I'd say its subtle. The instrumentalist just says that measurements are reality period and that there isn't anything to say about what goes on between measurements. You can sort of see this at an extreme in QBism. You just write down the quantum Bayes rule and that is all you commit to saying about the universe.

I'd argue that what physics finds are (possibly non-unique) mathematical structures which correlate to the physical world. Because they are non-unique it is, in my opinion, silly to say things like "spacetime is a 4d manifold of Lorentzian signature with a connection like so and so, blah blah." There are other ways to describe the same dynamics which offer a different ontology and so we shouldn't just assume the terms in our physical models correspond to things in the universe.

I think we reasonably can say "there exists a thing in the universe or a part of the universe that has the property that we can describe it with a 4d manifold of Lorentzian signature and so on." So physics does identify properties that hold for parts of the universe on certain time, length, action scales. But I think the general vibe of physics in the 20th and 21st centuries is that we don't know what those things are beyond the properties we give them. Maybe that is what knowing what something is is?