OpenAI releases GPT-5.5 Instant, a new default model for ChatGPT
On May 5, 2026, OpenAI announced the launch of its latest foundation model, GPT-5.5 Instant, which will replace the previous default model, GPT-5.3 Instant, for ChatGPT. This new model aims to enhance user experience by reducing inaccuracies, commonly referred to as “hallucinations,” particularly in sensitive fields such as law, medicine, and finance.
Key Features of GPT-5.5 Instant
The introduction of GPT-5.5 Instant comes with several notable improvements and features:
- Reduced Hallucination: The model is designed to provide more accurate information in critical areas.
- Improved Performance: GPT-5.5 achieved a score of 81.2 on the AIME 2025 math test, a significant increase from the 65.4 score of its predecessor.
- Enhanced Multimodal Reasoning: The new model scored 76 on the MMMU-Pro benchmark, outperforming GPT-5.3, which scored 69.2.
- Context Management: GPT-5.5 Instant can reference past conversations and files, including Gmail, to provide personalized responses.
Availability and User Access
OpenAI has made GPT-5.5 Instant available to Plus and Pro users on the web, with plans to extend this feature to mobile users shortly. Additionally, the company intends to roll out access to Free, Go Business, and enterprise users in the coming weeks. This accessibility aims to ensure that a broader audience can benefit from the advancements made in this model.
Memory Sources and User Control
One of the significant updates with GPT-5.5 Instant is the introduction of memory sources. Users will now be able to see where the model generated its answers from, allowing for greater transparency. This feature includes:
- Source Visibility: Users can view the memory sources used by the model for generating responses.
- Source Management: Users have the ability to delete outdated sources or correct inaccuracies in the answers provided by the model.
- Privacy Considerations: When sharing a chat, recipients will not have access to the memory sources, ensuring user privacy.
Developer Access and API Availability
For developers, GPT-5.5 will be accessible through the API as “chat-latest.” The previous model, GPT-5.3, will remain available for paid users for a limited time of three months. This allows developers to transition smoothly while still having access to the older model if needed.
User Feedback and Model Transition
OpenAI’s decision to retire older models has not been without controversy. The withdrawal of the GPT-4o model earlier in February 2026 faced significant backlash from users who had developed a strong connection with it. Many users described GPT-4o as their “best friend” or “a mirror,” emphasizing the emotional bond they formed with the model. This feedback highlights the importance of user experience and connection in AI interactions.
Conclusion
The release of GPT-5.5 Instant marks a significant step forward in OpenAI’s ongoing efforts to improve its AI models. With enhanced accuracy, better context management, and increased transparency, this new model is poised to offer users a more reliable and personalized experience. As OpenAI continues to evolve its offerings, user feedback will play a crucial role in shaping future developments.
Note: This article is based on information available as of May 2026 and may be subject to change as OpenAI continues to develop its technologies.
