I aborted GAP after some hours. I wanted to approve my conjecture that $gnu(n)<n$ for all cubefree numbers $n>1$, where $gnu(n)$ is the number of groups of order $n$, but the case $p^2\times q^2\times r^2$ seems to be already complicated.
Does anyone know whether $gnu(815,409)<815,409$ holds, or even the value of $gnu(815,409)$ ?
Short answer: $415$.
In more details: for cube-free orders, one should use the Cubefree package by Heiko Dietrich. It takes about a minute to get the answer, which is $415$, using the beta-version of GAP 4.8 (namely, GAP 4.8.1) with most recent GrpConst and Cubefree: