Really? They both have anycast pops all over the world - and in my tests, very close performance (couple of ms of difference - if that) to be felt by anyone.
You do realise your anecdotal "brief testing" (with n=1) in no way is telling us anything not in the least because you keep your location and network concealed? What might be better in your for us unknown test case scenario might be different in other scenarios.
Again, if that's the case, you should open a trouble ticket, so the problem can be fixed. It won't get fixed if there isn't a work ticket and nobody knows where your traffic is coming from or going to.
ex: https://medium.com/@nykolas.z/dns-resolvers-performance-comp...