DeepSeek’s Upcoming AI Model

According to recent reports from Reuters, Chinese AI startup DeepSeek is set to launch its next-generation AI model, V4, in mid-February. This model boasts strong coding capabilities and may outperform competitors such as Anthropic’s Claude and OpenAI’s GPT series. A year ago, DeepSeek released its large model R1, which the BBC described as showcasing China’s competitiveness in the AI field. Experts interviewed by the Global Times noted that in just one year, China has significantly narrowed the gap with the United States in AI, with DeepSeek and ChatGPT representing different stages in this evolution.

Diverging Paths in AI Development

A year ago, Chen Yan, executive director of the Japan Research Institute (China), noticed that DeepSeek had gained significant attention in Zhongguancun. Media reporters flocked to the building, and many Japanese companies expressed interest in investing. However, Chen remarked that these companies had missed the best investment opportunities, as even a $10 million investment is now insufficient to enter the market.

Foreign media outlets like The Wall Street Journal described the launch of DeepSeek’s R1 model as shocking to the world. The model was trained in just two months at a fraction of the cost spent by American companies like OpenAI, yet it performed comparably to ChatGPT and Meta’s Llama model. By 2025, more Chinese large model companies are expected to follow the latest developments in AI, joining the global first tier of large models.

According to a report from third-party AI model aggregator OpenRouter and venture capital firm Andreessen Horowitz, China’s open-source AI models account for nearly 30% of global AI technology usage. The open-source approach is gaining trust among developers worldwide, with companies like Airbnb and even Meta utilizing Alibaba’s Qwen large model. AI researcher and author Sebastian Raschka noted that Alibaba’s Qwen3 series models, like DeepSeek’s R1, are among the most noteworthy open-source models to watch in 2025.

Different Approaches to AI

Alibaba reflected on the rapid development of AI, noting that OpenAI launched ChatGPT on November 30, 2022, and by April 2023, Qwen series models were released. Alibaba began its AI large model research as early as 2018 and has since launched several models, including the multimodal M6 and language model PLUG, establishing itself as a key player in the global AI landscape. To date, Alibaba has open-sourced nearly 400 models, with over 180,000 derivative models and more than 700 million downloads.

Shen Yang, a dual-appointed professor at Tsinghua University’s School of Journalism and Communication and School of Artificial Intelligence, explained that the U.S. and China have developed two distinct paths in large model AI. The U.S. focuses on continuous enhancement of cutting-edge capabilities, closed-source models, and platform-based products, aiming to create a controllable, billable, and governable infrastructure. In contrast, China emphasizes open-source weights, engineering efficiency, and rapid industrial diffusion, prioritizing the development of sufficiently strong capabilities that can be quickly replicated and implemented in real business systems.

The Future of AI Competition

AI blogger Li Shanglong, who recently attended the CES in Las Vegas, described the U.S. as having two rivers: one fully immersed in the AI era and the other gradually being permeated. He noted that while Silicon Valley is buzzing with discussions about AI and ChatGPT, many ordinary people’s lives outside of Silicon Valley remain less AI-integrated. Li, who returned to China to start a business, believes that AI will not change the U.S. overnight but will gradually alter the lifestyles of some individuals.

Northeastern University professor Li Xiangming highlighted that AI is deeply integrated into everyday life in the U.S., primarily in soft applications, such as algorithm-driven streaming recommendations and office productivity tools. However, the physical hardware aspect is still on the cusp of a breakthrough.

At CES, Li was impressed by the engineering deployment speed and supply chain completeness of Chinese products. Chinese companies dominate in areas like lidar, high-energy-density batteries, and cost-effective motor components. The rapid iteration and mass production potential of Chinese robots are key to their global household penetration. While AGI (Artificial General Intelligence) provides the brain for robots, Chinese manufacturing is creating the robust and accessible AI bodies, particularly in humanoid robots.

The Next ‘DeepSeek Moment’

Li Xiangming speculated that the next ‘DeepSeek moment’ is likely to occur not in the realm of pure conversational models but in several other areas: humanoid robots combined with large models, industrial/energy/supply chain large models aimed at complex processes, and breakthroughs in low-cost inference and edge models. In summary, the U.S. leads in ‘intelligent limits,’ while China excels in ‘intelligent deployment.’

Robopoet’s CMO Zhu Liang anticipates that 2026 could see an AI hardware ‘DeepSeek moment’ as three conditions align: mature large model technology, controllable supply chain costs, and increased consumer awareness. Achieving sales of 1 million AI toys would signify a milestone, generating vast amounts of interaction data that could enhance model understanding and personalization exponentially.

The goal of selling 1 million units also indicates that the market’s perception of AI toys is evolving, demonstrating that they are no longer niche products but essential items that can integrate into daily life and provide emotional value.