Just Chi-square and you need to talk about why ANOVA is not appropriate even though GPA feels like ratio-level data (where you would want to compare 2.8 vs 3.1 vs 3.5 for example). Why? GPA is not ratio or interval data but categorial so we use Chi-square.