<div id="readability-page-1" class="page">China’s DeepSeek unveiled a preview version of its much-anticipated new model on Friday, promising to rival models from OpenAI, Anthropic and Google a year after the then little-known start up took the global AI industry by storm.  The Hangzhou-based company highlighted major upgrades in the new model V4’s reasoning and agentic abilities that could act autonomously on one’s behalf, like writing code. It also boasted new capabilities that enhance the model’s efficiency in processing larger numbers of tokens, the fundamental units of information which AI models use to comprehend instructions.  DeepSeek became the poster child for China’s AI boom after its groundbreaking 2025 release of the R1 model that delivered near industry-leading performance – for allegedly a fraction of the price.  The breakout success tumbled American AI stocks, raising questions about the ever-larger investments into data center buildouts. At the same time, it bolstered confidence in Chinese technology and heated up the tech race with the US.  But analysts said the new model is unlikely to send markets into the kind of frenzy the previous one did.  “R1 shocked US markets because no one expected a Chinese model to compete at that level. V4 is simply a follow-through on that same trend, and trends don’t make headlines the way shocks do,” said Ivan Su, senior equity analyst at financial services firm MorningStar.  The stock market has already priced in the reality that Chinese AI, like DeepSeek, is competitive and cheaper to use than US alternatives, so market reaction this time will be limited, Su added.  Like DeepSeek’s earlier models, V4 is open source, meaning it is available for anyone to use, unlike most American models. The “open” strategy has been one key channel through which China aims to compete with the US, by rapidly scaling up adoptions and rolling out real-life applications in various sectors from e-commerce to robotics.  That strategy also reflects the relative smaller pockets of Chinese AI firms and constraints in accessing cutting-edge chips under Washington’s export controls.  To overcome those restrictions, Chinese developers have been forced to work with domestic chipmakers as the most advanced AI processors from Nvidia and AMD remain out of reach. To fulfill V4’s computing needs, DeepSeek partnered with Chinese tech giant Huawei, which said in a statement Friday that it supports the AI startup with its “Supernode” technology by combining large clusters of its “Ascend 950” chips to provide more computing power.  Wei Sun, principal analyst at market analysis firm Counterpoint Research, highlighted the fact that V4 is run on domestic chips from Huawei and Cambricon, another Chinese AI chipmaker, in comparison to R1, which was trained on Nvidia hardware.  “It allows AI systems to be built and deployed without relying solely on Nvidia, which is why V4 could ultimately have an even bigger impact than R1 — accelerating adoption domestically and contributing to faster global AI development overall,” he said.  Anthropic logo is seen in this photo illustration taken on March 1, 2026.  Dado Ruvic/Reuters  While American proprietary models like Anthropic’s Claude, OpenAI’s ChatGPT and Google’s Gemini remain at the top of the industry ladder for now, Chinese firms are undeniably dominating open systems.  DeepSeek, via a statement on Friday, claimed the V4 has the best agentic coding capability among open-source models and achieves “world class” reasoning capabilities.  The company, in a research paper, also said the V4 outperformed other open models when it comes to broad world knowledge but acknowledged that it still trailed behind industry leaders like Gemini.  DeepSeek’s ascent since last year, however, has come with skepticism. Anthropic and OpenAI have accused the startup of illegally extracting capabilities – or distilling – from their models.  On Thursday, Michael Kratsios, White House director of the office of science and technology policy, also accused foreign entities primarily based in China of conducting “industrial-scale” campaigns to “distill” frontier AI models from US companies. While the memo by Kratsios did not directly name DeepSeek, it again put the company under the spotlight as tensions between the two superpowers continue.  CNN has reached out to DeepSeek for comment on those accusations.</div>

深度求索

<p>中國的人工智能初創公司 DeepSeek 推出了其新模型 V4 的預覽，旨在與 OpenAI、Anthropic 和谷歌競爭。該模型在推理和效率方面進行了重大升級，採用了華為和寒武紀的國產芯片。雖然 V4 是開源的，可能會加速中國的人工智能普及，但分析師表示，由於市場已經對中國人工智能競爭力的預期進行了定價，因此它可能不會像前身 R1 那樣引發市場熱潮。DeepSeek 還面臨着關於在持續的地緣政治緊張局勢中從美國模型中提取能力的指控的審查</p>

China’s AI upstart DeepSeek drops new model. Will it make waves like last year? | CNN Business

OpenAI

谷歌-A

谷歌-C

晨星信息

英偉達

華為

寒武紀

NVIDIA Corporation

中國的 AI 新秀深度求索發佈了新模型。它會像去年那樣引起轟動嗎？ | CNN 商業