OpenAI CEO Sam Altman said on Thursday that the artificial intelligence model released by Chiense technology startup DeepSeek has been largely overhyped.
Altman contended at an event in Washington, D.C., that DeepSeek’s recently announced R1 model, which purportedly cost a mere $5.6 million to train, does a “’couple of nice things’ but has been ‘wildly overstated,’” according to The Hill.
“Again, this is a model at a capability level that we had quite some time ago,” Altman asserted.
“We’ve got a sense of what it takes to run a cost-efficient model,” he added.
“So again, I don’t want to diminish the work. I think this is a great work, and it’s, in some sense, it’s like the first surprise or new competitor to OpenAI in quite a while.”
The reported cost of $5.6 million caused panic in the technology sector as investors questioned the assumption that training artificial intelligence models was extraordinarily expensive.
ChatGPT-4 from OpenAI had a technical creation cost between $41 million and $78 million, according to an analysis published last year by Forbes.
Altman previously said the model cost over $100 million to train.
Could this have been a strategic attempt to make Americans fear Chinese AI, thereby creating a greater mandate to funnel money into American AI?
Gemini from Google reportedly cost between $30 million and $191 million before considering staff salaries.
The unveiling of DeepSeek prompted a massive stock market selloff for artificial intelligence firms like Nvidia.
Altman meanwhile claimed that massive spending would still be required to advance the technology.
“The cost to do a frontier model, to be at the very edge, that goes up — the cost of training, compute, test time, compute, data, whatever you want,” he said, per The Hill.
“The amount of demand there is to use a given level of AI, as you bring that price down, goes up by much more than the decrease in price,” the entrepreneur added.
OpenAI previously said that the company would examine whether DeepSeek “improperly distilled” data from its models, meaning that the Chinese company could have transferred the knowledge of a large model to a smaller version, according to The Hill.
OpenAI forbids distillation to build competing models in its terms of service.
“People will distill our models,” Altman said. “Whatever, that’s fine.”
The consensus that massive costs are required to advance artificial intelligence was demonstrated by the $500 billion “investment in AI infrastructure” recently announced by Oracle, SoftBank, and OpenAI, and unveiled at the White House by President Donald Trump.
No funding for the deal will come from the federal government, but Trump signed an executive order revoking a previous action from now-former President Joe Biden that “hampered the private sector’s ability to innovate in AI by imposing government control over AI development and deployment,” according to a fact sheet from the White House.
The new order also “calls for departments and agencies to revise or rescind all policies, directives, regulations, orders, and other actions taken under the Biden AI order that are inconsistent with enhancing America’s leadership in AI.”
Advertise with The Western Journal and reach millions of highly engaged readers, while supporting our work. Advertise Today.