Here is the conventional use of Big O notation where the input is a number:

https://en.wikipedia.org/wiki/Miller%E2%80%93Rabin_primality...

As you'll see they are using the number of digits, not the value of the number, because the number of digits is the size of the input.

You can find countless examples of this in literature. So this is not a 'personally-interesting' way as you assert.