No, I have decided I don't trust benchmarks without using the model first.