A new chatbot developed by the Chinese firm DeepSeek is generating significant buzz in the A. I. community and has impacted the U. S. stock market. It reportedly rivals the offerings from top companies like OpenAI and Anthropic, all while utilizing only a fraction of the resources. During the morning on Monday, the app faced outages attributed to high traffic and temporarily restricted registrations due to a cyber attack. Nevertheless, DeepSeek quickly ascended to the position of the most downloaded free app on Apple’s app store, surpassing ChatGPT. I spent my morning experimenting with the chatbot, posing a variety of questions to it as well as to OpenAI’s ChatGPT and Anthropic’s Claude. After some initial exploration, I was genuinely impressed. DeepSeek managed to tackle complex math, physics, and reasoning problems I presented at twice the speed of OpenAI’s ChatGPT. When I posed programming-related questions typical of a technical interview, its answers were just as detailed and prompt as those of its competitors. When I asked DeepSeek to search the web for information—for instance, to write biographies for some co-workers—it exhibited fewer hallucinations compared to ChatGPT.

However, its responses seemed to be of slightly lower quality when creating poems, short stories, vacation plans, and dinner recipes. The model did have its drawbacks. DeepSeek was notably censored for users in the U. S. For example, when I inquired about a summary of the 1989 Tiananmen Square massacre, an event the Chinese government has sought to erase from online discourse, it replied that this topic was “beyond my current scope. ” Thank you for your patience while we verify access. If you are in Reader mode, please exit and log into your Times account, or subscribe for full access to The Times. Thank you for your patience while we verify access. Already a subscriber?Please log in. Want access to all of The Times?Subscribe now.