This contribution reports the results of Core Experiment 2.5.2 (CE2.5.2) on non-local mean (NLM) in-loop filters. Compared with VTM-1.0, the proposed method reportedly achieves -0.63%, -1.24%, and -1.00% luma BD-rates for the AI, RA, and LB, respectively, with 135%, 111%, and 79% decoding time increases. Compared with BMS-1.0, the proposed method reportedly achieves -0.31%, -0.57%, and -0.76% luma BD-rates for AI, RA, and LB, respectively, with 67%, 39%, and 38% decoding time increases. No coding gain shift between luma and chroma is applied, and the chroma BD-rate savings are approximately 1.5%, 3.5%, and 4.0% for AI, RA, and LB, respectively. The encoding time increases are smaller than 1%.