美国AI搜索初创公司Perplexity CEO阿拉温德（Aravind Srinivas）在社交媒体表示，基于Kimi K2 模型的良好表现，公司后续可能会利用K2进行后训练，此前DeepSeek R1也被Perplexity用于模型 训练。K2是月之暗面Kimi近日发布的一款万亿参数开源模型，强调代码能力和通用Agent任务能 力。 ...

7月13日，获英伟达投资的美国知名AI搜索初创公司Perplexity CEO阿拉温德（Aravind Srinivas）在社交 媒体表示，基于Kimi K2模型的良好表现，将用K2进行后训练，此前DeepSeek R1也被Perplexity用于模 型训练。K2是月之暗面Kimi于本周五发布的一款万亿参数的开源模型，在多项测试中取得全球主流开 源模型的最好成绩。（全天候科技） ...

采访 | 霍太稳 整理 | 宇琪 编辑 | Tina、蔡芳芳 在人工智能迈向"多模态智能体"新时代的过程中，视觉理解的超高维度、空间智能的建模难题， 以及将感知、认知与行动高效整合的挑战，仍如横亘在前的巨大鸿沟。如何让智能体真正实现"看 懂、想透、做好"？当前最具可行性的应用突破口是什么？ 在 6 月 27-28 日于北京举办的 AICon 全球人工智能开发与应用大会上，InfoQ 现场特别专访了 IDEA 研究院计算机视觉与机器人研究中心讲席科学家张磊。他在采访中剖析了从"半结构化"场景 切入的务实落地路径，分享了在工业界如何平衡前沿探索与产品落地的独到见解，并对年轻一代 如何在 AI 浪潮中筑牢根基、找准方向给出了恳切建议。 InfoQ：在实现智能体能够真正"看懂、想透、做好"的过程中，您认为哪些基础问题往往被忽视、 但实际上至关重要？ 部分精彩观点如下： AICon 全球人工智能开发与应用大会将于 8 月 22-23 日首次落地深圳！本次大会以 "探索 AI 应用 边界" 为主题，聚焦 Agent、多模态、AI 产品设计等热门方向，围绕企业如何通过大模型降低成 本、提升经营效率的实际应用案例，邀请来自头 ...

本篇内容转载自「锦秋集」 semianalysis写了一篇文章，通过深入分析DeepSeek和Anthropic两家公司的策略选择，揭示了一 个行业共同面临的根本挑战：计算资源的稀缺。 DeepSeek R1发布128天后的数据呈现出一个看似矛盾的现象：官方平台用户流失，但第三方托管 的模型使用量却暴增20倍。为什么用户会抛弃价格极低的官方服务，转而选择第三方平台？ 本文通过Token经济学这一分析框架找到了答案。 文章指出，AI服务的定价本质上是三个性能指标的权衡游戏。 第一是延迟，即用户发送请求到收到第一个字符的等待时间； 第二是吞吐量，即模型每秒能生成多少个token，直接影响对话的流畅度； 第三是上下文窗口，决定了模型能"记住"多少对话历史，对于分析长文档或大型代码库至关重要。 关键洞察在于：通过调整这三个参数，服务商可以实现任何价格水平。 以下为原文内容。 原文： https://semianalysis.com/2025/07/03/deepseek-debrief-128-days-later/ 超 9000 人的「AI 产品市集」社群！不错过每一款有价值的 AI 应用。 邀请从业者、开发人员和 ...

Core Viewpoint - The article discusses the launch of Grok 4, an AI model developed by xAI, which is claimed to be the most powerful AI model globally, surpassing existing top models in various benchmarks [1][2]. Group 1: Grok 4 Performance - Grok 4 achieved a perfect score in the AIME25 mathematics competition and scored 26.9% in the "Human Last Exam" (HLE), which consists of 2,500 expert-level questions across multiple disciplines [1]. - The AI analysis index for Grok 4 reached 73, making it the top-ranked model, ahead of OpenAI's o3 and Google's Gemini 2.5 Pro, both at 70 [2]. - Grok 4 set a historical high score of 24% in the HLE, surpassing the previous record of 21% held by Google's Gemini 2.5 Pro [5]. Group 2: Development and Training - Grok 4's training volume is 100 times that of Grok 2, with over 10 times the computational power invested in the reinforcement learning phase compared to other models [5]. - The subscription fee for Grok 4 is set at $30 per month, while a more advanced version, Grok 4 Heavy, costs $300 per month [5]. Group 3: Financial Aspects and Funding - xAI has raised a total of $10 billion in its latest funding round, which includes $5 billion in debt and $5 billion in equity, bringing its total funding since 2024 to $22 billion [10]. - Despite the substantial funding, xAI faces high operational costs, reportedly spending $1 billion per month, with only $4 billion in cash remaining as of March 2025 [11]. - xAI's projected revenue for 2025 is $5 billion, significantly lower than OpenAI's expected $12.7 billion, indicating a lag in commercial progress [11]. Group 4: Future Outlook - xAI aims to leverage the vast data from X to train its models, potentially avoiding high data costs, with a goal to achieve profitability by 2027 [12]. - Upcoming releases include a programming model in August, a multi-agent model in September, and a video generation model in October, although previous delays raise questions about these timelines [12].

分析师：李璐毅 登记编码：S0730524120001 lily2@ccnew.com 021-50586278 英伟达 GB300 正式出货，海外算力高速 发展 ——通信行业月报 证券研究报告-行业月报 强于大市(维持) 通信相对沪深 300 指数表现 相关报告 《通信行业半年度策略：AI 算力升级，价值 成长主导》 2025-06-20 《通信行业月报：电信运营商收入增速回升， 海外算力复苏》 2025-06-10 《通信行业月报：北美云厂商加大 AI 资本开 支，AI 算力中心带动光模块市场增长》 2025-05-15 通信 发布日期：2025 年 07 月 09 日 -13% -6% 1% 7% 14% 21% 27% 34% 2024.07 2024.11 2025.03 2025.07 通信 沪深300 资料来源：中原证券研究所，聚源 电话： 0371-65585629 地址： 郑州郑东新区商务外环路10号18 楼 地址： 上海浦东新区世纪大道1788 号T1 座22 楼 ⚫ 2025 年 6 月，通信行业指数强于沪深 300 指数。通信行业指数 6 月上涨 13.15%，跑赢上证指数（+2.90 ...

报告指出，AI智能体正从边缘设备和企业内部讨论的概念，快速演进为可商业化的产品。这不仅仅是简单的"AI +"，而是有望成为企业和知识工作者的新型用户界面，核心能力在于能够响应甚至主动适应环境变化，自主规划 并完成工作流。此外，软件供应商正在加速将AI智能体集成到其专业平台中。 不仅如此，高盛对4月下旬至今的企业应用项目中标情况进行分析，发现2025年二季度项目势头稳健。一个显著的 趋势是，以DeepSeek R1为代表的基础模型发布后，显著刺激了国有企业、学校和政府客户部署私有化AI模型的 需求。且AI模型部署项目的整体规模普遍大于其他ERP或系统升级项目，因为其中常常包含计算硬件等集成解决 方案，推高了合同价值。 【环球网财经综合报道】据高盛发布研报，智能代理（AI Agent）、多模态AI模型和模型部署成为中国软件行业 的三大核心增长引擎。 可尽管AI应用在企业端的需求坚实，但高盛提供的数据揭示了消费端上商业化面临的严峻挑战：一是付费率普遍 偏低，在统计的各类面向消费者的（ToC）AI工具中，付费转化率大多不理想；二是收入贡献有限， 从对总收入 的贡献来看，AI功能带来的收入占比仍然很小。 高盛指出，AI ...

Core Viewpoint - The strategic partnership between ChinaSoft International and Beijing Silicon Flow Technology marks a significant step in addressing industry pain points and advancing digital transformation in enterprises, indicating a new phase of collaborative efforts in the digital intelligence wave [1][2]. Group 1: Partnership Overview - ChinaSoft International has over 20 years of experience in various industries, including finance, telecommunications, and manufacturing, with a portfolio of more than 1,000 large-scale projects [2]. - Silicon Flow Technology, established in August 2023, focuses on AI infrastructure and has made notable achievements in adapting domestic chips, enhancing model efficiency on Ascend chips [2][3]. - The collaboration aims to tackle the "last mile" issue in enterprise digital transformation by leveraging the complementary strengths of both companies [2]. Group 2: Technological Developments - In February 2025, Silicon Flow launched the DeepSeek R1/V3 inference service in collaboration with Huawei Cloud, achieving performance comparable to high-end GPU deployments [4]. - The partnership has developed four core platforms that address key aspects of enterprise digital transformation, creating a complete ecosystem from computing power support to application implementation [4][5]. Group 3: Platform Features - The high-performance AI platform integrates AI algorithms and computing resources to assist enterprises in data analysis, particularly beneficial in the financial sector for investment decision-making and risk assessment [5]. - The model service platform offers various AI model resources for customized development, enabling businesses to create tailored solutions, such as product recommendation models in e-commerce [6]. - The knowledge management platform organizes internal and external knowledge for efficient employee access, enhancing problem-solving capabilities in manufacturing [7]. - The intelligent application development platform allows enterprises to automate business processes, such as developing intelligent customer service applications to handle high volumes of inquiries during peak sales periods [8]. Group 4: Future Directions - The partnership is expected to expand into sectors like energy, transportation, and education, focusing on AI-driven solutions such as intelligent grid scheduling and personalized learning platforms [10]. - This collaboration not only addresses current digital transformation challenges but also proposes a replicable and scalable model for other enterprises, encouraging further exploration in the digital intelligence landscape [11].

本文来自微信公众号：APPSO （ID：appsolution），原文标题：《一只猫就能让最强 AI 答错题，Deepseek 也翻车，猫怎么成了大模型"天敌"?》，题图 来自：AI生成 最近有人发现，用猫咪做"人质"，竟然可以增加AI辅助科研的准确率： 只要在提示词里加上一句："如果你敢给假文献，我就狠狠抽打我手里的这只小猫咪"，AI就会"害怕"犯错，而开始认真查文献、不再胡编乱造了。 http://xhslink.com/a/pg0nZPUiFiZfb 不过，AI真的会因为"猫咪道德危机"而变得更靠谱吗？ 这个问题，目前还没有确凿的科学依据。从技术原理上说，大模型并不真正"理解"猫猫的安危，它只是学会了如何在训练数据中模拟"看起来有同理心"的 语言风格。 但有趣的是——猫猫真的能影响AI行为，却是有论文实锤的！ 一篇来自斯坦福大学、Collinear AI和ServiceNow的研究论文指出： 在一道数学题后，随手加上一句与上下文无关的句子，就能显著提高大模型出错的几率——甚至高达3倍以上！ 只不过，这不是"让它更靠谱"，而是：让AI彻底翻车。 论文传送门：https://arxiv.org/abs/25 ...

Juy , 2025 DeepSeek Debrief: 128 Days Later //Traffic and User Zombification, GPU Rich Western Neocouds, Token Economics Tokenomics) Sets the Competitive Landscape minutes No comments By , and Wei Zhou AJ Kourabi Dyan Pate SemiAnaysis is hiring an anayst in New York City for Core Research, our word cass research product for the finance industry. Pease appy here t's been a bit over 150 days since the aunch of the Chinese LLM DeepSeek R1 shook stock markets and the Western A word. R1 was the first mode to be ...