DeepSeek生态圈持续扩大。 记者2月16日从腾讯公司获悉，微信搜一搜在调用混元大模型丰富AI搜索的同时，正式灰度测试接入 DeepSeek。 被灰度到的用户，可在对话框顶部搜索入口，看到"AI搜索"字样，点击进入后，可免费使用DeepSeek- R1满血版模型，获得更多元化的搜索体验。 腾讯表示，此次接入DeepSeek的微信AI搜索功能，不仅整合了微信公众号、视频号等腾讯生态内的信 息源，还支持联网搜索，能为用户提供更全面、实时、准确的回答。 目前该能力还在灰度测试中，将根据用户体验和反馈持续优化。 赵刚表示，DeepSeek生态圈不断扩大，将推动大模型上下游合作伙伴、企业和个人用户等向核心模型 集聚，促进我国人工智能大模型产业做大做强，同时激励其他预训练大模型企业改变竞争策略，加快产 品创新步伐，不断降低模型成本。 此外，DeepSeek生态圈持续扩大将加快AI行业渗透的速度，带动AI算力快速发展。 国际数据公司IDC联合浪潮信息日前发布的《2025年中国人工智能计算力发展评估报告》提出， DeepSeek通过其先进的算法优化和高效的模型性能，激发了新的应用场景的需求，显著拉动了数据中 心、端侧及边 ...

百度2月13日发布消息，文心一言将于4月1日0时起全面免费，所有PC端和APP端用户均可体验其最新 模型，包括超长文档处理、专业检索增强、高级AI绘画、多语种对话等功能。 同日，OpenAI也宣布免费版ChatGPT将在标准智能设置下无限制使用GPT-5进行对话。此外，谷歌最新 人工智能模型套件也于近期宣布正式向所有用户开放使用。 据报道，中国信息通信研究院技术与标准研究所工程师龚正表示，头部厂商的技术优势为免费铺平了道 路。免费的基础服务将成为流量入口，通过技术输出、企业级解决方案等增值服务实现商业闭环。这场 竞争的本质，是争夺下一代人机交互标准的制定权。 近日，有消息称，微信正在灰测接入DeepSeek-R1模型。对此，腾讯方面回应称，微信搜一搜在调用混 元大模型丰富AI搜索的同时，近日正式灰度测试接入DeepSeek。 据广州日报报道，腾讯相关人士确认：微信搜一搜在调用混元大模型丰富 AI 搜索的同时，近日正式灰 度测试接入DeepSeek。被灰度到的用户，可在对话框顶部搜索入口，看到"AI 搜索"字样，点击进入 后，可免费使用 DeepSeek-R1 满血版模型，获得更多元化的搜索体验。 近日，国内外 ...

并行科技以超过215%的涨幅位居98家公司的首位；公司表示，其智算云平台已部署智谱清言、 DeepSeek-R1等主流模型。 DeepSeek大热，风险与机遇并存。 自蛇年新春开市以来，A股刮起两股风，一是"哪吒风"，二是"DeepSeek风"。A股市场上，一场围绕着 以DeepSeek为中心的投资盛宴火热开启，几乎沾上DeepSeek的公司均备受二级市场的追捧，并且受到 一众国内上市公司的纷纷力挺。 从百度文心一言，到Kimi以及字节跳动的豆包，再到DeepSeek，中国大模型反复打破大众的固有认 知，这一系列成功的背后，笔者借用《哪吒》系列电影中的经典台词——我命由我不由天。中国在全球 科技领域的话语权更上一个台阶。 近百家公司接入DeepSeek 根据万得热门指数，截至2月14日，蛇年开市以来（2月5日起），DeepSeek指数（含接入DeepSeek大模 型、以及与DeepSeek有合作的公司）累计涨幅超过48%，呈现断崖式领涨。 记者根据公开信息以及投资者互动平台信息梳理，截至2月14日，已有98家A股上市公司宣布接入 DeepSeek大模型。然而，互动易的信息仍在持续更新中，意味着DeepSeek ...

其中，顺网科技获得81家机构到访。 公司在调研中称，DeepSeek的出现将推动AI技术的普及和应用，推动人工智能行业和科技行业的发 展，其低成本和高性能的特点，打破了传统人工智能大模型的限制，也提供了更具性价比的选择，降低 了人工智能的应用门槛。 节后机构调研个股数量有180多只，斯菱股份成为唯一一只调研机构数超百家的股票。 斯菱股份有112家机构调研。 公司在调研中透露，目前公司主要聚焦于谐波减速器的研发和量产，该产品适用于工业机器人、协作机 器人、人形机器人等多个应用场景。 谐波减速器的核心设备已于2024年底陆续到位，安装调试结束后可开始小批量量产。 同时，公司计划将于今年上半年投资第二条产线，旨在储备更多产能，以应对未来市场需求的增长。 在机器人零部件业务的优势方面，斯菱股份指出，公司主业的产业优势是精密零部件制造，与机器人零 部件制造具有一定的协同性，部分研发技术及生产设备可共用。 另一方面，公司将依据积累的市场和量产经验，着力于机器人零部件产品的技术和质量突破。 近期DeepSeek概念火爆，一批概念股获得机构调研。 公司同时透露，目前顺网雲系列云电脑产品现已全区内置Deepseek模型服务，本 ...

DeepSeek大热，风险与机遇并存。 自蛇年新春开市以来，A股刮起两股风，一是"哪吒风"，二是"DeepSeek风"。A股市场上，一场围绕着 以DeepSeek为中心的投资盛宴火热开启，几乎沾上DeepSeek的公司均备受二级市场的追捧，并且受到 一众国内上市公司的纷纷力挺。 从百度文心一言，到Kimi以及字节跳动的豆包，再到DeepSeek，中国大模型反复打破大众的固有认 知，这一系列成功的背后，笔者借用《哪吒》系列电影中的经典台词——我命由我不由天。中国在全球 科技领域的话语权更上一个台阶。 近百家公司接入DeepSeek 根据万得热门指数，截至2月14日，蛇年开市以来（2月5日起），DeepSeek指数（含接入DeepSeek大模 型、以及与DeepSeek有合作的公司）累计涨幅超过48%，呈现断崖式领涨。 并行科技以超过215%的涨幅位居98家公司的首位；公司表示，其智算云平台已部署智谱清言、 DeepSeek-R1等主流模型。 青云科技-U、优刻得-W、杭钢股份、永信至诚股价均翻倍，前2家公司相关产品已上线DeepSeek系列模 型；杭钢股份全资子公司浙江省数据管理有限公司成功完成DeepSeek ...

随着DeepSeek的爆火，炒作DeepSeek概念、打着AI大旗的网络课程也成了"香饽饽"，在互联网上掀起 一波"卖课潮"。 "1000个DeepSeek神级提示词""万字长文教会DeepSeek""DeepSeek15天教程：从入门到精通"……在网络 平台上，与DeepSeek使用相关的课程大量涌现，售价从几元至几十元不等。 然而，这些"包装"精美的课程，不少只是拼凑公开资料，利用消费者的"信息差"牟利。 记者购买了一份"DeepSeek高效使用技巧"的课程，接收的资料中主要内容包含下载软件、注册账号等基 础操作，而所谓的技巧大全则是在官方网站上即可免费查询的提示词样例。而在另一份"开发DeepSeek 百分百潜能"的课程中，课程标题说的是DeepSeek，内容却是另一个人工智能模型的指令技巧。 随着国产AI大模型DeepSeek的爆火，近期市场上出现大批炒作DeepSeek概念的网络课程，更有卖课机 构推出利用DeepSeek"搞钱"的"生财之课"。 "新华视点"记者调查发现，"井喷式"的DeepSeek网课背后，存在一些利用技术焦虑"围剿"消费者的陷 阱。 低价课靠复制粘贴 高价课靠话术"注水" 除了 ...

Group 1: Company Insights - Siling Co., Ltd. has become the only stock with over 100 institutional surveys, receiving attention from 112 institutions, focusing on the development and mass production of harmonic reducers for various robotic applications [1] - The company plans to invest in a second production line in the first half of this year to increase capacity in response to future market demand [1] - Shunwang Technology has been visited by 81 institutions, highlighting the impact of the DeepSeek concept on AI technology proliferation and its cost-effective solutions [2] - The company has integrated DeepSeek models into its cloud computer products, enhancing user experience for over a million active users [2] - UCloud has adapted DeepSeek open-source models to provide various cloud services, with a self-owned computing power scale exceeding 10,000 PetaFLOPS by early 2025 [3] - Capital Online has successfully deployed multiple domestic and international large models on its cloud platform, with over 20,000 chips for computing power, primarily for inference tasks [4] Group 2: Market Performance - The average increase in stock prices of companies surveyed post-holiday is nearly 13%, significantly outperforming the market average [5] - DeepSeek concept stocks like UCloud and Meiri Interactive have seen their prices double, while Capital Online, Yingfang Software, and Dian Diagnostics have increased by over 70% [5] - The largest decline was observed in Jerry Holdings, with a cumulative drop of over 10% post-holiday [6]

Core Viewpoint - The rapid rise of the domestic AI model DeepSeek has led to a surge in online courses capitalizing on its popularity, but many of these courses exploit consumer anxiety and offer little real value [1][2][3] Group 1: Course Quality and Pricing - A variety of online courses related to DeepSeek have emerged, with prices ranging from a few yuan to several hundred yuan, but many are simply repackaged public information [1][2] - High-priced courses promise quick wealth accumulation through DeepSeek, with one course priced at 1999 yuan selling over 114 copies, generating more than 220,000 yuan in sales [2] - Consumers have reported that many courses do not deliver on their promises, with complaints about misleading content and ineffective teaching methods [2][3] Group 2: Marketing Tactics - Course providers often use anxiety-inducing language and celebrity endorsements to attract consumers, claiming that knowledge of DeepSeek is essential for success [4][5] - Tactics such as "limited availability" and showcasing past student success stories are employed to create urgency and pressure consumers into purchasing [5][6] Group 3: Regulatory Concerns - The explosion of AI-related courses highlights the need for better regulation in the online education market, as current oversight is insufficient to protect consumers [6] - Experts suggest that regulatory bodies should enhance guidelines for digital platforms and course providers, and establish a penalty system for fraudulent practices [6] Group 4: Consumer Advice - Consumers are advised to maintain a rational approach to purchasing online courses and to be wary of claims promising easy money [7]

Core Viewpoint - The rapid rise of the domestic AI model DeepSeek has led to a surge in online courses capitalizing on its popularity, but many of these courses exploit consumer anxiety and provide little real value [1][2][3] Group 1: Course Quality and Consumer Experience - Many low-priced courses rely on copying and pasting publicly available information, misleading consumers into thinking they are gaining exclusive insights [1][2] - High-priced courses promise quick wealth accumulation through DeepSeek but often fail to deliver, with consumers reporting dissatisfaction and claims of false advertising [2][3] - Complaints from consumers highlight that many courses do not provide practical skills or actionable insights, with instructors often reading from scripts without offering real value [2][4] Group 2: Marketing Tactics and Consumer Manipulation - Course providers use anxiety-inducing language and exaggerated claims to attract consumers, suggesting that not using AI tools like DeepSeek equates to being illiterate in business [4][5] - The marketing strategies include leveraging the credibility of "expert" instructors and creating a sense of urgency through tactics like limited availability [5][6] - Some institutions falsely claim to have successful alumni and guarantee income generation, but these promises often do not materialize, leading to consumer frustration [5][6] Group 3: Regulatory and Industry Response - Experts emphasize the need for stronger regulatory oversight in the online education sector to protect consumers from misleading courses [6][7] - Recommendations include improving legal frameworks, enhancing platform accountability, and establishing penalties for fraudulent practices [6][7] - The responsibility also lies with digital platforms to ensure that course providers meet necessary qualifications and adhere to ethical standards [6][7]

Core Viewpoint - The integration of DeepSeek-R1 model into Zhihu's platform is expected to enhance its commercial potential and user engagement, leading to a significant increase in stock price and market interest [1][3]. Company Summary - Zhihu's stock price surged nearly 50% within a week, with a notable increase of 19.4% on February 14, closing at HKD 12.68 per share [1]. - The integration of DeepSeek-R1 model is seen as a strategic move to improve Zhihu's AI search capabilities, enhancing the accuracy and efficiency of content retrieval [2][3]. - Zhihu's financial performance showed a revenue of CNY 845 million for Q3 2024, with a significant reduction in net loss to CNY 9 million, and a gross margin of 63.9% [3]. Industry Summary - The AI search tool usage in China has seen a 230% year-on-year increase, highlighting the growing demand for advanced AI solutions [2]. - The integration of DeepSeek-R1 is expected to improve answer generation efficiency by 40%, addressing the high error rate of over 35% in existing general models within professional fields [2]. - Major tech companies, including Tencent and ByteDance, are rapidly adopting DeepSeek models, indicating a trend towards enhanced AI capabilities across various sectors [4][5].