First, getting the best performance is not the ultimate goal of programming. Thus, even if option B was always faster and consumed less memory than A, this does not mean that it is always the best option if it is more complex. More complex code takes longer to write, more difficult to understand, and most likely contains errors. So, if the simpler but less efficient option A is good enough for you, then this is the best choice.
Now, if you want to compare the AVL tree with a simple binary search (BST) tree without balancing, then the AVL will consume more memory (each node must remember its balance ratio), and each operation can be slower (because you need to maintain the balance ratio and perform rotations sometimes).
, BST () . , , , , BST , AVL.