Besides all the pros already mentioned, there is a high chance other researchers will use your code and cite you. If they have to compare their results with some previous state of the art, it will be the one with available code. The whole thing about “the paper is enough to reproduce” never happens, ever.