How to evaluate one step ahead prediction using the new ARIMA class in Matlab 2012a?

I used the new ARIMA class in Matlab to apply a multiplicative seasonal arima model on my data. Although I have estimated a model already, but I now I want to apply that model on test data in order to have a one step ahead prediction at each time step and then compare the results of predicted and original series. The 'forecast' method only forecasts x values in the future. The old 'predict' method also does not work with this new ARIMA class.

Any ideas how to do this one step ahead prediction or just how to apply the model estimated to a test data for predicting the values in the time range same as the test data?

[James Howard]

I have the same question. I would like to perform K steps ahead prediction on a test dataset. Is this doable with matlab 2012a or b?