Evaluation of Language Models in the Medical Context Under Resource-Constrained Settings