Chinese ChatGPT rival DeepSeek claims it spent just $294,000 training its R1 foundational AI model.

The announcement was made in a peer-reviewed article in the academic journal Nature, which listed DeepSeek founder Liang Wenfeng as one of the co-authors.

The United States has expressed doubt about DeepSeek’s past claims that it can train models at a far lower cost. Sam Altman, CEO of ChatGPT maker OpenAI, said in 2023 the training of foundational models had cost “much more” than $100 million, as reported by Reuters.

DeepSeek has openly admitted using data trained by rival models to train its own, although it says this is done indirectly.