OpenAI's newest default model for ChatGPT, GPT-5.5 Instant, has been released with significant improvements in factuality. The company claims that this new model has produced 52.5% fewer hallucinated claims than its previous Instant model for GPT-5.3 on high-stakes prompts covering areas like medicine, law, and finance. Additionally, GPT-5.5 Instant has reduced inaccurate claims by 37.3% on especially challenging conversations users had flagged for factual errors.
Overview
GPT-5.5 Instant has tighter retrieval-augmented grounding and confidence-gated response thresholds, which contribute to its improved factuality. This new model is also more capable across everyday tasks, such as analyzing image uploads and knowing when to turn to the web for an answer. Furthermore, GPT-5.5 Instant has tighter and more to-the-point responses and will avoid using gratuitous emojis.
What it does
With GPT-5.5 Instant, ChatGPT is now more effective at pulling in context from things like previous chats and your Gmail to give you more personalized responses. A new "memory sources" feature will also let the chatbot show what context was used to inform personalized responses, and you can delete or correct information if you need. This feature is rolling out to ChatGPT consumer plans now on the web and soon on mobile.
Tradeoffs
While OpenAI claims that GPT-5.5 Instant has significant improvements in factuality, critics question whether these gains hold under adversarial prompting or real-world deployment. The company will start rolling out GPT-5.5 Instant to all ChatGPT users, though GPT-5.3 Instant will be an option for three months until it's retired. The enhanced personalization will roll out first to Plus and Pro users on the web and is coming soon to the mobile apps.
In practical terms, the release of GPT-5.5 Instant means that ChatGPT users can expect more accurate and personalized responses. However, it's essential to note that the model's performance under adversarial prompting or real-world deployment is still uncertain. As with any AI model, it's crucial to critically evaluate the responses and consider multiple sources before making decisions.