You’re likely using different Monte Caro variates (rand and randn). i.e., different problem instances.
The number of Constraints, Cones, and Scalar variables reported by Mosek changed each time I ran different random numbers.
Thank you Mr. Stone for your information.