Core Insights - DeepSeek's R1 model has undergone a minor version upgrade, enhancing semantic understanding, complex logical reasoning, and long text processing stability [1] - The upgraded model shows significant improvements in understanding capabilities and programming skills, capable of generating over 1000 lines of error-free code [1] - The R1 model's cost-effectiveness is highlighted, being priced at 1/11 of Claude-3.7-Sonnet and 1/277 of GPT-4.5, while being open-source for commercial use [1] Group 1 - The R1 model has gained global attention since its January release, outperforming Western competitors and causing a drop in tech stocks [2] - Following the release of the V3 model, interest in DeepSeek has shifted towards the anticipated R2 model, which is expected to utilize a mixture of experts model with 1.2 trillion parameters [2] - The latest version R1-0528 has sparked renewed media interest, showcasing competitive performance against OpenAI's models in code generation [2] Group 2 - DeepSeek's low-cost, high-performance R1 model has positively influenced the Chinese tech stock market and reflects optimistic market expectations regarding China's AI capabilities [2] - The upgrade has also shown improvements in reducing hallucinations, indicating that DeepSeek is not only catching up but competing with top models [1]

Core Viewpoint - The article discusses the evolving landscape of AI and entrepreneurship, emphasizing the shift from developing large models to focusing on practical applications and user experience as the core of business growth [4][11][12]. Group 1: AI Model Development and Strategy - The debate on the viability of large models for startups has shifted towards a consensus that practical applications are more important than the models themselves [4][6]. - The emergence of the DeepSeek-R1 model has changed the competitive landscape, leading many companies to pivot from foundational model development to application-focused strategies [5][11]. - Companies are increasingly recognizing that large models will become a common infrastructure, akin to utilities like water and electricity, with a focus on applications driving revenue [11][12]. Group 2: User Experience and Market Dynamics - User experience is identified as the most critical growth metric, with companies needing to adapt quickly to user needs and behaviors [16][22]. - The rapid evolution of foundational models means that companies must continuously innovate and improve their applications to retain user engagement [15][19]. - The article highlights that user habits are hard to change, and once established, they can sustain a product's market position even in the face of new competition [18][22]. Group 3: Robotics and Practical Applications - The article discusses the challenges of human-like robots, emphasizing that practical applications and stability are more important than flashy demonstrations [31][36]. - The development of robots should focus on specific tasks and environments, with a timeline of 3 to 5 years for significant advancements in functionality [34][36]. - The importance of creating reliable products that meet user expectations is stressed, as high accuracy is crucial for user acceptance [36][37]. Group 4: Organizational Changes and Future Trends - Companies are encouraged to adopt a culture of AI integration, with all employees expected to engage with AI technologies [42][43]. - The article suggests that organizations should restructure to incorporate AI capabilities into their core operations, enhancing overall productivity and innovation [42][44]. - The need for entrepreneurs to explore global trends and ideas, particularly from Silicon Valley, is emphasized as a way to foster innovation and avoid homogenization in the startup ecosystem [44][45].

Core Insights - Beijing is emerging as a strategic hub in the AI large model sector, driven by technological innovation and a supportive ecosystem for breakthroughs [1] - The role of angel investors is crucial in the AI industry, providing essential support to startups and helping them take their first steps [4] - The AI large model wave has gained momentum globally since 2023, with early investments in generative models proving to be prescient [5][6] Group 1: AI Development and Investment Trends - The AI large model trend is characterized by a shift from previous waves focused on computer vision and autonomous driving to the current emphasis on AI agents and embodied intelligence [5][6] - Investors are increasingly favoring experienced founders with strong academic and research backgrounds, as seen in the case of companies like DeepMind and the Tsinghua NLP team [12][16] - The emergence of open-source models like Llama has accelerated competition among AI companies, allowing them to shorten development timelines [13] Group 2: Investment Strategies and Market Dynamics - Angel investors are focusing on a select number of projects, often operating in a "water under the bridge" manner, avoiding fully marketized projects [14][15] - The investment landscape is divided between long-term oriented funds that prioritize innovation and those focused on immediate revenue generation [21][22] - The success of companies like DeepSeek highlights the challenges faced by startups in competing with established giants, as the consensus around large models has solidified post-ChatGPT [26][27] Group 3: Entrepreneurial Characteristics and Market Challenges - Current AI entrepreneurs are predominantly scientists or technical experts, forming a close-knit community that is easier to identify and engage with [18][19] - The academic foundation of AI startups is critical, as many successful ventures are built on decades of research and development from their respective institutions [16][20] - The market is witnessing a shift where the ability to innovate is becoming more important than merely having financial resources, as the previous model of "buying capability" is no longer sustainable [27][28]

Group 1 - Elon Musk is attempting to block a major AI deal in Abu Dhabi led by OpenAI unless his own AI startup is involved [1] - Nvidia CEO Jensen Huang stated that China is one of the largest AI markets globally, with a $50 billion market, and emphasized the importance of winning the Chinese platform for global success [2] - DeepSeek has released an open-source version of its R1 model, which reportedly matches the performance of OpenAI's latest o3 model [3] Group 2 - Reed Hastings, co-founder of Netflix, has joined the board of AI startup Anthropic, which aims to explore the impact of AI on work, relationships, and education [4] - OpenAI plans to increase investments in Asia following its expansions in South Korea and Japan, expressing optimism about growth prospects in the region [5]

每经记者｜可杨 每经编辑｜张海妮 丨 2025年5月30日 星期五 丨 NO.1 宇树科技从有限公司变更为股份公司 5月29日，宇树科技向合作伙伴发布通知称，因公司发展需要，杭州宇树科技有限公司即日起名称变更 为"杭州宇树科技股份有限公司"。原公司所有业务由"新公司"继续经营，原公司签订的所有合同继续有 效。 点评：宇树科技从有限责任公司到股份有限公司的转身，是企业自身发展壮大的必然选择，也是科技行 业创新发展的生动缩影。在新的股份制架构下，期待宇树科技凭借更强大的资本实力、更灵活的运营机 制与更高效的治理结构，在科技领域开启新的征程。 大模型明星企业DeepSeek深夜"上新"。5月29日凌晨，DeepSeek开源了R1最新0528版本。DeepSeek目前 没有对该版本进行任何说明，只是"悄悄"地开放了模型。著名代码测试平台Live CodeBench显示，其性 能可以媲美OpenAI最新的o3模型的高版本。也有网友对新版R1的风格进行了测试，结果几乎和OpenAI 的o3差不多。 点评：开源是推动技术进步和生态发展的重要方式。DeepSeek R1新版本的开源，为开发者提供了更多 选择和创新机会，有助于 ...

「DeepSeek 一更新，我们就知道又要放假了。」 昨天，DeepSeek 宣布其 R1 系列推理模型小版本升级，最新版本 DeepSeek-R1-0528 参数量高达 6850 亿，模型在思维深度和推理方面的能力显著提升。 刚刚，DeepSeek 公布了 R1-0528 在各类基准测评上的具体得分情况。R1-0528 在数学、编程与通用逻辑等多个基准测评中成绩亮眼，整体表现接近 o3 与 Gemini-2.5-Pro。 | Benchmarks | DeepSeek-R1- | OpenAI- | Gemini-2.5- | Qwen3- | DeepSeek-R1 | | --- | --- | --- | --- | --- | --- | | | 0528 | o3 | Pro-0506 | 235B | | | AIME 2024 数学竞赛 pass@1 | 91.4 | 91.6 | 90.8 | 85.7 | 79.8 | | AIME 2025 数学竞赛 pass@1 | 87.5 | 88.9 | 83.0 | 81.5 | 70.0 | | GPQA Diamond 科学测试 pass@ ...

Core Insights - The updated R1 model from DeepSeek has significantly improved its capabilities, particularly in reducing the "hallucination" rate, which previously stood at around 21% [1][4]. Model Performance - The new R1 model has achieved top-tier performance in various benchmark tests, surpassing all domestic models and nearing the performance of international leaders like o3 and Gemini-2.5-Pro [4]. - The hallucination rate has been reduced by approximately 45%-50% in tasks such as rewriting, summarization, and reading comprehension, providing more accurate and reliable results [4][18]. - In the AIME 2025 test, the model's accuracy improved from 70% to 87.5% in complex reasoning tasks [18]. Model Features and Capabilities - The updated R1 model can generate longer and more structured pieces of writing, including essays, novels, and prose, while aligning more closely with human writing styles [18]. - The model's coding capabilities have also seen significant enhancements, performing nearly on par with OpenAI's o3-high model in code testing environments [18]. - The new model has a parameter count of 685 billion and supports a context length of 128K in the open-source version [19]. Future Developments - There is considerable anticipation in the industry for the next-generation R2 model, with users expressing their eagerness for its release [19]. - DeepSeek has not commented on speculations regarding the R2 model, but the ongoing competition in the foundational model space remains intense [19].

新浪科技讯 5月29日晚间消息，DeepSeek今日宣布，DeepSeek R1模型已完成小版本升级，当前版本为 DeepSeek-R1-0528。用户通过官方网站、App或小程序进入对话界面后，开启"深度思考"功能即可体验 最新版本。API 也已同步更新，调用方式不变。 工具调用，DeepSeek-R1-0528 支持工具调用（不支持在 thinking 中进行工具调用）； 据介绍，DeepSeek-R1-0528 仍然使用 2024 年 12 月所发布的 DeepSeek V3 Base 模型作为基座，但在后 训练过程中投入了更多算力，显著提升了模型的思维深度与推理能力。官方称更新后的 R1 模型在数 学、编程与通用逻辑等多个基准测评中取得了当前国内所有模型中首屈一指的优异成绩，并且在整体表 现上已接近其他国际顶尖模型，如o3与Gemini-2.5-Pro。 其他能力更新方面，包括幻觉改善，新版 DeepSeek R1 针对"幻觉"问题进行了优化。与旧版相比，更新 后的模型在改写润色、总结摘要、阅读理解等场景中，幻觉率降低了45~50%左右，能够有效地提供更 为准确、可靠的结果； 创意写作，在旧版 R1 ...

目前该升级版的DeepSeek-R1-0528已经全量上线官方网页、APP、小程序等等，API也已经可以接入。 关于DeepSeek官方多么有诚意，我们已经在V3版本的升级上看到了——模型性能大幅提升只是开胃小菜，成本价格比 更是再度优化。这回的更新也是一样，新版本的DeepSeek-R1主要在编程能力上大幅提升。据一家LLM API接入网站 OpenRouter，这回的新版本R1的输入输出价格几乎与先前版本毫无变化！ | or DeepSeek: R1 0528 | | DeepSeek: R1 | (2) | | --- | --- | --- | --- | | Author | V deepseek @ | Author or deepseek @ | | | Context Length | 164K | Context Length | 164K | | May 28th update to the original DeepSeek R1 Performance | A | DeepSeek R1 is here: Performance on par with OpenAl o1. | 14 ...

此外，5月28日晚，小鹏MONA M03 Max版正式上市，根据续航分为两个版本，其中续航502km的版本 售价为12.98万元，600km续航的版本售价为13.98万元。另外，小鹏MONA M03还推出了515长续航Plus 版售价11.98万元，以及620超长续航Plus版售价12.98万元。据小鹏汽车官方微博披露，小鹏MONA M03 加推Max、Plus新版型，上市1小时大定12566台，超过去年上市同期，其中Max版订单占比83%。 5月29日，港股三大指数持续走高，恒生科技指数午后一度涨超2.5%。盘面上，科网股普涨，生物技术 股大涨，中资券商股普涨。主流ETF方面，恒生科技指数ETF（513180）跟随指数强势上扬，小鹏汽 车、美团、同程旅行、舜宇光学科技、金蝶国际等持仓股涨幅居前，其中小鹏汽车午后一度涨近7%。 消息面上，近日DeepSeek在官方交流群中公布，DeepSeek R1模型已完成小版本试升级。用户可在官方 网页、APP、小程序测试（打开深度思考），API接口和使用方式保持不变。DeepSeek在开源社区 Hugging Face也开源了新版R1模型（R1-0528）。目前，市场在 ...