Deepseek China tops the upgrade of the R1 AI model in the Openai competition

The Chinese company Deepseek, which has shocked the markets this year, released a developed version of the artificial smart thinking model.

The company has not issued an official advertisement, but the Deepsek R1 was released on the face of the artificial intelligence model warehouse.

Dibsic has risen to the lead this year after the open source freedom R1 thinking model has been out of offers From competitors including Meta and Openai. Low -cost and short times were shocked by global markets, which raised fears that American technology giants were spending on infrastructure and erasing billions of major American technology shares such as NVIDIA. These companies have since been widely regained.

As was the case with the appearance of Deepseek R1 for the first time, the model that was also promoted with Little Fanfare was released. It is a thinking model, which means artificial intelligence can carry out more complex tasks through a step -by -step logical thinking.

The DEPSEK R1, which was promoted behind the OPENAI thinking models, is located on LiveCodebeench, a site that enumerates models against various standards.

“The latest upgrade to Deepseek is more intense in thinking, stronger on mathematics and symbols, and closing on top models such as GEMINI and O3,” Adina Yakavo, an Amnesty International researcher at Huging Face, told CNBC.

Yakefu said that the model that has been promoted has “significant improvements in inference and hallucinations,” Yakefu said, adding that “this version explains Deepseek not only to catch up with the knees, it is competing.” Hells refer to artificial intelligence that provides incorrect information.

Deepseek has become a poster on how Chinese artificial intelligence has evolved despite our attempts to restrict the country’s access to chips and other technology. This month, Chinese technology giants Bido and Tinetnet Reveal how they were Make artificial intelligence models more efficient To deal with the export restrictions of the American conductors.

Jensen Huang, CEO of the company NafidiaIt designs the graphics processing units required to train huge artificial intelligence models, He criticized US export controls on Wednesday.

“The United States has relied on its policy on the assumption that China cannot make artificial intelligence chips,” Huang said. “This assumption has always been doubtful, and now it is wrong now.”

“The question is not whether China will get artificial intelligence,” Huang added. “It is indeed.”