EQ the difference between 2 and 3 inquiries is probably enough right there frankly; I know there was a breakpoint for me there, so not surprised by that.
The EX one though, the fact your EX FICO 2 is absolutely identical and your FICO 8 is not, suggests the AU might not be counted for you.
Now why you have a score if that tradeline is discounted I don't know, possible we don't quite understand how that works... but what I probably would do in the name of science would be pay off the Discover, let the reports update and keep the balance on the OSCU card basically the same, and then repull and look for score shifts.
I suspect though your files are both so new that changes might get lost in the noise... I wouldn't worry about it really, but as stated it's probably the AU.
AU's generally count the same in FICO 04 and before, FICO 8 has an anti-abuse algorithm in it and none of us really know for any given set of files if it gets counted or not when it comes to AU's... we just don't know what they're pattern matching on.
We had thought they either counted or didn't, and usually the way to try to sort that was with number of tradelines with balances (vis a vis 1 balance, 2 $0, with the balance being the AU, and then 1 balance, 2 $0 with the balance being a non-AU card and see if the scores are the same or different) but maybe you can still generate a FICO score with an AU but not have it counted for revolving utillization? AU's can be a bit strange and FICO 8 made them well stranger.
It won't matter long term as I suspect you both will probably thicken your file some, but it's interesting given that your files are basically identical on EX to see the score difference. I can't recall a similar data point honestly as comparing between users is nearly a non-starter but with everything either joint or AU, you're sort of an ideal test case haha.