Efficiently developing effective antipsychotic drugs is one of the most daunting challenges in modern psychiatric drug discovery. Although there has been extensive research in the past 70 years since chlorpromazine was first used to treat schizophrenia, the therapeutic strategies that are available today are still largely dependent on the manipulation of dopamine D₂ receptors, which provides only partial relief of symptoms and has significant side effects. This comprehensive review critically discusses the development, present status and future directions of the preclinical models that are used to test antipsychotic drug properties. We critically review classic pharmacologic paradigms, as well as genetic and neurodevelopmental research strategies, human stem cell-based technology, computational strategies, and novel technologies that are changing the landscape of the field. This enduring disconnect between the drug discovery and drug development preclinical and clinical phases is reviewed in terms of model validity, model predictive value and strategic changes to improve drug development pipelines. We envision a third-generation combination of recent advances in optogenetics, 3D brain organoids, machine learning algorithms, and integration of multi-omics to produce a truly novel approach to antipsychotic screening that could end the therapeutic stagnation associated with schizophrenia.