Imagine you are applying a decision tree based model to a numerical dataset and you run a experiment where you scale the features. What model would give the best results the original or the scaleted one?

They should give the same result because tree-based models are scale invariant.