Amodei said he thinks a fair statement is that DeepSeek produced a model close in performance to US models 7-10 months older, at a good deal less cost but not near the ratios people have suggested.
Resolved true The record shows it happened or was so.
Factual: Statements about the present or the past, checked against the record: did they hold up?
What would count as true or false
The proposition is that Amodei made this statement, so it resolves on whether a record of him saying it exists. The recorded date (2022-10-13) is metadata, not part of the proposition. True: a verifiable statement or writing by Amodei says, in substance, that a fair statement is that DeepSeek produced a model close in performance to US models 7-10 months older, at a good deal less cost but not near the cost ratios people have suggested. Mostly true: he made this statement with minor wording differences (e.g. a slightly different month range). Mixed: he made a related statement that leaves out or changes one of the key elements (the 7-10 month gap, the lower cost, or the rejection of the claimed ratios). Mostly false: he discussed DeepSeek's cost or performance, but in substantially different terms. False: no such statement is attributable to him, or he said the opposite.
Set before the research began.
Evidence (2)
-
E1 Anthropic CEO downplays DeepSeek models, calls for more Chinese chip sanctions DatacenterDynamics · 2 November 2026
“I think a fair statement is 'DeepSeek produced a model close to the performance of US models seven-10 months older, for a good deal less cost (but not anywhere near the ratios people have suggested).'”
-
E2 Anthropic's CEO says DeepSeek shows US export rules are working | TechCrunch TechCrunch · 29 January 2025
““DeepSeek produced a model close to the performance of U.S. models 7-10 months older, for a good deal less cost (but not anywhere near the ratios people have suggested),” Amodei said.”
Reasoning
TechCrunch [E2], in a 2025-01-29 report, quotes Amodei saying DeepSeek produced a model close to the performance of US models 7-10 months older, for a good deal less cost but not anywhere near the ratios people have suggested. DatacenterDynamics [E1] quotes the same sentence and its "I think a fair statement is" framing. Both are independent sources. Under the criteria, the 2022-10-13 recorded date is only metadata, so the claim depends on whether a record of the statement exists, and these two reports show that it does.
Cite this claim
Dario Amodei, “Amodei said he thinks a fair statement is that DeepSeek produced a model close in performance to US models 7-10 months older, at a good deal less cost but not near the ratios people have suggested.” (said 13 October 2022). Verdict: resolved true. Reliability API, claim c_d9e88382099c, method 0.1.0+c876030c, dataset release v1.1.2. https://epistacklabs.com/claims/c_d9e88382099c
This claim is in dataset release v1.1.2, with the same verdict.
Reply to this claim
More factual claims by Dario Amodei
-
Mostly true
“Because the value of having a more intelligent system is so high, this shifting of the curve typically causes companies to spend more, not less, on training models”
-
Mixed
“Human economic planners made the decisions that led to this success, not by directly controlling entire economies but by pulling a few key levers”
-
Mostly true
“Diseases have been eradicated and many countries have gone from poor to rich, and it is clear that the decisions involved in these tasks exhibit high returns to intelligence”
-
Mostly true
“biomedicine is unique in that although the process of developing drugs is overly cumbersome, once developed they generally are successfully deployed and used.”
-
Mixed
“It’s easy to dismiss this as a one-off or curiosity, but we’ve repeatedly seen a pattern where AI performance in new intellectual domains goes from weak to superhuman in a matter of a few years.”
Verdicts are made automatically; how verdicts are made.