The difference in how well a model fits differing datasets is called variance.

References⌗