You built the same model in two different programs and got two very different results. How do you know which model is closer to the "truth" and which model has the error in it?

I recommend you build a much simpler model that has a known solution in each program and see if you can get the same answer on the simple model first.