Will the evaluation section include examples of testing the same agent across local and hosted models?