Nothing incredibly good, neither really bad, I guess they used an old lady to show more of the blendshapes that can be quite tricky because of the pronounced cheeks, and they're trying pretty "extreme" versions.
They don't really need expensive mocap to do a good job though
https://www.youtube.com/watch?v=XiCazsg4X2s
This is probably automated blendshapes from text/speech, in cutscenes they'll probably animate them separatedly.
It looks ok, not really AAA standard, but good enough.
EDIT: TBH I prefer the second character, but man, it kinda clash quite a bit with the others characters we've seen, weird that they juxtaposed those just like that.