By Ross Douthat,Sophia Alvarez Boyd
Publication Date: 2026-05-14 09:05:00
Chan: Overall, I think the consensus is that Chinese models are between three, six and nine months behind, depending on the time of year and which was the newest model that just came out.
This means that the Chinese models released today are slowly approaching the American models released a few months ago when looking at certain benchmarks and specific evaluations to understand how well they perform, for example, on math or coding tasks or even on new agent tasks. This is what this lead time means.
The fact is that it’s not just about having the absolute most modern model. You can have very, very powerful models that can do a lot, that can do a lot of useful agent tasks, like creating a whole PowerPoint presentation for you and all the research and analysis that goes into that, or answering your emails.
I think there’s this strategy in China right now where they hope it’s not just about…
