OpenAI releases GPT-5.5 Instant, with comprehensive improvements in intelligence, factual accuracy, and response quality

According to Data Insight Beating monitoring, OpenAI has released GPT-5.5 Instant, a major upgrade to ChatGPT’s default model and now available to all users. GPT-5.5 Instant delivers comprehensive improvements in intelligence, factual accuracy, and response quality. Core improvements include: a significant increase in factual accuracy, with hallucinations reduced by 52.5% in high-risk fields such as medicine, law, and finance, and overall inaccurate statements reduced by 37.3%; at the same time, the model’s capabilities have been strengthened in STEM questions, image analysis, tool use (such as web search), and visual reasoning. The response style is also better optimized—answers are more compact and to the point (about 30% fewer words), with less redundant formatting, fewer excessive emojis, and fewer unnecessary follow-up questions. While maintaining warmth and personality, it avoids clutter.

The new model performs especially well in personalization: it can more intelligently leverage chat history, uploaded files, and connected contexts such as Gmail to provide highly targeted suggestions and plans, while users always remain in full control. In everyday conversations, GPT-5.5 Instant’s tone is more natural and practical—for example, giving concise and useful workplace advice or personalized recommendations based on the user’s preferences (such as choosing a tea shop). OpenAI emphasizes that because Instant is the “daily driver” assistant for most people, these small but comprehensive improvements can significantly enhance the user experience, making ChatGPT a more reliable and efficient AI partner.