I remember that me and jack few years ago solved a stochastich growth model using the value function iteration. Onestly, i have no clue where to find that script. Maybe jack can help.
Instead, if you are referring to a traditional log linearized model i have somewhere (i Need to verify) a very simple script that solve an rbc model.
However, if i remember correctly gretl Is not able to handle complex numbers in the QZ decompostion which Is essential for solving most of the forward looking model. It was a long time ago, maybe in the meantime things changed. Maybe the community can help.
Best
Federico