Top: comparison of cumulative template TS values for the eB and Loop I templates across the 64 GALPROP diffuse-emission templates. Each value is computed relative to the corresponding baseline model without the additional large-scale template. The benchmark model is highlighted by the red star, and the navy dashed line indicates equal likelihood improvement for the two non-nested spatial hypotheses. Bottom: SEDs obtained using the 64 GALPROP diffuse-emission templates for the northern (left panel) and southern (right panel) large-scale structures. Dashed lines indicate the ensemble mean, and shaded regions denote the 1σ spread across the diffuse-emission templates. The data points show the SEDs obtained with the benchmark background model, with circles and squares denoting the eB and Loop I templates, respectively.