AbstractCharacteristics of the atmospheric kinetic energy (KE) spectrum follow a distinct power‐law dependence in the synoptic (k−3) and mesoscales (k−5/3), where k is the horizontal wave‐number. The representation of this canonical spectral behaviour provides a testimony of the acceptability of the design, configuration and performance of a numerical weather prediction (NWP) model. Therefore, the characteristics of the KE spectrum are studied using an operational version of the high‐resolution National Centre for Medium Range Weather Forecasting (NCMRWF) Unified Model (NCUM) system considering both global (∼12 km mesh) and regional (∼4 km mesh) configurations during the northern hemispheric summer period (June–August 2020). The synoptic‐scale spectral characteristics of the NCUM global (NCUM‐G) model match well with the observed spectrum in the tropics, midlatitudes and polar regions. However, the regional NCUM model (NCUM‐R) outperforms the NCUM‐G in representing the canonical slope at mesoscales. As the NCMRWF is accelerating towards a high‐resolution 6 km global (NCUM‐G6) operational model, the spectral characteristics from NCUM‐G6 are also studied during the monsoon period. Though the representation of the mesoscale spectrum is better in NCUM‐G6 relative to NCUM‐G, the regional model surpasses both the global models while demonstrating the mesoscale characteristics close to the observed spectral slopes. Further, the mesoscale spectrum deviates gradually from the Lindborg (1999) analytical fit in the mesoscales around 20Δx (where Δx ∼ 4 km), which specifies an effective resolution of the regional model. Hence, the regional model needs refinements optimally to further reduce the energy dissipation at high wave‐numbers to improve the mesoscale features at the limits of its grid resolution.