If you want it to read letters, all you have to do is make your tokens be letters. That's easier than normal tokenization.
If it was top priority, every company that can't find a post training fix would go disable half their tokenizer code and it would be solved in the next model.