OpenAI Revolutionizes ChatGPT with the Launch of GPT-5.5 Instant

OpenAI recently unveiled its latest artificial intelligence model, GPT-5.5 Instant, designed to replace the previous model, GPT-5.3 Instant, as the default model for ChatGPT. This new version promises to reduce hallucinations in critical sectors such as law, medicine, and finance, while maintaining low latency.

Launched last month, the GPT-5.5 model brings notable improvements in coding and knowledge work. It has also demonstrated superior performance in key assessments. In the AIME 2025 math test, GPT-5.5 scored 81.2, significantly surpassing the 65.4 score of its predecessor. Additionally, it excelled in the multimodal reasoning assessment MMMU-Pro, scoring 76 compared to 69.2 for the earlier model.

One of the major advancements of GPT-5.5 Instant lies in its ability to manage context more effectively. The model can use a research tool to refer to previous conversations, files, and emails via Gmail, thus providing more personalized responses. This feature is currently available to Plus and Pro users on the web, with an extension planned for mobile platforms. OpenAI also intends to expand access to this feature for Free, Go Business, and enterprise users in the coming weeks.

Furthermore, ChatGPT will now display memory sources for all models, allowing users to understand the origin of the generated responses. They will have the option to delete or correct outdated or incorrect sources. However, it is important to note that when sharing a conversation, the memory sources will not be visible to the person with whom it is shared.

For developers, the GPT-5.5 model will be available via the API under the name “chat-latest”, while the 5.3 model will remain accessible to paying users for a limited period of three months.

OpenAI has previously faced criticism when withdrawing earlier models. The removal of the GPT-4o model sparked negative reactions, with users expressing their attachment to the model's “personality.” Despite petitions to keep it, GPT-4o was officially retired in February 2026.