At 7 o'clock in the morning, Grandma Li in Guangzhou checked her sleep data with the smart wristband health assistant and asked Ant Ah Fu for breakfast recommendations for hypertensive patients through voice chat; At 9am, Xiaochen, a white-collar worker in Wuhan, wore Quark AI glasses and woke up his AI assistant to start cycling, navigation, and commuting. With a "Quark photo", he left behind the scenery he saw; At 10am, Mr. Zhang from Shanghai used Feishu Miaoji to organize the meeting to-do list, quickly generated a list, and synchronized it with the team; At 2 pm, Shenzhen financial researcher Xiao Zhou used Qianwen to organize the latest financial hotspots and data reports, and synchronized them to the work group; At 5:30 pm, Ms. Yu from Beijing used bean buns to generate mental arithmetic questions for her son, and the wrong questions were automatically analyzed and explained; At 7 pm, Mr. Wang from Chengdu obtained home cooked dishes through the Xiao Ai smart speaker; At 9pm, Xi'an video creator Xiao Song used the Jianying AI editing tool to automatically match video BGM and add subtitles; At 11pm late at night, Hangzhou designer Xiaolin used the Tide AI Sleep App to adjust white noise for sleep... This set of daily scenes spanning the north and south is a vivid epitome of the accelerated landing of China's AI industry in 2025. From parenting at home, working in the workplace, to clothing, food, housing, transportation, leisure and health preservation, artificial intelligence technology that used to stay in the laboratory has shed its cold coat and infiltrated into daily life with a practical and approachable attitude. Behind all these changes, the rapid rise of Chinese AI is indispensable, reflecting the comprehensive advancement of China's AI industry from technological breakthroughs to ecological construction. Carrying out the "Artificial Intelligence+" campaign is an important strategic deployment of the Party and the country. "AI+" is also an innovative evolution of "Internet plus". The 2025 government work report points out that we will continue to promote the "Artificial Intelligence+" initiative, better integrate digital technology with manufacturing and market advantages, support the widespread application of large models, vigorously develop new generation intelligent terminals such as intelligent connected new energy vehicles, artificial intelligence smartphones and computers, intelligent robots, and intelligent manufacturing equipment. The combination of policy guidance and market vitality has accelerated the development of China's AI open-source ecosystem. According to HuggingFace data, as of August 2025, the cumulative download volume of open-source models in China has surpassed that of the United States. Global developers vote with their feet, confirming the global appeal of China's open source technology. When daily fireworks and technological innovation meet, Chinese AI is embarking on a development path that combines commercial value and social warmth, while also facing dual challenges of opportunities and challenges in its rapid development. Technological breakthrough: The open source revolution has led China's AI industry to break through from daily scenarios to the depths of the industry. The reason why the domestic AI industry can achieve such widespread implementation is that it has embarked on a path of "open source innovation" that is different from Silicon Valley in the United States. Not long ago, the website of the British journal Nature released the list of the top ten scientific figures for 2025, and Liang Wenfeng, founder of Hangzhou artificial intelligence company DeepSeek, was selected. This spring, the emergence of the open-source model DeepSeeker R1 is an important milestone in China's AI technology breakthrough. Previously, it was widely believed both inside and outside the industry that the world's best big language models were in the hands of a few large American companies. These large models are closed source, expensive to invest in, and have superior performance. They are developed and maintained by a single institution or commercial company, and have absolute commercial competitive advantages. They are the "exclusive games" of a few giants. DeepSeek breaks this situation. With limited computing power and investment, DeepSeek has achieved a significant breakthrough in the paradigm of large model reinforcement learning technology, with performance comparable to the world's most advanced closed source models. In the Artificial Analysis ranking of the mainstream performance testing platform for global artificial intelligence big models, DeepSeek's big model score reached 66 points, significantly narrowing the gap with Gemini and GPT's 73 points. Behind the technological breakthrough is the benign competition situation of Chinese AI, which is both open source and speed oriented. If technological breakthroughs are the 'hard power' of Chinese AI, then the open source model is the 'soft power' that unleashes value. DeepSeek actively breaks the technological monopoly and opens up its model architecture with an open attitude. In February 2025, DeepSeek launched a 5-day "Open Source Week" event to promote transparency and ecological development of AI technology by continuously opening up multiple key projects. In September 2025, in a cover article published in the journal Nature, the model's replicated "cooking recipe" was openly disclosed without reservation, including key parameters, engineering choices, and other core information, which was referred to as "textbook transparency" in the industry. When explaining the reason for choosing Liang Wenfeng, Nature magazine used a thought-provoking term: 'touchable excellence'. Persisting in open source and free means that researchers and developers around the world can continue to innovate on the shoulders of their predecessors. DeepSeek senior researcher Chen Deli said that open source has been a firm path for DeepSeek since its inception. In communicating with community technology developers, we have received a lot of meaningful feedback, which in turn has driven DeepSeek's own development. ”In the DeepSeek V3.2 series launched in December, V3.2 Specialty won gold medals in four major competitions, including the International Mathematical Olympiad and the Global Finals of the International Collegiate Programming Competition, further narrowing the gap between open source and closed source models. DeepSeek's open-source move will make AI as accessible as water, electricity, and networks. Open source and lightweight technology will vigorously promote 'artificial intelligence+', and a wide range of scenarios have emerged. ”Wu Fei, director of the Institute of Artificial Intelligence at Zhejiang University, said. In the ChatbotArena open source big model list, Chinese models such as Zhipu, Moon Dark Face Kimi, and Qwen are all at the forefront, demonstrating the technological strength of China's open source AI. The "2025 Artificial Intelligence Index Report" by Stanford University in the United States clearly states that the quantity and quality of high-performance artificial intelligence models in China are constantly improving, and the performance of artificial intelligence models in China and the United States has almost reached parity. DeepSeek's open source initiative has also brought amazing ecological explosive force: the number of active users of its app has exceeded 30 million in 21 days after its launch, topped the list of free app downloads in Apple's app stores in the United States, Ali and other Internet platforms, WeChat and other "national" applications, new energy cars and other industrial ports have announced access, AI chip manufacturers and computing industry chain enterprises have completed adaptation, and computing demand has significantly increased... Under the positive influence of DeepSeek, from government departments, investment institutions to the upstream and downstream of the industry chain, all sectors are confident and excited. Qwen (Tongyi Qianwen) continues to launch a series of full-size and full modal large models, with over 300 open source models and over 600 million global downloads, resulting in over 170000 derivative models; After the release of the open-source model MiniMaxM2, it surged to the top of the OpenRouter trend list, the world's largest model calling platform, and ranked third in terms of call volume; Volcano Engine recently delivered a remarkable AI performance report. As of December this year, the daily token usage of Doubao Big Model has exceeded 50 trillion yuan, an increase of more than 10 times compared to the same period last year. Not long ago, The Economist reported that Martin Casado, a partner at Silicon Valley venture capital firm a16z, revealed that about 80% of current American AI startups use Chinese open-source models during their financing roadshows, rather than mainstream American products. In the eyes of countries in the global South, open source big models are a precious gift for China to help developing countries keep up with global technological progress. These models are affordable, capable of localization and free deployment. Stephen Minas, a professor at Peking University School of International Law, believes that as a result, there are fewer obstacles and greater opportunities for the development and deployment of AI in the global South. Empowering the industry: AI has become an important engine for improving the quality and efficiency of industries, and breakthroughs in technology ultimately need to be focused on industrial transformation. By 2025, AI will have moved beyond the laboratory and become an important productivity tool that permeates multiple fields such as creative creation, cultural entertainment, software development, and educational audio. It will also enter the stage of deep application from point to surface. The application of current large-scale models has entered a more complex deep-water area, gradually deepening from the initial single point application to the main business scenarios of various industries. ”The judgment of Liu Cong, Vice President and Dean of the Research Institute of iFlytek Co., Ltd., accurately outlines the evolution trend of this industry. The reshaping of industries by AI mainly lies in its core advantages of cost reduction, efficiency improvement, breaking down barriers, and universal empowerment, which are particularly prominent in the field of creative creation. AI is not just another pen for me, it may be my companion, helping me go to more places and achieve creativity that we currently cannot. ”Director Lu Chuan recently admitted as the chairman of the judging panel at the 7th Hainan Island International Film Festival that AI is not just a technology showcase, but also an experiment to reconstruct creative relationships. In early 2024, OpenAI in the United States released a clip generated by the Sora video model, sparking a global discussion on AI video technology generation. Although the AI generated video effects at that time were still crude and could only meet the needs of users to "try and watch", this technological seed grew into a towering tree driving industrial transformation in just two years. With AI video technology as a breakthrough point, the Chinese industry has embarked on a leapfrog catch-up from "following" to "running in parallel" and then to "leading in stages". In the second half of 2024, multiple Chinese AI video generation models will be showcased intensively, forming a competitive situation of fierce competition and driving technological iteration into the "fast lane". Yan Yijun, Vice President of MiniMax (Xiyu Technology), stated that the AI video industry is generally in the early stage of technological exploration, and the generation effect and stability need to be improved. However, Chinese companies have quickly achieved technological breakthroughs by leveraging the collaborative advantages of the open source ecosystem. In June 2024, Ke Ling AI launched the DiT video generation big model, achieving technological breakthroughs in the field of video generation big models. Among them, Ke Ling 2.6 achieved core upgrades and became one of the first models in the industry to achieve AI video audio and visual synchronization generation. In the past year and a half, Ke Ling AI has iterated more than 30 times. Wen Yuanxu, the R&D director of Kwai Kering Technology, introduced that Kering O1, which was officially released not long ago, solved the problem of cross shot consistency of roles and scenes in video generation by relying on multimodal deep collaboration capabilities, and achieved a key breakthrough at the architecture level. Coincidentally, MiniMax is also fully committed to accelerating product technology iteration in 2025, and has successively released new generation video generation models such as Conch 2.3 and Speech 2.6. Conch 2.3 focuses on its powerful dynamic performance as its core highlight, enhancing its physical expression and command following ability. It can present more complex and smooth character body movements, greatly improving accuracy and controllability. Speech2.6 further reduces end-to-end delay in audio generation, reproducing the target tone while generating text fitting, fluent, and natural speech. With the rapid transformation of China's AI video industry from entertainment and experimentation to large-scale productivity applications, domestic models have shown significant advantages in segmented scene adaptation, commercial landing capabilities, and cost-effectiveness. Different platforms rely on their own characteristics to form differentiated competitive patterns. Director Lu Chuan's practice is a vivid footnote to this trend. The company he founded, Yuandongli, has already established a deep cooperation with MiniMax, using its Conch AI platform to develop high-quality AI animated series projects for movies. In the future, it will also land in theatrical films and boutique dramas, exploring new possibilities for creation through the integration of technology and art. Technological advantages are constantly being transformed into industrial dividends. In the field of film and television production, the traditional pain points of "high cost, long cycle" have been completely broken: in Amazon PrimeVideo's "Dynasty of David", more than 70 special effects shots are produced through AI generation tools