For classification algorithms modelling using classification learner app, which among the two model diagnostics is most important for selection of model - AUC or Accuracy?

I am using classification learner app for modelling data. I came accross two algorithms (using the same number of features) with the following results.

Gaussian Naive Bayes - Accuracy 89.3; AUC 0.78

Fine Gaussian SVM - Accuracy 91.3; AUC 0.75

Which among the two is best. Among the model diagnostics, more weightage should be given to Accuracy or AUC?