OpenAI is replacing ChatGPT’s default model, GPT-5.3 Instant, with GPT-5.5 Instant in a rollout that began May 5. The company says the update improves factual accuracy, produces more concise responses and makes better use of personal context.
OpenAI says the model is rolling out to all ChatGPT users and is available through the API as chat-latest. The company says paid ChatGPT users will retain access to GPT-5.3 Instant through model configuration settings for three months before its planned retirement.
Accuracy and response changes
OpenAI says GPT-5.5 Instant produced 52.5% fewer hallucinated claims than GPT-5.3 Instant in its internal evaluations of high-stakes prompts covering areas such as medicine, law and finance. The figure measures the relative reduction in hallucinated claims on those evaluation prompts.
The company also says the model performs better on image analysis, mathematics and science tasks. These improvements and the reduction in hallucinated claims remain company assertions without independent performance validation in the available evidence.
OpenAI says responses are more concise while retaining substance, with less unnecessary formatting, fewer gratuitous emojis and fewer unnecessary follow-up questions.
Personalization and memory controls
OpenAI says GPT-5.5 Instant makes more effective use of past chats, uploaded files and connected Gmail accounts when that information is relevant to a response. The company describes improvements in finding context from earlier conversations and deciding when personalization would help.
Enhanced personalization is initially rolling out to Plus and Pro users on the web. OpenAI plans to add mobile support and expand access to Free, Go, Business and Enterprise users in the coming weeks. Availability of specific personalization sources may vary by region.
Separately, OpenAI is introducing memory sources across ChatGPT models. The feature shows some of the context used to personalize an answer, such as saved memories or relevant past chats, and provides controls to correct or remove outdated information.
OpenAI says memory sources may not show every factor that shaped a response or every past conversation the model searched. Users can manage saved memories in settings, delete chats they do not want referenced and use temporary chats that do not use or update memory.
Memory sources are rolling out across consumer ChatGPT plans on the web, with mobile support planned to follow.
