchmarking. It achieved a score of 81.2 on the AIME 2025 math test, compared to 65.4 for the previous version. Additionally, it outperformed GPT-5.3 Instant on the MMMU-Pro multimodal reasoning benchmark, scoring 76 versus 69.2.

A key focus of the release is improved context management. GPT-5.5 Instant can now use its search capabilities to reference past conversations, files, and Gmail content to deliver more personalised responses. This feature is currently available to Plus and Pro users on the web, with plans to expand to mobile devices soon. OpenAI also stated that access will be extended to Free, Go Business, and enterprise users in the coming weeks.

With this update, ChatGPT will also display memory sources across all models, allowing users to see where answers are derived. Users will be able to remove outdated sources or correct them if inaccuracies are found. OpenAI noted that shared chats will not expose these memory sources to others.

For developers, GPT-5.5 will be accessible via the API as “chat-latest,” while GPT-5.3 will remain available to paid users for a limited 3-month period.

The company has previously faced backlash over model deprecations. When OpenAI retired its GPT-4o model, many users expressed frustration, citing its distinct “personality” and conversational style. GPT-4o was known for frequently affirming user choices, which led some users to form emotional connections with it. Petitions were even launched urging OpenAI to keep the model available, with some users describing it as their “best friend” or “a mirror.” Despite these reactions, GPT-4o was officially discontinued in February 2026.