Accurate management of powder bed temperature is essential in binder jetting (BJ) to achieve dimensional accuracy and adequate mechanical properties. Three different models using the finite element method operating at different scales were developed in this study to compare their accuracy in predicting the thermal history of the powder bed during the BJ process. The simulated temperatures were compared with in situ experimental thermal measurements of the powder bed during printing. The first model relied on 2D Gaussian heat sources to model the movement of infrared lamps, achieving an absolute average error of just 1.5°C with the experimental data, but taking 28 h to simulate only 20 layers using eight CPUs. The second model employed a layer heating (LH) approach to reduce computation time while maintaining accuracy similar to that of the previous model. The second model was able to simulate 200 layers in 33 h with an average error of 1.7°C in comparison with the thermal measurements. The third model combines the LH and lumping approach to further reduce the model computational time, enabling simulation of the full process (2000 layers) in 33 h with an average error of 2.3°C compared to the experimental case. This parametric study suggests that thermal management of the powder bed during BJ can be improved using a combination of infrared lamps above the powder bed and a heated build box. This avoids bleeding, interlayer stitching issues, and other detrimental phenomena regardless of the nesting configuration.