（文章来源：第一财经） 12月17日，第一财经了解到，大模型初创公司智谱已通过港交所上市聆讯，港股也将迎来以AGI基座模 型为核心业务的上市公司。截至发稿，智谱官方暂未回应。 ...

Core Insights - Tencent has made a significant structural adjustment by hiring Vinces Yao from OpenAI as its Chief AI Scientist, indicating a shift towards prioritizing AI at the highest corporate level [1][5][20] - This move reflects Tencent's recognition that developing advanced AI models requires a foundational approach, focusing on infrastructure and data rather than merely application-level innovations [6][10][24] Group 1: Structural Changes - Vinces Yao will report directly to Tencent's President, a departure from the norm where technical leaders report to lower-level executives, highlighting the importance of AI within the company [4][5] - Tencent has established two independent departments: AI Infra, led by Vinces Yao, and AI Data, led by Liu Yuhong, to streamline operations and enhance focus on AI development [7][10] Group 2: Strategic Intent - The restructuring signifies Tencent's strategic shift from a "guerrilla warfare" approach to a more consolidated "group army" strategy in AI development, aiming to compete effectively against rivals like ByteDance [16][19] - Tencent's focus is on building a robust AI foundation, as opposed to merely racing to develop applications, recognizing that it cannot outpace competitors in speed [15][17] Group 3: Competitive Landscape - Tencent's aggressive hiring strategy includes offering salaries up to double the market rate to attract top talent from competitors, indicating a fierce talent war in the AI sector [18][19] - The competition is intensifying, with ByteDance's "Doubao" application capturing significant market share, prompting Tencent to rethink its approach to AI applications [12][13] Group 4: Future Prospects - The integration of advanced AI models with high-quality data from Tencent's platforms, particularly WeChat, could create a powerful AI search capability, potentially transforming the AI landscape in China [21][23] - The cultural shift brought by new hires from OpenAI may challenge Tencent's traditional focus on user experience and incremental improvements, raising questions about the adaptability of its corporate culture [24]

Group 1 - The core viewpoint of the articles revolves around the insights of He Xiaopeng, Chairman of XPeng Motors, regarding the competition in AI between China and the US, emphasizing the differences in entrepreneurial focus and technological development in robotics and AI applications [1][2]. - He Xiaopeng highlights that the US has a strong emphasis on startups in AI, biotechnology, and finance, particularly in the fields of SaaS and physical AI robotics, with high valuations observed in these sectors [1]. - In contrast, Chinese robotics companies tend to focus on joints and control, while US companies are more model-oriented, indicating a competitive landscape where humanoid robots will be a key area of competition among major players [1]. Group 2 - XPeng Motors is set to host its 2025 Technology Day in mid-November, where He Xiaopeng will introduce the concept of "Physical AI" and unveil four significant applications, including the second-generation VLA, Robotaxi, a new generation of IRON, and two flight systems [1]. - The company asserts that these AI applications have clear mass production plans, suggesting that "Physical AI" is not a distant concept but a tangible reality accessible to everyone [1]. - He Xiaopeng expresses optimism about the future of AI, stating that while there may be bubbles in technology phases, the overall market for AI presents significant opportunities, with China focusing more on market applications compared to the US's emphasis on cutting-edge research [2]. Group 3 - He Xiaopeng discusses the concept of AGI (Artificial General Intelligence), noting that current AI primarily relies on imitation learning and reinforcement learning, lacking true creativity [3]. - He emphasizes that true AGI will require advancements in multi-modal learning, world modeling, continuous learning, and long-term planning capabilities, which are still years away from full realization [3]. - The recent shift in XPeng's corporate positioning from "future mobility explorer" to "physical AI world mobility explorer" reflects a broader market perspective that views XPeng as a global technology company driven by AI [3].

以下文章来源于腾讯科技 ，作者晓静 腾讯科技 . 腾讯新闻旗下腾讯科技官方账号，在这里读懂科技！ 来源丨 腾讯科技（ID：qqtech） 作者丨 郭晓静 编辑丨 徐 青 阳 12月17日，2025小米 " 人车家全生态合作伙伴大会 " 举办。在这次大会上，小米MiMo团队负责人 罗福莉完成了首秀演讲。作为前DeepSeek核心成员，罗福莉自从加入小米，就被视为小米AI研发走 向"正规军化"和"极客化"的里程碑。 市场曾期待她能带来像DeepSeek那样颠覆性的"小而美"模型， 而 罗福莉 身上 AI 时代 的 极客特 质， 也许 能与小米初创时期的极客基因完美相融，产生 新 的 化学反应。 罗福莉的首秀 略显 紧张 ， 但 不负众望 ， 她 带来 了一个高效的模型 MiMo-V2-Flash ， 也 抛出 了新的AGI梦想 。 虽然 有 惊喜 ， 但是 此次 小米并没有打破端侧AI的算力天花板， 对于期待"AI手机"变革的用户而 言， 还需要 继续 等待 。 但 今天 的 罗福莉 ， 确实 在 给 小米 ， 讲了 一个 可 围绕 " 人车家 生态 的 " 、 新的 AGI 故事 在她看来，现在的模型大多只是" ...

Group 1 - The core viewpoint of the article revolves around the competitive landscape of AI between China and the US, with a focus on the advancements in robotics and the concept of "Physical AI" as a transformative force in the industry [1][4] - He Xiaopeng, the chairman of XPeng Motors, emphasizes that the US has a higher concentration of startups in AI, particularly in SaaS and physical AI robotics, with high valuations, while Chinese companies focus more on joint and control aspects [1][2] - The upcoming XPeng Technology Day in mid-November 2023 will showcase significant applications of "Physical AI," including the second-generation VLA, Robotaxi, and new flight systems, all with clear mass production plans [1][4] Group 2 - He Xiaopeng discusses the future of AI, predicting breakthroughs in "Physical AI" over the next three years, particularly in areas like autonomous driving and humanoid robots, which may achieve significant advancements [2][3] - The article highlights that while there may be bubbles in technology sectors, the overall AI market presents substantial opportunities, with Chinese valuations being more reasonable compared to the US [3] - The concept of AGI (Artificial General Intelligence) is explored, with He Xiaopeng noting that current AI primarily relies on imitation and reinforcement learning, and true AGI will require further advancements in multi-modal learning and world modeling [3][4]

2025 年12月17日，北京小米 "人车家全生态" 合作伙伴大会现场，展区展台被围观人群层层围住，讨论声与快门声交织 —— 这场年度生态盛会的焦点， 并非缺席的创始人雷军，而是首次以小米 MiMo 大模型负责人身份登台的 "天才少女" 罗福莉。这位原 DeepSeek 核心成员，用一场稍显紧张却金句密集的 演讲，揭开了小米 AI 战略的全新篇章，也抛出了关于人工智能进化路径的行业思考。 一、从 "6 亿年进化" 到 "投影论"：罗福莉的 AI 认知框架 "AI 正以非线性的方式重演人类大脑 6 亿年的进化史。" 演讲开篇，罗福莉便以生物进化为喻，搭建起她对 AI 发展的认知坐标系。在她看来，当前大模型 的 "智能爆发" 并非偶然 —— 语言作为 "人类思维和物理世界在符号空间的投影"，恰好被大模型解码了文本空间的思维轨迹，这也是小米选择从 "语言" 切入，构建 Agent 语言基座模型 MiMo-V2-Flash 的核心逻辑。 这份逻辑在技术细节中得到印证。当天发布的 MiMo-V2-Flash，虽非参数规模顶尖，却在全球开源模型代码与 Agent 测评基准中跻身 TOP2。其关键突破 在于 "极致推理效率 ...

Core Viewpoint - Xiaomi has introduced its new large model MiMo-V2-Flash, which emphasizes efficiency and practical deployment over sheer size, marking a significant step in its AI exploration journey [4][9]. Group 1: Model Overview - MiMo-V2-Flash features a total parameter scale of 309 billion, with only about 15 billion parameters activated during inference, utilizing a MoE (Mixture of Experts) architecture [8]. - The model incorporates Multi-Token Prediction (MTP) technology, designed for high-speed inference and agent workflows, aiming for efficiency rather than just increasing parameter size [8][21]. - Xiaomi's approach to MiMo-V2-Flash is driven by the need for models that are not only intelligent but also practical and deployable in real-world scenarios [21][22]. Group 2: Performance Metrics - During inference, the model achieves a throughput of 5000 to 15000 tokens per second in a single-machine environment, with a single request output speed of 150 tokens per second, representing a speed increase of approximately 2-3 times compared to models without MTP [24][47]. - MiMo-V2-Flash has entered the first tier in seven mainstream evaluations, particularly excelling in the SWE-Bench test with a 71.7% accuracy rate [27][28]. Group 3: Future Directions - The next generation of intelligent agents must be capable of continuous interaction with the real environment, moving beyond mere language processing to a unified, dynamic world model [30][32]. - Xiaomi emphasizes that true intelligence arises from interaction rather than just textual understanding, indicating a shift towards models that can engage with the physical world [52][53].

Core Viewpoint - Xiaomi announced the launch of the Xiaomi MiMo-V2-Flash large model at the "Human-Vehicle-Home Ecosystem" partner conference, highlighting advancements in AI technology and the importance of open-source models in achieving AGI [1][11]. Group 1: Xiaomi MiMo-V2-Flash Model - The Xiaomi MiMo-V2-Flash model has a total of 309 billion parameters, with only 15 billion activated parameters, positioning it among the top two globally in terms of code capability [4]. - The model outperforms DeepSeek V3.2 and K2-Thinking in most benchmark tests, achieving lower costs and higher speeds while maintaining a competitive edge in the global AI landscape [6]. - Xiaomi aims to integrate AI with the physical world, focusing on the development of intelligent agents that can understand and interact with human environments [9][15]. Group 2: Role of Luo Fuli - Luo Fuli, a prominent figure in AI research, emphasized the need for intelligent agents to evolve from merely answering questions to completing tasks, incorporating memory, reasoning, and decision-making capabilities [9]. - She believes that open-source models are crucial for the democratization of AGI, allowing collective human intelligence to evolve [9]. - Luo Fuli's background includes significant academic achievements and contributions to AI, having published multiple papers in top international conferences [10]. Group 3: Xiaomi's AI Strategy - Xiaomi has initiated a "pressure investment" strategy in AI, with the development of various models including MiMo-7B and MiMo-VL, focusing on deep integration of AI with physical applications [15]. - The company is committed to providing a strong platform for top talent in the AI field, as indicated by the recruitment of Luo Fuli [11][15].

Group 1 - The core viewpoint of the articles revolves around the insights of He Xiaopeng, Chairman of XPeng Motors, regarding the competition in AI between China and the US, emphasizing the differences in entrepreneurial focus and technological development in robotics and AI applications [1][2][3] - He Xiaopeng highlights that the US has a strong emphasis on startups in AI, biotechnology, and finance, particularly in the fields of SaaS and physical AI robotics, with high valuations observed in these sectors [1] - In contrast, Chinese robotics companies tend to focus on joints and control, while US companies are more model-oriented, indicating a competitive landscape where humanoid robots will be a key area of competition among major players [1] Group 2 - XPeng Motors is set to host the 2025 XPeng Technology Day in mid-November, where He Xiaopeng will introduce the concept of "Physical AI" and unveil four significant applications, including the second-generation VLA, Robotaxi, a new generation of IRON, and two flight systems [1] - The company asserts that these AI applications have clear mass production plans, suggesting that "Physical AI" is not a distant concept but a tangible reality accessible to everyone [1] - He Xiaopeng expresses optimism about the future of AI, stating that while there may be bubbles in technology phases, the overall market for AI presents significant opportunities, with China focusing more on market applications compared to the US's emphasis on cutting-edge research [2]

12月17日，小鹏汽车CEO何小鹏在微博发文称，近日在美国，和差不多近三十位AI领域的外部和内部 朋友聊天，并分享了一些感受。 何小鹏称，任何科技时代都是阶段或局部有泡沫的，无论是互联网时代，还是新能源汽车时代，但是这 是一个市场从混沌到有序的必然竞争过程。"整体来说，AI必然会推动整个社会巨大的变革，今天还是0 到0.1的最开始阶段。如果一定要说泡沫，感觉中国估值比较合理，美国估值实在有一点高；中国更关 注市场应用，美国更关注前沿研究。" 整体来说，何小鹏指出，当前没有AI泡沫，未来AI的市场有着巨大机遇。 关于AGI的到来 何小鹏表示，今天的AI主要是类似人类的模仿学习（例如学习课本上的知识，学习别人如何操作一个 事情）+ 强化学习（例如反复刷题，反复将一个事情的效果和效率做好最快）。类似自动驾驶，因为可 以快速从百万人的驾驶习惯中学习，所以会看到很快可以比大部分人开车更好更安全，以及出现全新的 涌现能力，但是这些并不是真正的创造力。 关于美国的新创业和机器人 何小鹏表示，美国在AI，生物，金融方面的创业多。而AI方面的创业，从硅谷看，在SaaS和物理AI机 器人领域特别多，估值也特别高，非常集中。中国的 ...