![[dab3bb53-06fe-4045-a9ae-ca53d0da9390.mp3]]

GPT-5 Pro独立推导数学新结论，被视为AI拥有生命迹象

简报：

• OpenAI研究人员透露，GPT-5 Pro在阅读一篇关于凸优化问题的数学论文后，独立推导并证明了一个比原文更精确的数学结论。
• GPT-5 Pro将论文中一个边界问题的阈值从1/L精确到了1.5/L，其证明思路与人类学者完全不同，展现了原创性推理能力。
• 尽管原论文作者随后更新了更优的结果，但GPT-5 Pro的独立创新能力已得到证实，标志着AI发展的一个重要转折点。
• OpenAI总裁Brockman将此表现评价为“生命迹象”，认为AI已不再仅仅是知识的复述者，而是具备了独立思考和创新推理的能力。

相关链接：

• AI数学天才诞生：GPT-5 Pro独立证明新定理震撼学界，OpenAI总裁惊呼这是"生命迹象"
• Altman押注GPT-6记忆革命 OpenAI战略转向代理系统开发

GPT-5 Pro Independently Derives New Mathematical Conclusions, Seen as AI Showing Signs of Life 🧠✨

Briefing:

• OpenAI researchers revealed that after GPT-5 Pro read a mathematical paper on convex optimization problems, it independently derived and proved a more precise mathematical conclusion than the original paper.
• GPT-5 Pro refined the threshold of a boundary problem in the paper from 1/L to 1.5/L. Its proof approach was entirely different from human scholars, demonstrating original reasoning capabilities.
• Although the original paper's author later updated with a superior result, GPT-5 Pro's independent innovation capability has been confirmed, marking a significant turning point in AI development. 🚀
• OpenAI President Brockman described this performance as "signs of life," believing that AI is no longer merely a relayer of knowledge but possesses the ability for independent thought and innovative reasoning. 🤔

Related Links:

• AI Math Genius Born: GPT-5 Pro Independently Proves New Theorem, Shaking Academia, OpenAI President Exclaims "Signs of Life"
• Altman Bets on GPT-6 Memory Revolution, OpenAI Shifts Strategy to Agent System Development

字节跳动开源大模型Seed-OSS-36B，支持512K超长上下文

简报：

• 字节跳动宣布开源其最新的Seed-OSS-36B大语言模型，该模型拥有360亿参数。
• 模型具备原生512K超长上下文窗口，远超当前主流开源模型的128K限制。
• Seed-OSS引入了“思考预算”机制，允许用户通过设置token数量来控制模型推理的深度。
• 该模型采用Apache-2.0开源协议，可免费用于学术研究和商业部署。

ByteDance Open-Sources Seed-OSS-36B Large Model with 512K Ultra-Long Context Support

Briefing:

• ByteDance has announced the open-sourcing of its latest Seed-OSS-36B large language model, boasting 36 billion parameters. 🚀
• The model features a native 512K ultra-long context window, significantly surpassing the 128K limit of current mainstream open-source models. 🤯
• Seed-OSS introduces a "thinking budget" mechanism, allowing users to control the depth of the model's inference by setting the number of tokens.
• The model is released under the Apache-2.0 open-source license, making it free for academic research and commercial deployment. ✅

英特尔调整AI芯片战略，发布机架级方案Jaguar Shores

简报：

• 英特尔首次曝光其新一代AI芯片Jaguar Shores，该芯片是其首款机架级解决方案，采用18A工艺并搭载HBM4内存。
• 为集中资源开发Jaguar Shores，英特尔已宣布取消原定的Falcon Shores AI GPU系列。
• 该芯片封装尺寸为92.5mm x 92.5mm，专为高性能计算（HPC）平台设计。
• Jaguar Shores未来将与即将推出的Diamond Rapids Xeon CPU相结合，共同面向AI市场。

相关链接：

• Intel 发布全新机架级 AI 芯片 Jaguar Shores，搭载 HBM4 内存

Intel Adjusts AI Chip Strategy, Unveils Rack-Scale Solution Jaguar Shores

Brief:

• Intel has for the first time unveiled its next-generation AI chip, Jaguar Shores, which is its first rack-scale solution, built on the 18A process and featuring HBM4 memory. 🚀
• To concentrate resources on developing Jaguar Shores, Intel has announced the cancellation of its previously planned Falcon Shores AI GPU series. 🛑
• The chip has a package size of 92.5mm x 92.5mm and is designed specifically for High-Performance Computing (HPC) platforms.
• Jaguar Shores will be combined with the upcoming Diamond Rapids Xeon CPU to jointly target the AI market. 🧠

Related Links:

• Intel Unveils New Rack-Scale AI Chip Jaguar Shores with HBM4 Memory