DeepSeek aims to launch new AI model soon
Chinese startup DeepSeek is ramping up efforts to deploy a new artificial intelligence model (Ia), following the global outcry over its predecessor R1 last month, Hong Kong-based newspaper The Standard reported, citing three people familiar with the company's plans, founded by Liang Wenfeng, Agenzia Nova reports.
DeepSeek reportedly planned to release the R2 model in early May, but then accelerated its development plans to “get it out as soon as possible.” The company is aiming to provide better coding and reasoning in languages other than English.
On January 20, DeepSeek unveiled an advanced AI model called R1, which – according to the company itself – is capable of “competing with the o1 model” developed by US rival OpenAI. The main cause of discussion was the competitive cost of the new virtual assistant from the Chinese startup, which presented it as 20 to 50 times cheaper than OpenAI’s o1 model. Liang Wenfeng, founder of the Hangzhou-based startup, was among the participants in the meeting held on February 17 by President Xi Jinping with the main names in the national technology industry, with the aim of giving new impetus to the development of the private economy in the context of technological rivalry with the United States.
Despite its global success, DeepSeek’s R1 virtual assistant has not failed to raise widespread security concerns. In Italy, the model was blocked on January 30 by the Data Protection Authority, as well as in government departments in Taiwan and South Korea. At the same time, Australian Treasurer Jim Chalmers has urged citizens to be cautious when using the Chinese AI model, and U.S. officials are also examining DeepSeek’s national security implications.