Fluorescence image-guided surgery (FIGS) offers high spatial resolution and real-time feedback but is limited by shallow tissue penetration and autofluorescence from current clinically approved fluorophores. The near-infrared (NIR) spectrum, specifically the NIR-I (700-900 nm) and NIR-II (950-1700 nm), addresses these limitations with deeper tissue penetration and improved signal-to-noise ratios. However, biological barriers and suboptimal optical performance under surgical conditions have hindered the clinical translation of NIR-I/II nanoprobes. In vivo mouse models have shown promise, but these models do not replicate the complex optical scenarios encountered during real-world surgeries. Existing tissue-mimicking phantoms used to evaluate NIR-I/II imaging systems are useful but fall short when assessing nanoprobes in surgical environments. These phantoms often fail to replicate the tumor microenvironment, limiting their predictive assessment. To overcome these challenges, we propose developing tumor-mimicking phantom models (TMPs) that integrate key tumor features, such as tunable tumor cell densities, in vivo-like nanoparticle concentrations, biologically relevant factors (pH, enzymes), replicate light absorption components (hemoglobin), and light scattering components (intralipid). These TMPs enable more clinically relevant assessments of NIR-I/II nanoprobes, including optical tissue penetration profiling, tumor margin delineation, and ex vivo thoracic surgery on porcine lungs. The components of TMPs can be further modulated to closely match the optical profiles of in vivo and ex vivo tumors. Additionally, 3D bioprinting technology facilitates a high-throughput platform for screening nanoprobes under realistic conditions. This approach will identify high-performing NIR-I/II probes with superior surgical utility, bridging the gap between preclinical findings and clinical applications, and ensuring results extend beyond traditional in vivo mouse studies.