This document presents a prediction model for the yield of wheat, rice, and corn crops, which are the ones that provide the most significant source of protein consumption in the world. This model is the result of the analysis, application, and computation of several different proposals that combine a fuzzy Mamdani and Sugeno inference system, along with a deterministic (Quasi-Newton algorithm) and meta-heuristic (genetic algorithm) optimization techniques. This combination generates, respectively, four different variants of each crop’s model, obtaining a total of twelve alternatives. The applied methodology for the establishment of the model is based upon the analysis of a set of inputs and outputs, characterized by biomass, solar radiation, rain, extractable water fractions, and yield linguistic variables; and is also based upon construction of inference rules, membership functions configuration, and fuzzy sets optimization. This model returned an efficacy index of over 94% for each crop; determination coefficient was found under a 0.90 threshold in each case. This results represent great reliability on the proposed method, offering greater precision in the calculation without compromising interpretability in the yield variable. This approach presents itself as an alternative for known food production solutions, all while it provides a mechanism so that an estimation of the final product of sowings in the agricultural field can be made.