Someone had this happen for a 15MB download, wich makes me think, this is quite chance based. But larger files will of course be affected more often.
It is interesting that you could replicate it with exact number. When I had it happen, the file was halted about at the same size, but not exactly. Nothing had changed in between that I was aware of.
If clearing cache changed you to a new number, the second try without clearing cache sounds like a ... well, caching issue. Since the issues seems to be, that the file host delivers the file incomplete but claims to be done, this would fit. It was delivered incomplete to your personal caching system, wherever that might be, including your browser's cache, wich would be the simplest answer. But could also be a proxy.