arXiv positions itself in a place before the journals in the food chain; anyone can upload a paper there, including authors who submit to journals, but all of these papers are "preliminary", as in, "before journal edits". It's questionable if journal edits actually bring any value, but nonetheless, their status is that they're going to be "likely different in some way" than whatever journals publish.

SciHub allows free access to the "final" pieces of work, even if they're equivalent (or even equal) to the "non-final" ones available on arXiv. That's a big mental difference.