Currently the RZ location of the coordiates are randomly selected each time a plasma source is created.
So they will be slightly different each time. This can be reduced by sampling more that 1000 points. Or alternatively you can change the sampling method used one these two lines of code openmc-plasma-source/src/openmc_plasma_source/tokamak_source.py at 64e4aa13f1e81efb8d0e16d7b76acb40bacce468 · fusion-energy/openmc-plasma-source · GitHub
Im actually working on that package at the moment and trying to improve it. This can be one of the improvements to look into. I’m hoping to first improve the energy distribution then move it to using mesh sources (which will fix the reproducibility bug you noticed). Next openmc release supports mesh sources so hopefully a release of openmc-plasma-source will get it fixed just after the next openmc release.