单词学习：![[2c196ff6-4462-4161-a181-155895861ff5.mp3]]

新闻播报：![[d3279073-8918-47de-9e44-a9a6db91b4bf.mp3]]

OpenAI推出GPT-5，赋能科学与数学研究新突破

简报：

• OpenAI近日宣布推出其最新的GPT-5模型，该模型具备更强的计算能力和语言理解能力，能够更准确地理解和生成数学公式及科学概念。
• 在药物发现方面，GPT-5能通过分析大量生物医学文献和数据，快速识别潜在的药物分子和疗法，以加速临床试验进程。
• 在新材料研究领域，该模型可以帮助科研人员模拟和预测材料特性，从而加速新材料的研发过程。

相关链接：

• OpenAI 发布新 GPT-5 模型，加速数学与科学研究

OpenAI Launches GPT-5, Empowering New Breakthroughs in Science and Math Research

Brief:

• OpenAI recently announced the launch of its latest GPT-5 model, which boasts enhanced computational power and language understanding capabilities, enabling it to more accurately comprehend and generate mathematical formulas and scientific concepts. 🚀
• In drug discovery, GPT-5 can rapidly identify potential drug molecules and therapies by analyzing vast amounts of biomedical literature and data, thereby accelerating the clinical trial process. 🧪
• In new materials research, the model can assist researchers in simulating and predicting material properties, thus speeding up the development of novel materials. ✨

Related Links:

• OpenAI Releases New GPT-5 Model to Accelerate Math and Science Research

小米开源MiMo大模型，打通具身智能与自动驾驶

简报：小米发布并全面开源了其具身智能大模型MiMo-Embodied，该模型成功打通了自动驾驶与具身智能两大领域，实现了任务统一建模，并在涵盖感知、决策与规划的29项核心基准测试中确立了新的性能标杆。

相关链接：

• 小米发布具身大模型 MiMo-Embodied 并开源
• MiMo-Embodied 开源模型地址

Xiaomi Open-Sources MiMo Large Model, Bridging Embodied AI and Autonomous Driving 🤖🚗

Brief: Xiaomi has launched and fully open-sourced its embodied AI large model, MiMo-Embodied. This model successfully integrates the two major fields of autonomous driving and embodied AI, achieving unified task modeling. It has also established new performance benchmarks across 29 core tests covering perception, decision-making, and planning. ✨

Related Links: 🔗

• Xiaomi Launches and Open-Sources Embodied Large Model MiMo-Embodied
• MiMo-Embodied Open-Source Model Address

马斯克与黄仁勋激辩太空AI算力可行性

简报：埃隆·马斯克预测，鉴于地球在电力和散热方面的限制，太空AI算力将在未来四至五年内凭借免费太阳能和便捷的辐射冷却成为最具成本效益的选择，但英伟达CEO黄仁勋对此表示质疑，认为在散热装置、发射成本、芯片抗辐射及维护技术等方面仍存在巨大挑战。

相关链接：

• 马斯克：未来五年太空 AI 算力将成新趋势，黄仁勋对此表示质疑

Elon Musk and Jensen Huang Clash Over Space AI Computing Feasibility

Brief: Elon Musk predicts that within the next four to five years, space AI computing will become the most cost-effective option, leveraging free solar energy and convenient radiative cooling, given Earth's limitations in power and heat dissipation. ☀️ However, NVIDIA CEO Jensen Huang questions this ambitious vision, believing significant challenges persist in cooling systems, launch costs, chip radiation resistance, and maintenance technologies. 🚀🧠

Related Link:

• Musk: Space AI Computing to Be New Trend in Five Years, Jensen Huang Expresses Doubts

马斯克称月球制造与质量投射器可年产超100太瓦太空能源

简报：

• 11月23日，埃隆·马斯克表示，通过在月球进行制造并结合使用质量投射器，将有能力实现每年超过100太瓦的太空能源产能。
• 此前在11月20日，他曾提出星舰每年应能向轨道运送约300至500吉瓦的太阳能人工智能卫星。

相关链接：

• 月球制造+质量投射器，马斯克瞄准每年100太瓦太空能源产能
• 马斯克计划通过月球制造和质量投射器，实现每年100太瓦太空能源产能

Musk: Lunar Manufacturing and Mass Drivers Could Produce Over 100 Terawatts of Space Energy Annually 💡

Brief:

• On November 23rd, Elon Musk stated that by combining lunar manufacturing with the use of mass drivers, it would be possible to achieve an annual space energy production capacity exceeding 100 terawatts. 🚀
• Previously, on November 20th, he had suggested that Starship should be capable of delivering approximately 300 to 500 gigawatts of solar AI satellites to orbit each year. 🛰️

Related Links:

• Lunar Manufacturing + Mass Drivers: Musk Aims for 100 TW Annual Space Energy Capacity
• Musk Plans 100 TW Annual Space Energy Capacity Through Lunar Manufacturing and Mass Drivers

OpenAI为K-12教师发布免费专用版ChatGPT

简报：

• OpenAI面向美国K-12教师推出了免费的“教师专属ChatGPT”，该平台将运行至2027年6月，为经过验证的教师提供一个访问GPT-5.1 Auto模型并集成Canva等工具的安全工作空间，且默认不将用户数据用于模型训练。

相关链接：

• OpenAI 推出 “教师专属 ChatGPT”，助力 K-12 教育

OpenAI Releases Free Dedicated ChatGPT Version for K-12 Teachers 🍎

Briefing:

• OpenAI has launched a free "ChatGPT for Teachers" exclusively for K-12 educators in the U.S. 🇺🇸 This platform will run until June 2027, providing a secure workspace for verified teachers to access the GPT-5.1 Auto model and integrate tools like Canva. Crucially, user data will not be used for model training by default. 🔒

Related Links:

• OpenAI 推出 “教师专属 ChatGPT”，助力 K-12 教育

谷歌发布Nano Banana Pro模型，支持4K生成与联网搜索

简报：

• 谷歌发布基于Gemini 3的新一代图像生成模型Nano Banana Pro，支持最高4K分辨率输出。
• 新模型具备联网生成能力，可实时搜索网络信息（如食谱、股价）并生成对应图表。
• 为验证来源，模型会为生成内容同时写入SynthID隐形水印和C2PA内容凭证。
• 该模型已在NotebookLM、Slides等应用中集成，定价较前代上涨3-6倍，4K图像为0.24美元/张。

相关链接：

• 谷歌推出影像模型Nano Banana Pro，基于Gemini 3并支持多语言

Google Releases Nano Banana Pro Model, Supporting 4K Generation and Web Search

Brief:

• Google has unveiled Nano Banana Pro, a new generation image generation model based on Gemini 3, supporting up to 4K resolution output. 🖼️
• The new model features web-connected generation capabilities, allowing it to search online information (e.g., recipes, stock prices) in real-time and generate corresponding charts. 📈
• To verify sources, the model embeds both SynthID invisible watermarks and C2PA content credentials into the generated content.
• The model has been integrated into applications like NotebookLM and Slides. Its pricing has increased 3-6 times compared to its predecessor, with 4K images costing $0.24 per image. 💰

Related Links:

• Google Launches Image Model Nano Banana Pro, Based on Gemini 3 and Supporting Multilingual Capabilities