In coupled climate models, land-surface models (LSMs) simulate the exchange of carbon, water and energy fluxes between the terrestrial biosphere and the atmosphere. LSMs typically employ empirical soil moisture stress factors to reduce vegetation function as water availability declines. However, these functions have been pinpointed as a key source of uncertainty, varying widely across LSMs. Multiple alternative models, which hypothesise that plants optimally regulate their photosynthetic and hydraulic functions, have recently been proposed as solutions to reduce existing model biases. Yet, to date, a systematic inter-model evaluation is lacking (i.e. inter-model comparison is needed to understand how different mechanistic assumptions across these optimal models affect plant behaviour). Here, we asked how, and under what conditions, five of these optimal models differ from one another, from a reference stomatal model, and from a reference model of plant hydraulics. All the models were trained to match under average conditions, before being subjected to (i) well-watered conditions, (ii) a dry-down, and (iii) high vapour pressure deficit. We therefore identified the models’ specific responses and sensitivities to climate extremes. To further assess whether model-specific responses were realistic, we tested them against photosynthetic and hydraulic field data measured along mesic-xeric gradients in Europe and Australia. Finally, we evaluated model performance versus expected model performance — given model complexity and the amount of information taken in by each model — which enables us to make recommendations regarding the use of these schemes in LSMs.