@User I do plan to design in moderate inversion. My question for you is, what is the deficit that caused the odd gm/Id curve you plotted? Is it that the model being used in the PDK is poor for the transition from weak to strong inversion? Or is it that the device characterization did not cover that part of the design space, causing some of the model coefficients to be wrong or omitted? Tim Edwards mentioned discontinuities in the current model, that's probably a strong clue.