(This is a redraft to be clearer. And also to fix @mathematician123's point about the possibility that $N+1$ is a perfect square.)
I'm going to do this for any large enough number $n$ of perfect squares. And then show that $1000$ is large enough.
Let $g(x) = x^{3/2}$. Then the number of perfect squares below $N^3$ is $\lfloor g(N) \rfloor$. What we want to show is that there exists an interval of length at least $3$, inside of which $n < g'(x) < n + 0.5$. In that interval we will find three consecutive integers $N$, $N+1$, and $N+2$. Of them, either there are $n$ perfect squares between $N^3$ and $(N+1)^3$, or else there must be $n$ perfect squares between $(N+1)^3$ and $(N+2)^3$.
Let's start. $g'(x) = 1.5 \sqrt{x}$. $g'(0) = 0$, $g'(x)$ is increasing, continuous, and goes to infinity. Therefore for any positive integer $n$ there is always a solution to $g'(x) = n$.
$g''(x) = \frac{0.75}{\sqrt{x}}$. This is positive, decreasing, and goes to $0$ in the limit. By inspection, $g'(21) < 7$ and $g''(21) < \frac{1}{6}$. If $7 \le n$ let $x_0$ be the solution to $g'(x_0) = n$. $x_0$ is in the interval $(21, \infty)$ and therefore $0 < g''(x) < \frac{1}{6}$ in the interval $(x, x+3)$. By the intermediate value theorem $g'(x)$ cannot change by more than $3 * \frac{1}{6} = 0.5$, so in that interval $n < g'(x) < n + 0.5$.
Let $N$ be the smallest integer bigger than $x_0$. Then $N$, $N+1$ and $N+2$ are all in the interval $(x_0, x_0+3]$. By the intermediate value theorem, $g(N)+n < g(N+1) < g(N)+n + 0.5$. Likewise $g(N+1)+n < g(N+2) < g(N+1) + n + 0.5$. Therefore the triple $g(N), g(N+1), g(N+2)$ has the form $g(N), g(N) + n + \epsilon_1, g(N) + 2n + \epsilon_2$ where $0 < \epsilon_1 < \epsilon_2 < 1$.
Now there can be at most one integer in the interval $[g(N), g(N) + \epsilon_2]$. If there are none or it occurs after $g(N) + \epsilon_1$, then $\lfloor g(N+1) \rfloor - \lfloor g(N) \rfloor = n$ and those perfect squares are strictly between the cubes. If it occurs before $g(N) + \epsilon_1$ then $\lfloor g(N+2) \rfloor - \lfloor g(N+1) \rfloor = n$ and those perfect squares are strictly between the cubes. If it occurs exactly at $g(N) + \epsilon1$, then $\lfloor g(N+1) \rfloor - \lfloor g(N) \rfloor = n+1$, but one of those perfect squares is at $(N+1)^3$, and so there are exactly $n$ perfect squares strictly between $N^3$ and $(N+1)^3$. In all cases, the result holds.
This shows that if $7 \le n$, there is always an $N$ with $n$ perfect squares between $N^3$ and $(N+1)^3$. Since $7 < 1000$, this solves the problem.
As @TonyK already pointed out in the comments, examples of $N$ can be found for $n = 2, 3, 4, 5, 6, 7$. Therefore this actually holds for $2 \le n$.
This argument can be generalized to looking for counts of perfect $k$ powers between perfect $r$th powers as long as $k < r$. Though, of course, with different lower bounds on $n$.