Group 1 - xAI announced the free global availability of Grok 4, limiting usage to 5 times every 12 hours, which has led to dissatisfaction among paid users who feel betrayed by the subscription model [1] - Inspur released the "Yuan Nao SD200" super-node AI server, integrating 64 cards into a unified memory system, capable of running multiple domestic open-source models simultaneously [2] - Zhiyuan published the GLM-4.5 technical report, revealing details on pre-training and post-training, achieving native integration of reasoning, coding, and agent capabilities in a single model [3] Group 2 - Kunlun Wanwei launched the SkyReels-A3 model, capable of generating high-quality digital human videos up to one minute long, optimized for hand motion interaction and camera control [4] - Chuangxiang Sanwei partnered with Tencent Cloud to enhance 3D generation capabilities for its AI modeling platform MakeNow, utilizing Tencent's mixed model [5][6] - Alibaba's DAMO Academy open-sourced three core components for embodied intelligence, including a visual-language-action model and a robot context protocol [7] Group 3 - Baichuan Intelligent released the 32B parameter medical enhancement model Baichuan-M2, outperforming all open-source models in the OpenAI HealthBench evaluation, second only to GPT-5 [8] - Lingqiao Intelligent showcased the DexHand021 Pro, a highly dexterous robotic hand with 22 degrees of freedom, designed to simulate human hand functions accurately [9] - A report indicated that 45% of enterprises have deployed large models in production, with users averaging 4.7 different products, highlighting low brand loyalty in a competitive landscape [10][12]

智谱：今天，我们推出全球 100B 级效果最佳的开源视觉推理模型 GLM-4.5V（总参数 106B，激活参数 12B），并同步在魔搭社区与 Hugging Face 开源。这是我们在通向 AGI 道路上的又一探索性成果。此 外，在保持高精度的同时，GLM-4.5V 兼顾推理速度与部署成本，为企业与开发者提供高性价比的多模 态 AI 解决方案。 API 调用价格：低至输入 2 元/M tokens，输出 6 元/M tokens。 响应速度：达到 60-80 tokens/s。 市场有风险，投资需谨慎。本文不构成个人投资建议，也未考虑到个别用户特殊的投资目标、财务状况或需要。用户应考虑本文中的任何 意见、观点或结论是否符合其特定状况。据此投资，责任自负。 风险提示及免责条款 ...

Core Insights - The article highlights the aggressive recruitment of top AI talent from China by major tech companies in Silicon Valley, particularly Meta and OpenAI, leading to unprecedented salary offers [1][2][26]. - It emphasizes the significant contribution of Chinese educational institutions in producing leading AI researchers, with a notable percentage of top AI talent in the U.S. being of Chinese origin [13][22]. - The article raises concerns about the brain drain of Chinese talent to foreign companies, questioning the reasons behind their choices and the implications for China's own AI development [15][30]. Group 1 - Meta has offered a record salary of $200 million to former Apple executive Pang Ruoming, surpassing the salary of Apple CEO Tim Cook [1][4]. - OpenAI has lost key personnel, including Yu Jiahui, who received an $80 million signing bonus and over $300 million in equity [2][3]. - The article notes that many of the top AI talents, such as Yu Jiahui and Pang Ruoming, have backgrounds from prestigious Chinese universities, indicating a strong educational foundation [4][11]. Group 2 - The article cites a report stating that 47% of top AI researchers globally graduated from Chinese institutions, with 38% of researchers in leading U.S. AI firms being Chinese [13][22]. - It discusses the disparity in compensation, with average salaries for graduates from top Chinese universities being significantly lower than those offered in Silicon Valley [16][17]. - The article mentions that over 200 top scholars from China have moved to Silicon Valley in the past five years, with a significant number of Tsinghua and Peking University graduates being "reserved" by U.S. tech companies in 2024 [22][24]. Group 3 - The competition for AI talent is described as a critical phase in the global AI arms race, with companies like Meta striving to catch up with competitors like OpenAI and Google [26][28]. - The article suggests that the recruitment of Chinese talent is not merely a personal choice but reflects deeper systemic issues in China's high-end research ecosystem [24][30]. - It concludes with a call for China to better support and retain its talent to ensure that they contribute to the domestic AI landscape rather than seeking opportunities abroad [30].

很难说GPT-5是失败的。尽管它没有提高太多前沿模型的上限，但通过减少幻觉，大幅提升了它的下限。它在基准测试中表现仍然全面领先，尽管领先优 势微弱，但它又足够便宜，主导了性价比的帕累托边界；而且，它将前沿模型推到了每个用户面前，包括免费用户。 问题出在它的自动切换模型的"路由"。按照GPT-5的系统卡的描述，GPT-5是一个统一的模型系统，包含多款模型，以及一个实时路由系统，根据对话类 型、复杂度、所需工具和明确意图（例如提示中写"请认真思考"）快速决定调用哪个模型。但是，它没有很好地发挥作用。奥特曼的解释是，昨天，"路 由"坏了，一天中的大部分时间都不能使用，结果让GPT-5看起来很傻。他还承诺继续改进"路由"的决策机制。 GPT-5发布还没多久，OpenAI在Reddit的AMA（问我一切）上就被用户希望GPT-4o回来的评论淹没。有人形容它的消失就像"老朋友"的"突然离世"。 GPT-5是OpenAI迫不及待的一次商业化尝试，但它显然在技术上与营销上都没有准备好。用户开始呼唤GPT-4o回来，一方面证明GPT-5没有给他们带去 足够好的用户体验，一方面也证明当前的AI表现已经匹配了市场。 奥特曼显然知道 ...

Core Insights - GPT-5 was launched after multiple delays, showcasing improvements in intelligence, programming, and task reasoning, but did not significantly surpass its predecessor GPT-4 in performance [1][2] - The pricing strategy for GPT-5 is a major highlight, with API call costs being significantly lower than competitors, indicating strong cost-effectiveness in the current large model market [1][6] - The release of GPT-5 reflects a shift in focus from technical breakthroughs to practical applications of AI in daily life, as the excitement around large model parameters diminishes [2][3] Group 1: Evolution of GPT Models - The GPT series has evolved from GPT-1 in 2018 to GPT-5, with significant milestones including the jump from 1.5 billion parameters in GPT-1 to 175 billion in GPT-3, and the introduction of multimodal capabilities in GPT-4 [3][4] - The narrative around large model parameters is shifting towards a focus on multimodal capabilities and practical applications in sectors like healthcare and education [3][4] Group 2: Market Dynamics and Competition - The launch of GPT-5 coincides with a wave of new model releases from competitors, indicating a competitive landscape where companies are racing to capture user attention [7][8] - The slowdown in significant upgrades for large models suggests a maturation of the market, with recent models showing limited improvements compared to earlier iterations [7][8] Group 3: AI Agent Emergence - The rise of AI Agents, which are specialized applications built on large models, is shifting the commercialization focus from general-purpose models to tailored solutions for specific industries [11][12] - OpenAI anticipates that revenue from AI Agents will surpass that from ChatGPT by the end of 2025, reflecting a significant shift in business strategy [11][12] Group 4: OpenAI's Strategic Moves - OpenAI has introduced multiple versions of GPT-5, including a mini and a pro version, indicating a move towards a more customizable and tiered pricing model [6][13] - The company has also resumed releasing open-source models, suggesting a response to industry demands for transparency and collaboration [13]

Core Insights - MiroMind ODR (Open Deep Research) is introduced as a powerful open-source deep research model, achieving a GAIA score of 82.4, surpassing other models like OpenAI's Deep Research and Manus [1][4][32] - The project is fully open-source, including its core model, data, training processes, AI infrastructure, and DR agent framework, allowing for reproducibility [3][4][15] - The team plans to maintain a monthly update schedule for open-source contributions, indicating a commitment to continuous improvement and community engagement [4][21] Performance Metrics - MiroMind ODR achieved a GAIA score of 82.4, while OpenAI's Deep Research scored 67.4 and Manus scored 73.3, highlighting MiroMind's superior performance [4][19] - The model MiroThinker, part of the ODR project, has shown state-of-the-art performance with a score of 60.2% on GAIA-Text-103 [19][21] Project Components - MiroMind ODR consists of four sub-projects: MiroFlow (Agent framework), MiroThinker (model), MiroVerse (data), and MiroTrain (training infrastructure), each contributing to the overall functionality and performance of the deep research model [15][21] - MiroFlow supports multiple mainstream tool calls and extends large language models for tool-assisted deep research reasoning [18][21] - MiroVerse provides 147,000 open-source training datasets, focusing on community feedback and continuous updates [21][32] Leadership and Vision - Dai Jifeng, a prominent figure in the project, has a strong academic background and extensive experience in computer vision and deep learning, having published over 80 papers with significant citations [26][30] - The mission of MiroMind is to develop self-aware digital entities that evolve with the community, aiming for safe and beneficial AGI [30][32]

RT David Sacks (@DavidSacks)A BEST CASE SCENARIO FOR AI?The Doomer narratives were wrong. Predicated on a “rapid take-off” to AGI, they predicted that the leading AI model would use its intelligence to self-improve, leaving others in the dust, and quickly achieving a godlike superintelligence. Instead, we are seeing the opposite:— the leading models are clustering around similar performance benchmarks;— model companies continue to leapfrog each other with their latest versions (which shouldn’t be possible i ...

What do you think about Meta buying up talent with huge salaries and and the heating up of this battle for talent. You know, there's a strategy that that Meta is taking right now. I think that from my perspective at least, I think the people that are real believers in the mission of AGI and what it can do and understand the real consequences both good and bad from that and what's what that responsibility entails.I think they're mostly doing it to be like myself to be on the frontier of that research. So, yo ...

RT Steve Jurvetson (@FutureJurvetson)Updated with the latest GPT-5Grok is still in a league of its own.And learning more rapidly.In the path to AGI, process >> product. https://t.co/yZMRkH8sbv ...

以下文章来源于新皮层NewNewThing ，作者王杰夫 新皮层NewNewThing . 关注AI，提供洞察。 2025.08. 09 本文字数：2684，阅读时长大约4分钟 作者 | 新皮层NewNewThing 王杰夫 8月8日凌晨，OpenAI终于发布了新一代有整数编号的GPT模型GPT-5，距上一代GPT-4发布已经过去2年4个月零24天。 过去，每一代GPT模型都标志着某种技术上的突破：随着参数规模扩大，GPT-3「涌现」出了GPT-2没有的智能水平；到了GPT-4，模型开始具有 图像理解相关的多模态能力。相比之下，打磨了2年之久的GPT-5却显得有些「平庸」：OpenAI说GPT-5是个「博士」，但在各项能力上，除了更低 的幻觉——比GPT-4o低约45%，比OpenAI o3低约 80%，GPT-5没有展示出先前模型没有的能力，AGI也没有到来。 幻觉降低是模型最大优化。 甚至连OpenAI自己都不再将GPT-5称作「模型」，而是将其定义为「一个统一的系统」（One unified system）。 不过，对于这个曾被寄予AGI理想的「模型」，OpenAI给出了所有竞争对手中最低的调用价格：G ...