Artículos Relacionados con Large Model

El Centro de Noticias de HTX ofrece los artículos más recientes y un análisis profundo sobre "Large Model", cubriendo tendencias del mercado, actualizaciones de proyectos, desarrollos tecnológicos y políticas regulatorias en la industria de cripto.

Report: DeepSeek Resumes Financing, Pre-money Valuation at 500 Billion Yuan

Several transaction sources revealed to Caijing that AI company DeepSeek has restarted its second round of financing, aiming to raise 50 billion yuan with a pre-money valuation of approximately 500 billion yuan. The signing is planned for late August. DeepSeek completed its first funding round of 50 billion yuan in June with a valuation over 350 billion yuan, marking the largest first-round financing in China's AI large model history. The second round was reportedly paused in late July, partly due to the founder's dissatisfaction with leaked investor meeting details circulating online. It has since been restarted with a desire for a more discreet process. The second-round valuation represents a 43% increase from the first round. If successful, DeepSeek will have raised over 100 billion yuan across both rounds. Investor interest remains high; the first round saw over 100 billion yuan in expressed intent, leaving significant unmet demand. DeepSeek's fundraising scale surpasses competitors like Moonshot AI, which recently raised funds at a $31.5 billion valuation and is planning a Pre-IPO round. Market pricing for these model companies is seen more as an "option" on reaching AGI rather than based on traditional financial metrics. Regarding model development, DeepSeek recently publicly tested the DeepSeek-V4-Flash official version in late July, featuring enhanced Agent capabilities and strong performance on benchmarks. It follows a high-value, low-cost strategy, with its pricing being competitive against rivals like Zhipu GLM-5.2 and OpenAI's GPT-4o. On the OpenRouter platform, DeepSeek-V4-Flash recently ranked first in weekly token consumption. The company faces pressure in training new models, and the release timeline for the V4-Pro official version remains unannounced.

marsbitAyer 02:56

Report: DeepSeek Resumes Financing, Pre-money Valuation at 500 Billion Yuan

marsbitAyer 02:56

Is the iPhone Moment for Embodied AI Coming Soon?

Is the "iPhone moment" for embodied AI approaching? This article, based on a roundtable discussion, presents expert insights on the current state and future of embodied AI. The consensus is that the pivotal "iPhone moment" is still distant. The field is likened to the "brick phone" era, with technology paths—such as VLA and world models—not yet converging. While robotic "motor skills" (e.g., walking) have matured, the "brain" (decision-making, generalization) remains far from commercial readiness. A major bottleneck is data: an estimated tens of millions of data points are needed for a breakthrough, but only around 500,000 currently exist globally. Currently, cost remains prohibitive for widespread labor replacement, making the economic case challenging. However, experts see a three-tiered market potential: a billion-level market for emotional companionship (e.g., entertainment, basic care), a trillion-level market for commercial services (e.g., guides, receptionists), and a massive, long-term opportunity for physical labor in factories and homes. The discussion suggests that while humanoid robots face hurdles, non-humanoid embodied AI applications (like existing service robots) can be deployed sooner. The ultimate vision is for AI to operate seamlessly in the physical world, not just behind screens. Regarding AI tools, participants noted their widespread use for boosting efficiency in coding, research, and teaching. However, they warned against over-reliance due to risks of AI "deception" and the erosion of critical thinking, emphasizing that core judgment must remain with humans. In summary, embodied AI holds immense promise but requires significant progress in brain models, data collection, and cost reduction before achieving its transformative potential. Its development is expected to be gradual, advancing through specific use cases rather than a single explosive moment.

marsbit07/14 05:07

Is the iPhone Moment for Embodied AI Coming Soon?

marsbit07/14 05:07

Hardcore First Look | Ocean Embodied Intelligence Company 'Shihang Intelligence' Secures Record-Breaking 1 Billion in Funding, Zhu Xiaohu, Temasek Place Bets

Breaking News | Ocean Embodied Intelligence company "Shihang Intelligent" secures a record-breaking 1 billion RMB (approximately 10 billion yuan) in Series A financing, with investment from Zhu Xiaohu and Temasek. Author: Qiu Xiaofen | Editor: Yuan Silai Ocean Embodied Intelligence company "Shihang Intelligent" has completed its Series A funding round, raising over 1 billion RMB. This marks the largest single funding round in the global marine robotics field to date. Investors include upstream momentum funds from chip companies "Moore Thread" and "Kunlunxin," Singapore's state-owned investment platform Vertex Growth, and listed company Dyneo, among others. Existing investors like GSR Ventures (whose founder Zhu Xiaohu has invested for the fifth time), Vertex Ventures China, Hua Ying Capital, and Long Capital also significantly increased their investments. Founder and CEO Chen Xiaobo, a 1989-born alumnus of Harbin Engineering University, is a long-time expert in underwater robotics. He received the National Defense Science and Technology Progress Award at age 28 (the youngest recipient) and led the development of China's first commercial underwater cleaning robot. The funds will be used for core technology R&D, global market expansion, and building the industry chain ecosystem to scale the application of marine robots in complex underwater scenarios. The ocean is considered one of the most challenging environments for robotics due to low light, high turbidity, complex currents, limited communication, high pressure, and corrosion. "Shihang Intelligent" focuses on developing core underlying technologies for marine robots, covering six key systems: power, control, sensing, navigation, sealing, and deployment. Its robots are capable of operating at depths from 0 to 10,000 meters with full degrees of freedom, performing complex maneuvers, autonomous navigation, and multi-robot collaboration. Applications include ship cleaning, underwater security, offshore wind power, marine ranching, and seabed inspection. The company's order value for the first half of 2026 alone has exceeded 1 billion RMB. Its "Orca Robot" is used by major shipping companies and has performed maintenance on over a thousand large vessels. In April of this year, the company launched its ocean embodied large model "Cangqiong CEORION." Unlike traditional remote-controlled or pre-programmed robots, this model integrates environmental perception, task understanding, and action generation into a single end-to-end architecture. Trained on millions of hours of commercial operation data and simulation data, it covers 12 major underwater operation scenarios. In simulations, it achieved over 90% task success rate and over 70% zero-shot adaptation capability to unseen environments. A built-in physics reasoning module reduces collision risk by 80%, enabling autonomous operation even with weak or no communication. Recently, "Shihang Intelligent" was selected as a core technology partner for Singapore's Maritime and Port Authority national hull inspection and cleaning program. These advancements indicate marine robotics is moving from pilot projects to scaled applications, with real-world operations generating valuable data to continuously improve robot capabilities. CEO Chen Xiaobo stated the company will continue investing in core marine robotics technology, the embodied intelligence model, and global application scenarios to expand into more high-risk, high-difficulty, and high-value underwater operations.

marsbit06/15 01:58

Hardcore First Look | Ocean Embodied Intelligence Company 'Shihang Intelligence' Secures Record-Breaking 1 Billion in Funding, Zhu Xiaohu, Temasek Place Bets

marsbit06/15 01:58

Alibaba's Yet Another New Business Division: What Signal Does It Send?

Alibaba has established a new "Token Foundry" business unit, merging its Tongyi large model division and Future Life Lab. Led directly by Group CEO Wu Yongming, this marks the company's third significant AI organizational reshuffle in 2026, following the creation of the Alibaba Token Hub (ATH) and a Group Technology Committee. The move signals a strategic shift from consolidating AI resources to accelerating productization and commercialization. The "Token Foundry" name reflects Alibaba's ambition to become a foundational supplier in the AI era, focusing on model development and commercial application. Key teams, including those behind the high-performing HappyHorse video generation model, have been integrated into the new unit. Concurrently, Zhou Jingren, architect of the Qwen model series, has been appointed Group Chief Scientist to lead a new AI Future Research Institute, focusing on long-term technological breakthroughs like Agent capabilities. This restructuring creates a clear four-layer AI architecture within Alibaba: the research institute for frontier exploration, Token Foundry for core models and commercialization, MaaS for platform services, and business units like Qianwen (C端) and Wukong (B端) for end-user applications. The adjustments align with a global trend among tech giants like Google and Microsoft to centralize AI leadership under the CEO and deeply integrate research with business units. The urgency is driven by a narrowing competitive window. Alibaba has announced its AI business is now entering a commercialization phase, with AI-related revenue seeing triple-digit growth for eleven consecutive quarters. The company faces intense competition in the MaaS (Model-as-a-Service) sector from rivals like ByteDance and Tencent. The Token Foundry initiative represents Alibaba's effort to streamline execution and enhance competitiveness in this critical, fast-evolving landscape.

marsbit06/11 10:36

Alibaba's Yet Another New Business Division: What Signal Does It Send?

marsbit06/11 10:36

Valuation Surpasses 200 Billion, Kimi Reportedly Raises 13.6 Billion More, Speeds Up Hong Kong IPO

Beijing-based AI unicoth MoonDark (Kimi) is reportedly in talks for a new funding round aiming to raise up to $20 billion (approximately RMB 136 billion), targeting a post-money valuation of $300 billion (approximately RMB 2.035 trillion). If successful, this would mark its third round in six months and a six-fold increase from its $43 billion valuation in December last year. Last month, the company completed a $20 billion funding round led by Meituan Longzhu, reaching a valuation exceeding $200 billion. According to reports, MoonDark has raised over RMB 376 billion across six rounds, making it the most funded large language model startup in China. Founded in 2023 by CEO Yang Zhilin, the company's core product is the Kimi AI Assistant. In April, it launched and open-sourced its flagship model, Kimi K2.6, which has demonstrated performance comparable to top models like GPT-5.4 in certain benchmarks. Recently, it began beta testing for Kimi Work, a local AI agent for knowledge workers. Commercially, the company's Annual Recurring Revenue (ARR) reportedly surpassed $2 billion in April. Regarding its IPO plans, Bloomberg reported in March that MoonDark is preparing for a listing in Hong Kong, though the process remains in early stages. The funding and IPO pace for leading Chinese AI firms has accelerated notably in 2026, mirroring global trends where companies like OpenAI and Anthropic are also setting new fundraising and valuation records. Securing substantial capital is becoming a critical factor in the competitive landscape alongside model capabilities.

marsbit06/08 07:45

Valuation Surpasses 200 Billion, Kimi Reportedly Raises 13.6 Billion More, Speeds Up Hong Kong IPO

marsbit06/08 07:45

Exclusive from Yingke | Tang Wenbin's 'Yuanli Lingji' Merges with Logistics Robotics Company, and Secures Investment from Zhipu, SenseTime, Jieyue, and Others

Exclusive report: Embodied AI company "Yuanli Lingji" recently completed a new round of financing from major AI model firms including Zhipu AI, Stepfun, and SenseTime, alongside continued investments from industrial backers like Huaqin and SAIC Hengxu. Founded in March 2025 by Tang Wenbin, former co-founder and CTO of Megvii, Yuanli Lingji is a general-purpose embodied AI model company. In a notable move, the company has merged with logistics robotics firm "Atomix" (formerly known as Yuanli Juhe) through a share acquisition. Atomix, which originated from Megvii's logistics robotics business led by Tang in 2016 and was spun off in July 2024, has grown to become the world's second-largest supplier of pallet shuttle robots, with annual revenue nearing 1 billion RMB and over 500 projects globally for clients like Uniqlo and CATL. This merger aims to break the industry's "data deadlock" by combining Atomix's extensive real-world operational data from more than 20 countries with Yuanli Lingji's model training capabilities. The company's embodied AI model "DM0" utilizes a cross-domain training approach, integrating internet semantics, autonomous driving rules, and robotics data to achieve hardware-agnostic, precise manipulation even with a compact 2.4B parameter size. The collective investment from key AI players and the strategic merger signal a shift in the competitive landscape, as major model companies pivot from language tokens to physical actions ("from Token to Action"). The industry is entering a consolidation phase where hardware, AI models, data, and application scenarios converge to scale embodied intelligence, a trend mirrored by recent moves from giants like ByteDance and Skild AI.

marsbit06/05 01:07

Exclusive from Yingke | Tang Wenbin's 'Yuanli Lingji' Merges with Logistics Robotics Company, and Secures Investment from Zhipu, SenseTime, Jieyue, and Others

marsbit06/05 01:07

Competitors Going Public, Kimi Can't Sit Still

Competitors Go Public, Kimi Feels the Pressure Yue Zhi An Mian (Moonshot AI), the company behind the AI assistant Kimi, has begun dismantling its VIE and red-chip structure, clearing a key obstacle for a potential Hong Kong IPO. This marks a significant shift from six months ago when founder Yang Zhilin stated the company was in "no hurry" to list. The move comes as rivals like Zhipu AI and MiniMax have successfully listed on the Hong Kong Stock Exchange in early 2026, experiencing massive surges in market value. This has reset valuation logic for AI companies, turning "going public" from an end goal into a competitive necessity. Analysts suggest Kimi is both seizing a favorable market window and responding to competitive pressure. Kimi's valuation has skyrocketed from around $3 billion at its 2023 founding to over $20 billion by May 2026. Capital is betting on its potential as a future AI platform and gateway, though some caution this "emotional valuation" depends on sustained technological leadership and successful commercialization. Traditionally focused on core model R&D over user growth, Kimi has recently pivoted strategy. While its monthly active users declined through 2025, it shifted focus to Agent development and reducing marketing spend. The release of its K2.5 model in early 2026 reportedly generated substantial revenue, with annual recurring revenue reaching $200 million by April, driven by subscriptions and API services. A $2 billion D-round financing in May signaled investor approval of this commercial shift. However, listing will bring new pressures. Experts predict a listed Kimi would face stricter scrutiny on financial controls, compliance, and R&D efficiency. The narrative must evolve from pure technological breakthroughs to demonstrating clear commercialization paths, sustainable income, and a defensible valuation, balancing model superiority with business performance.

marsbit05/28 10:02

Competitors Going Public, Kimi Can't Sit Still

marsbit05/28 10:02

活动图片