ChimeraX's PDB reader has no ability to skip MODELs in a multi-MODEL file.  So you would have to read the entire file into ChimeraX, but then with the help of a small bit of Python code you could write out your reduced PDB file.  The procedure would be:

1) Open the 3000-MODEL PDB file.

2) Run the following Python code:

s = session.models[0]

import numpy

s.add_coordsets(numpy.array([s.coordset(i).xyzs for i in s.coordset_ids[::5]]))

You would put the above code (but not indented) into a file with a ".py" suffix and open it in ChimeraX to run it.

3) Use the "save" command (with "allCoordsets true") to save the resulting 600-MODEL PDB file.

For convenience, I've attached a file with the Python code in it.