OpenAI o3-pro模型发布,但不能聊天
AI前线·2025-06-11 08:39

Core Insights - OpenAI has officially released o3-pro, a sub-version of its strongest model o3, aimed at providing more reliable responses by extending thinking time [1][3] - The o3-pro model has shown superior performance in key areas such as mathematics, science, and programming, as evaluated by users [1][2] - The model is designed for complex problems where reliability is prioritized over speed, with longer response times expected [1][2] Model Performance - Expert evaluations consistently favored o3-pro over o3 in clarity, comprehensiveness, instruction execution, and accuracy [2] - Academic assessments indicate that o3-pro outperforms both o1-pro and o3, using a strict "4/4 reliability" evaluation method [3] User Access and Features - o3-pro will replace o1-pro in the model selector for Pro and Team users, with Enterprise and Edu users gaining access the following week [3] - Temporary chat functionality for o3-pro is currently disabled due to unresolved technical issues, and it does not support image generation [3]