With paper-and-pencil method I found only a first $5$ cases:
$$1=3^2-2^3$$
$$2=3^3-5^2$$
$$3=2^7-5^3$$
$$4=5^3-11^2$$
$$5=2^5-3^3$$
This looks interesting and if a natural $n$ can be represented as difference of two powers (we do not take here $a^1$ into consideration but only exponents $\geq 2$ and we do not take into consideration powers $1^m$) we can call $n$ a power-representable natural number.
It is very reasonable to expect that some numbers can be represented in more than one way but I would like to know here is it known to be true and is it true a following statement:
Every natural number is power-representable.