|
推理模型DeepSeek-R1,，在測試表現(xiàn),、訓(xùn)練成本和開源開放程度等多個(gè)基準(zhǔn)測試中均超越美國開放人工智能研究中心（OpenAI）的最新模型o1，但成本僅為o1的三十分之一。 China's homegrown AI model, DeepSeek, has surged to the top of the free app download charts in the United States region of the Apple App Store, surpassing the once-dominant ChatGPT. It also secured the number one spot on the free app rankings in China.
▲ 蘋果APP Store中國區(qū)免費(fèi)榜On Jan 20, Chinese artificial intelligence firm DeepSeek officially launched its latest AI model, DeepSeek-R1, positioning its performance to rival OpenAI's o1 official version.In benchmark tests on Friday, DeepSeek-R1 ranked third across all categories on the international large model leaderboard, Arena. In the style control model (StyleCtrl) category, it tied with OpenAI's o1 for first place, with an Arena score of 1,357, slightly surpassing o1's score of 1,352.In addition to matching o1 in performance, the release of DeepSeek-R1 sent shockwaves through the US tech industry because it is fully open-source and achieved this breakthrough at an exceptionally low cost.
Alexandr Wang, CEO of Scale AI, stated that while the US has arguably maintained an edge over China in the AI race over the past decade, the release of DeepSeek's AI model could 'change everything'. He said that DeepSeek's AI model performance is roughly on par with the best models from the US.Meta CEO Mark Zuckerberg announced on Friday that Meta planned to spend over $60 billion to build a massive data center in Louisiana to power its newest AI model, Llama 4.Meanwhile, shares of chipmaker Nvidia dropped by 3.12 percent last Friday, marking its biggest single-day decline since the Consumer Technology Association exhibition, amid growing concerns over the overall risks of an AI bubble.Yin Mingyue contributed to this story.
|
