[tripletao]

"Running viruses through humanized mouse models" is a pretty normal (though frightening) part of virology. For example, Ralph Baric was doing it back in 2005:

https://www.pnas.org/content/102/23/8073

So if the Chinese military had in fact been doing this, I'd guess it was just basic research, in the same way that lots of American basic research links back to DARPA. Of course they fund it because they believe there might be a military application, but I see no reason to think that application would be bioweapons (vs. the same kind of beneficial applications described in the open literature).

[Throwawayaerlei]

Offensive bioweapons researchers don't publish their results in scientific journals.

[tripletao]

Perhaps "no evidence" would have been better phrasing than "no reason"? I do think it's possible that some Chinese (or American, or British, or ...) military officer has at some point wondered if coronaviruses would make good bioweapons, but there's still no evidence.

They don't seem like obvious candidates to me, though. Both SARS v1 and SARS-CoV-2 show unpredictable, stochastic person-to-person spread, via super-spreader events. For a bioweapon that would ideally infect all the enemy but no one else, that's the last thing you want, hard to reliably get started and hard to reliably stop once it starts. So that reinforces my belief that if SARS-CoV-2 was of unnatural origin, it was almost certainly an accident during basic research.