OpenAI has introduced GPT-5.5 Instant, updating ChatGPT’s default model for all users. The update focuses on making responses more accurate, clearer, and more concise, while improving how the model uses existing user context.
It replaces GPT-5.3 Instant and is designed to improve everyday interactions for hundreds of millions of users through stronger responses, a more natural conversational tone, and better use of personalization when relevant.
GPT-5.5 Instant
GPT-5.5 Instant is the updated default model focused on improving factual reliability, reasoning ability, and everyday task performance. It reduces errors compared to GPT-5.3 Instant, particularly in high-stakes areas such as medicine, law, finance, and STEM-related queries.
In internal evaluations, it recorded:
- 52.5% fewer hallucinated claims on high-stakes prompts
- 37.3% fewer inaccurate claims in difficult user-reported conversations involving factual issues
The model also improves performance in image understanding, STEM-related questions, and web search decision-making, along with stronger results in visual reasoning, math, and science evaluations.
Response style has been refined to be more direct and structured, reducing verbosity, overformatting, and clutter. It also limits unnecessary follow-up questions and reduces excessive emoji usage while maintaining clarity and a natural conversational tone.
Key capability areas
- STEM, Math & Science: Improved accuracy in STEM-related queries, with stronger performance in math and science reasoning evaluations.
- Visual Understanding: Better analysis of image and photo inputs with improved visual reasoning capabilities.
- Web Search Intelligence: Improved ability to decide when web search improves response quality and usefulness.
- Response Quality & Communication: More concise answers with reduced verbosity, fewer follow-ups, cleaner formatting, and more structured output while maintaining natural flow and tone.
Memory sources
GPT-5.5 Instant improves personalization using context from past chats, files, and connected Gmail (if enabled). It can retrieve relevant information faster and decide when additional context improves responses, reducing repetition in ongoing conversations.
OpenAI has introduced memory sources across ChatGPT models, allowing users to see what context influenced a response, such as saved memories or past chats.
Users can:
- View context used in responses
- Delete or correct saved memories
- Remove chats from being referenced
- Use temporary chats that do not use or update memory
- Note: memory sources are not shown when chats are shared
Memory sources may show only the most relevant past chats rather than the full retrieval history, and the system is designed to improve transparency over time.
Availability
GPT-5.5 Instant is rolling out as the default ChatGPT model for all users, replacing GPT-5.3 Instant. It is also available in the API as chat-latest.
- GPT-5.3 Instant: Will remain available for paid users for three months via model settings, after which it will be retired.
- Personalization (past chats, files, Gmail): Rolling out first to Plus and Pro users on web, followed by mobile, then expanding to Free, Go, Business, and Enterprise users.
- Memory sources: Rolling out across all consumer plans on web and soon on mobile, with availability of specific personalization sources varying by region.