An AI assistant created by Chinese startup DeepSeek became the number one most-downloaded app in the Apple App Store over the weekend, sending shockwaves through Silicon Valley and sending major tech stock prices plummeting. Nvidia saw more than $460 billion removed from its market capitalization on Monday, a decline Bloomberg characterized as “the largest in the history of the US stock market. The company said it was competing with the current industry leader: OpenAI’s 01. But what surprised the technology industry the most was that DeepSeek claimed has created a model using only a fraction of the specialized computer chips that AI companies usually have to develop- On Monday, DeepSeek said it was limiting new registrations, citing “large-scale malicious attacks” on the company’s services, according to a message on its website. DeepSeek’s R1 model “challenges the notion that Western AI companies have a significant lead over the Chinese,” said Jack Clark, co-founder of AI startup Anthropic, in a newsletter. Venture capitalist Marc Andreessen called it a “Sputnik AI moment.” Cheng Lu, a research scientist at OpenAI, said that the DeepSeek chatbot showed excellent Chinese conversational skills. “For the first time I can feel the beauty of the Chinese language created by a chatbot,” he said in a post on Sunday. DeepSeek’s AI assistant is now available for free and comes with three main functions. First, users can ask the chatbot questions and receive immediate answers. For example, when WIRED asked for recipe ideas incorporating pomegranate seeds, the DeepSeek chatbot quickly provided a list of 15 options ranging from yogurt parfait to “Middle Eastern-inspired” rice pilaf, but did not mention any specific chefs or recipes. DeepSeek’s app also has a search mode that reveals answers from the internet. When WIRED asked “What are some important news stories today?” Chatbot DeepSeek documented the Israel-Hamas ceasefire and linked to some Western news outlets such as BBC News, but not all of the stories were relevant to the topic. Ironically, one of them was a New York Times story about DeepSeek’s impact on the stock market. Finally, there’s a “DeepThink” mode that allows users to tap DeepSeek’s R1 model, which builds on the company’s existing V3 model. The difference between the two is that R1 has so-called “reasoning” abilities that allow to explain step by step how to reach the Conclusion. For example, when asked “What was the most important historical event of the 20th century?” DeepSeek initially provided very long answers that started with a series of broad questions. “I would have to exclude decades or major themes like war, political change, technological progress, social movements, etc.” The DeepSeek chatbot then went on to mention World War II, the Cold War, and the Holocaust. But before R1 could finish his reply, all the answers disappeared and were replaced by a message that read “Sorry, I’m not sure how to approach this type of question yet. Let’s talk about math, coding, and logic problems!” Some experts and early adopters have noted that DeepSeek, like other technology platforms operating in China, appears to extensively censor topics considered sensitive by the Chinese Communist Party. $20 per month to access the most powerful AI models. Unlike its Chinese counterpart, OpenAI does not reveal the underlying “weights” of the model, which determine how the AI processes information. It has also refused to make public the complete “chain of thought” produced by the model of reason itself.
