Since the Chinese AI company Deepseek released the open version of the Reasoning Model R1 earlier this week, many high -tech industries have made a magnificent declaration on what the company has achieved and what it means for AI. I have come.
For example, Venture Capitalist Mark and Leisen posted DeepSeek as “one of the most surprising and most impressive breakthroughs I’ve ever seen.”
R1 seems to match the O1 model of Openai on a specific AI benchmark. Compared to hundreds of millions of dollars paid by major American companies to their training, it costs only $ 5.6 million in training.
It also seems to have achieved that it has faced US sanctions that prohibit highly selling high chips to Chinese companies. Mit Technology Review states that the company’s success believes that sanctions will “revolutionize the startups like DeepSeek, and to innovate in a way to prioritize efficiency, resource pools, and collaboration.” (On the other hand, the Wall Street Journal reported that DeepSeek’s Liang Wenfeng recently told the Chinese Prime Minister that US export restrictions still have a bottleneck.
Curai CEO’s Neal Khosla provides a simpler explanation, and the company claims that “the cost is low to justify the price settings,” all of them impair US AI competitiveness. I want that. ” (KHOSLA does not provide evidence about this, pointing out that his father VINOD is an Openai investor, and the community notes are attached to his post.)
Meanwhile, journalist Holger Zschaepitz suggested that DeepSeek is “the potential for the greatest threat to the US stock market.” If a Chinese company can build a state -of -the -art model at low cost that cannot be accessed to advanced chips, it is called “utility.” Hundreds of billions of aptitude has been poured into this industry. “
In response, GARRY TAN, the CEO of the Y combinator, argued that the success of DeepSeek is actually good for American competitors. “If the training model is faster and easier, he writes in X:
Meta’s AI scientist Yang Lekun argued that he opposed the Deep Shek’s presentation through Chinese and US lenses. Instead, he suggested that the real lesson was “the open source model exceeded its own model.”
“DeepSeek is profitable from open research and open source (such as Meta Pytorch and LLAMA),” Lecun wrote in Linkedin this week. “They came up with new ideas and built them on other people’s work. Because their jobs are open and open source, everyone can make a profit.”
All discussions seem to be trying to try products for consumers. As of Sunday afternoon, Deepseek’s AI Assistant is the top free application of Apple App Store just before Chatgpt.