I looked into this package about a year ago and found the same thing. I believe they wrote the documentation intending to allow for this, but haven’t implemented it yet. On GitHub, there are open issues for minimum time problems and hybrid dynamics, either of which should address this. It’s possible this is pretty easy to implement, because continuous-time problems just get discretized. At that conversion, code could be added to allow for unusual discrete dynamics. For example, a discrete model might have trivial dynamics e.g. x[1] = u[0] where x[1] is the final time (used to scale the continuous-time dynamics as others have said above), and u[0] is a control input that acts as the single decision variable for final time. That is trivial, but an exception to the package’s rule of discretizing a continuous-time model.