[Palmik]

I inferred that from the other statement:

> Since DeepSeek-V3 is worse than those US frontier models — let’s say by ~2x on the scaling curve, which I think is quite generous to DeepSeek-V3

He says that it's 2x worse. So if it has the ~same quality [1], it would imply it's 2x larger. Unless I misunderstood what he meant there, of course.

[1]: "DeepSeek produced a model close to the performance of US models" -- in his own words.

[kalkin]

Ah, I read that as him saying the quality isn't as good-equivalent to a model with 50% less compute. But you might be right.