Why class weight is outperforming oversampling?

Question

I am applying both class_weight and oversampling (SMOTE) techniques on a multiclass classification problem and getting better results when using the class_weight technique. Could someone please explain what could be the cause of this difference?