Note that what makes this notable is that the time bound is proven (according to the paper, anyway). N^1/5 is much slower than we normally think of GNFS as being, but the thing to note is that GNFS isn't actually proven to run that fast.
However, those algorithms are _probabilistic_, not _deterministic_, which is what sets this linked method apart.
This algorithm, much like Pollard-Strassen before it, is pretty much irrelevant to integer factorization in practice. But it is nevertheless remarkable to see some progress on deterministic factorization after such a long time.