Chinese artificial intelligence company DeepSeek recently released an open version of its inference model R1, quickly sparking discussions in the tech community. Its remarkable breakthrough achievements have astonished venture capitalist Marc Andreessen, who described it as “one of the most stunning and impressive breakthroughs I have ever seen.” The model has also demonstrated capabilities that rival or even surpass those of OpenAI's o1 model in AI benchmark tests.
Particularly noteworthy is DeepSeek's claim that its model training cost is only $5.6 million, while leading companies in the United States require hundreds of millions. This undoubtedly challenges the conventional understanding of AI model development costs.
Garry Tan, CEO of Y Combinator, believes that DeepSeek's success will accelerate the demand for AI inference, driving the development of the entire industry. Yann LeCun, Chief AI Scientist at Meta, also emphasized that DeepSeek's success is not a reflection of the competition between China and the U.S., but rather a strong proof that “open-source models are surpassing proprietary models.” He believes that DeepSeek's growth is attributed to open-source research and tools, which further promote technological advancement, ultimately benefiting everyone.
It is worth mentioning that DeepSeek's AI assistant quickly gained popularity after its release. As of Sunday afternoon, the application had surpassed ChatGPT to become the top free app on the Apple App Store, further demonstrating its popularity.
DeepSeek's rise will undoubtedly have a profound impact on the global artificial intelligence landscape. Its low-cost, high-efficiency model, open-source philosophy, and the complex relationship between China and the U.S. will all become key variables in the future development of the industry.
