But it was on the NatGeo channel so it must be true
For almost all types of hacks, you generally don't need more than 100 tools, as a lot of tools are multi-purpose tools.
Anyway, if they count in all the tools from the beginning till now, including various stealers, bots, viruses, trojans, etc., then 68k is actually wrong, but without viruses, worms, trojans, etc. and only manual tools (+ automated scanners), 68k could be right if you as previously mentioned include stealers, bots, rats, and of course, copies of these that goes under other names but are actually 100% the same version.
If you count an exploit, for a tool, which is generally is as it's a piece of code that you can run, and use as a tool to e.g., exploit an ftp service and thereby gaining access to the actual server, then 68k may sound more reasonable.
It really depends on how you look at it, because anything from a script to a stealer can almost be classified as a tool somehow, thus 68k isn't really that much.
If you count in all the variants of trojans, etc. then you could probably say 300k or even more (probably a million?)
No idea, I think it's either a random number they came up with that sounds cool or from an official report that isn't accurate