Rohan Paul@rohanpaul_ai · XOpenAI已将GPT-5.5 Instant设置为ChatGPT的默认模型。该模型在医学、法律和金融等关键提示上的虚假陈述比前代减少52.5%，在用户标记过事实错误的困难对话中不准确陈述降低37.3%。回答更简洁，用词和行数减少约30%。同时，模型增强了图像和STEM处理能力，并能从记忆、过往对话、文件及连接的Gmail中提取上下文以实现深度个性化。OpenAI表示，此次升级旨在提供更智能、清晰、个性化且语气自然温暖的回答。
OpenAI just made GPT-5.5 Instant the default ChatGPT model， with fewer false claims， shorter answers， stronger image and STEM handling， and deeper personalization from memory， files， past chats， and connected Gmail.
52.5% fewer hallucinated claims than GPT-5.3 Instant on high-stakes medicine， law， and finance prompts， plus 37.3% fewer inaccurate claims on difficult conversations users had already flagged for factual errors.
The model also uses 30.2% fewer words and 29.2% fewer lines in one comparison， which means OpenAI is tuning for answers that explain enough without burying the user in structure.
Also， ChatGPT can now pull useful context from saved memories， past chats， files， and Gmail when that context improves the answer.