Update improved accuracy, reduced errors, and delivered clearer, more personalized answers for everyday users.
OpenAI introduced an updated version of its default ChatGPT model, GPT‑5.5 Instant, aimed at delivering smarter, clearer, and more personalized responses. The update was made available to all users, enhancing everyday interactions across a wide range of topics.
The new model showed significant improvements in accuracy and reliability. According to internal evaluations, GPT-5.5 Instant produced 52.5% fewer hallucinated claims compared to GPT‑5.3 Instant, particularly in high-stakes areas such as medicine, law, and finance. It also reduced inaccurate claims by 37.3% in complex conversations previously flagged for factual errors.
The upgrade focused on making responses more concise and easier to understand, while maintaining a natural conversational tone. Users experienced tighter answers with less unnecessary detail, helping them find information more quickly without sacrificing depth.
GPT-5.5 Instant also improved its ability to handle everyday tasks, including better analysis of photo and image uploads, stronger performance in science and mathematics questions, and smarter decisions on when to use web search for more relevant answers.
One highlighted example showed how the model could identify and correct its own mistakes. After initially supporting an incorrect mathematical solution, GPT-5.5 Instant reassessed the error, identified the incorrect algebra step, and provided the correct answer using the quadratic formula. This demonstrated a stronger reasoning process compared to its predecessor.
Overall, the update aimed to make ChatGPT more dependable and user-friendly, offering more accurate, efficient, and personalized responses while reducing unnecessary follow-up questions and overly complex formatting.
Source: OpenAI
