Chinese Deepseek releases have upgraded the R1 AI model in Openai Competition

The Chinese launch of Deepseek, which this year caused shock waves in the markets, quietly released the upgraded version of its artificial intelligence reflection model.

The company did not make an official message, but the Model repository repository was released by Deepseek R1.

Deepseek grew on this year after free open source Model of reasoning R1 is superior to offers from competitors, including Meta and Openai. The low cost and short development shocked the world markets, causing concern that the US technological giants have exceeded the cost of infrastructure and destroying billions of major stock value in the US like AI Stalwart Nvidia. Since then, these companies have been widely resumed.

As in the case of Deepseek R1 debut, the upgraded model was also released from Little Fanfare. This is a model of reasoning, which means that AI can perform more difficult tasks through a step -by -step logical process of thinking.

The updated Deepseek R1 model is located immediately behind the O4-Mini and OPENAI reasoning models on Livecodebench, the benchmark that moderates different indicators.

Deepseek has become a child of a poster how Chinese artificial intelligence develops, despite attempts to restrict the country’s access to chips and other technologies. This month is Chinese technological giants Bake and Favorable revealed as they were By making their AI models more efficient To fight the American semiconductor export curb.

Jensen Juan, CEO Nvidiadeveloping graphical processing units needed to prepare huge AI models, On Wednesday, the US export was cut.

“The United States relied on its policy on the assumption that China could not do the Chips II,” Juan said. “This assumption has always been questionable, and now it is clearly wrong.”

“The question is not whether there will be in China,” Juan added. “That’s it.”