In Flux v.12.1, the Bertotti model has been modified. Now, it allows computing correctly the iron losses of such kind of machines.
See also the FAQ: Why Flux 12.1 (or higher version) is giving a different result on iron losses computed with Bertotti method than in previous version?