Glioblastoma (GBM), the most malignant brain cancer, remains deadly despite wide-margin surgical resection and concurrent chemo- & radiation therapies. Two pathological hallmarks of GBM are diffusive invasion along brain vasculature, and presence of therapy-resistant tumor initiating stem cells. Deconstructing the underlying mechanisms of GBM-vascular interaction may add a new therapeutic direction to curtail GBM progression. However, the lack of proper 3D models that recapitulate GBM hallmarks restricts investigating cell-cell/cell-molecular interactions in tumor microenvironments. In this study, we created GBM-vascular niche models through 3D bioprinting containing patient-derived glioma stem cells (GSCs), human brain microvascular endothelial cells (hBMVECs) cells, pericytes, astrocytes and various hydrogels to model glioma/endothelial cell-cell interactions in 3D. Three GBM-vascular models were designed: Model A with large vessels and GBM spheroid; Model B with large- and micro-vessels, and GBM spheroid; Model C with large- and micro-vessels and dispersed GBM cells. Large channels were created by sacrificial 3D bioprinting. Microvessel network was formed through self-assembly of ECs and mural cells (fibroblast, pericytes, and/or astrocytes). Three GBM cell types were used in the study: SD02 and SD03 are GSCs; U87MG is a commercially-available GBM cell line. GSCs cultured in these models maintained stemness and heterogeneity during the long-term cultures. In Model A, GSCs actively invaded into the surrounding tissues (~Day26), initially regressed in response to the drug (~Day50), then developed therapeutic resistance and resumed aggressive invasion (~Day57). In Model B and C, three GBM types presented distinctive invasion patterns and EC-interactions. SD02 cells showed a spiky invasion pattern with elongated morphology. SD03 cells showed a more dispersed invasion pattern with many single cell migrations towards surrounding microvessels. U87MG cells showed a blunt invasion pattern, caused EC death in the spheroid form. In summary we have created GBM-vascular niche models that can recapitulate various GBM characteristics such as cancer stemness, tumor type-specific invasion patterns, and drug responses with therapeutic resistance. Our models have a great potential in investigating patient-specific tumor behaviors under chemo-/radio-therapy conditions and consequentially helping to tailor personalized treatment strategy. The model platform is capable of modifying multiples variables including ECMs, cell types, vascular structures, and dynamic culture condition. Thus, it can be adapted to other biological systems and serve as a valuable tool for generating customized tumor microenvironments.