It might be more ... useful to think of models, understanding, and/or knowledge (largely interchangeable terms) as having use value rather than truth value.

That is, good models are useful, in some way.

Usefulness, like value, is itself relative and depends on intent, goals, and starting positions.

That said, there are models with greater or usefulness over a wide range of circumstances. Holding to a model which is demonstrably inferior in this regard might well be considered "wrong", in this sense.

Absolutely. This is exactly what I do, actually. I argue from the perspective of many models at once, assign weights and priorities and then arrive at some conclusion with an estimated probability of effectiveness or usefulness depending on use case. Truth simply is and is also unknowable at the same time.

Personally, I would only make one modification here for my own purposes:

> Holding to a model which is demonstrably inferior in this regard might well be considered "wrong" to me.

I know what is right and wrong for me and to me, but stop short of extending that judgement outside my person. That's just me though.