<p>Hot cracking remains a critical defect limiting alloy processability in powder bed fusion additive manufacturing (PBF-AM). While thermodynamics-based hot-cracking models (HCMs) offer rapid screening capabilities, their accuracy across diverse alloy systems remains unclear. This study systematically evaluated four HCMs—the Clyne–Davies CSC model, HCS model, Kou’s CSI model, and a situational solidification range adjustment model (Δ<i>T</i>)—using a comprehensive dataset of 460 PBF-AM alloys compiled from 128 literature sources. Analyses focused on five primary alloy systems with sufficient sample sizes containing both crack/no-crack data points: aluminum, stainless steels, other steels, nickel-based superalloys, and high-entropy alloys, representing ~ 79% of the total dataset. Scheil–Gulliver solidification simulations were performed using both classic and solute-trapping calculation modes to derive thermodynamic parameters for each model. Model performance was assessed using balanced accuracy to account for crack-class imbalance. Results demonstrated that system-specific models significantly outperformed mixed-alloy approaches, with the CSI model providing optimal predictions for aluminum (70.3% accuracy), stainless steels (87.5% accuracy), and HEAs (62.2% accuracy), while Δ<i>T</i> performed best for other steels (65.3% accuracy) and CSC for Ni-superalloys (57.6% accuracy). Solute-trapping calculations improved predictions only for stainless steels and aluminum alloys. These findings demonstrate that hot-cracking behavior is fundamentally system-dependent and therefore requires tailored model selection based on alloy classification rather than universal application. While this work provides practical guidelines for selecting current thermodynamics-based models, it also highlights the need for developing unified predictive approaches that can overcome these alloy-system boundaries.</p>