A model consists in one or more possible worlds. Necessity in a world is determined by its associated set of possible worlds.
I am curious whether there is any work that involves an account of something like "possible models."
So, whereas each possible world in a model makes each statement in the model either true or false, similarly, each possible model in a (supermodel?) varies some property of worlds in the supermodel. For example, each world in a model may have R to each other world in that model. In the supermodel it's part of, perhaps another possible model has some worlds failing to have R to each other.
A gloss on a possible model might be "ways the possibility relation could be."
Is this a concept anyone has written near or about?