Good question! The short answer is yes. The size of the context window is based on the model's capabilities. And the models go to their API max (not a lower UI max found on other platforms). Because that can get expensive fast (and it's not always needed) we have a setting