Classification and the probability scores are contradictory

Hi,

I have trained a simple binary classification model and trying to predict unlabeled items in my dataset. My prediction returns the classification and also probability scores (proba_0 and proba_1), but my results are contradictory. Say for example when my classification is 1, my proba_1 must be > 0.5, but its not consistent like below.

Operating system used: GCP