It's true that a model is not necessarily an explanation. Epicycles had predictive value (if tuned well), but almost all agree it's not how planetary motion "works". Newton gravity is considered a better explanation because it's the simplest working model: Occum's Razor.
If a better model comes along, we may make that the top model, ie, the "best explanation so far". Maybe there is no gravity, only something that happens to fit our gravity model. We do more experiments to find that out by testing the model more and testing alternative models.