Chi square Or Mann U Whitney for comparing student grades

Hi, for my study I have obtained the UK classification grades that are used at University. I've obtained data from two groups of students first generation and continuing generation. I want to see if there is a difference I'n grades between the groups. The problem I have is that I've obtained categorical data on this, so the participants have indicated their grade in terms of classification with the options 1st,2:1,2:2,3rd and not their % which would have been better. My question is if I should convert these grades into % by for example taking the minimum % that is required to receive such a grade (e.g first =70, 2:1 =60) and then compare the means of the groups using Mann U Whitney? Or would it be better to keep the answers as they are and just run a Chi-square test? I've done them both with sig results indicating that there is a sig difference. Just want some input on which would be a better model to use! Thanks!
I would used the Mann Whitney because your data deserves more than a chi-square test. Also a MW test is more often used to compare two samples whereas a chi-square test is called when you want to hypothesis there is some sort of relationship between those variables. But I could be wrong, any comment is welcome.