Nervous system cancers are the 10th leading cause of death worldwide, many of which are difficult to diagnose and exhibit varying degrees of treatment resistance. The limitations of existing cancer models, such as patient-derived xenograft (PDX) models and genetically engineered mouse (GEM) models, call for the development of novel preclinical cancer models to more faithfully mimic the patient's cancer and offer additional insights. Recent advances in human stem cell biology, organoid, and genome-editing techniques allow us to model nervous system tumors in three types of next-generation tumor models: cell-of-origin models, tumor organoids, and 3D multicellular coculture models. In this review, we introduced and compared different human stem cell/organoid-derived models, and comprehensively summarized and discussed the recently developed models for various primary tumors in the central and peripheral nervous systems, including glioblastoma (GBM), H3K27M-mutant Diffuse Midline Glioma (DMG) and H3G34R-mutant High-grade Glioma (HGG), Low-grade Glioma (LGG), Neurofibromatosis Type 1 (NF1), Neurofibromatosis Type 2 (NF2), Medulloblastoma (MB), Atypical Teratoid/rhabdoid Tumor (AT/RT), and meningioma. We further compared these models with PDX and GEM models, and discussed the opportunities and challenges of precision nervous cancer modeling with human stem cells and organoids.