The most efficient artificial intelligence (AI) and metaheuristic algorithms have been assessed for the design of nanovector (NV) drug delivery systems. This was done through a thorough search of modeling and optimization algorithms for nanoparticles. Adaptive neuro-fuzzy inference system (ANFIS), a modeling algorithm not currently used for NV drug delivery systems, was compared to the most used algorithm, multilayer perceptron artificial neural network (MLP-ANN), by modeling a dataset of nanovectors. ANFIS provided consistently a better modeling than MLP-ANN for nanovector data regarding particle size, polydispersity index, zeta potential, and drug loading. Similarly, optimization algorithms not currently used for NV drug delivery systems, cuckoo search (CS), symbiotic organism search (SOS), and firefly algorithm (FA), were compared to genetic algorithm (GA), the most used algorithm, by optimizing several benchmark functions. CS and SOS outperformed in function optimization both GA and FA. Comparison among AI or metaheuristic algorithms is scarce. Our results show that current most used NV drug delivery systems AI and metaheuristic algorithms are not the most efficient for modeling and optimization of physicochemical properties of NVs. Implementation of algorithms here suggested, in addition to sample size calculations, should yield better predictions to optimize NV properties.