OpenAI发布最强模型o3-pro，奥尔特曼称“温和的奇点”到来

Core Insights - OpenAI has launched the o3-pro model, which reportedly outperforms competitors like Google's Gemini 2.5 Pro and Anthropic's Claude 4 Opus in various benchmarks, indicating a significant advancement in reasoning model capabilities [1][4] - The pricing for the previous o3 model has been reduced by 80%, with costs dropping from $10 and $40 per million tokens to $2 and $8, respectively, while o3-pro is priced at $20 for input and $80 for output per million tokens, representing an 87% reduction compared to o1-pro [4] - OpenAI has entered a cloud service partnership with Google, marking a strategic move to reduce reliance on Microsoft, which previously held exclusive cloud service rights [4][5] Model Development and Future Projections - OpenAI's CEO, Sam Altman, anticipates the emergence of cognitive agent systems by 2025, systems capable of novel insights by 2026, and robots capable of executing real-world tasks by 2027, suggesting a transformative decade ahead [7] - Altman emphasizes that advancements in AI could lead to unprecedented productivity increases, with scientists reporting productivity levels two to three times higher than before AI's introduction [7] - The company is also developing the next-generation model, GPT-5, with a tentative release planned for summer 2023, contingent on meeting internal performance benchmarks [9]