Photochemical mechanisms are core components of air quality models, yet differences among them introduce substantial uncertainty in ozone (O₃) simulations. Here, we systematically evaluate three widely used mechanisms (CB06, SAPRC07, and RACM2) using the CMAQ model during an O₃ pollution episode in Chengdu, China. To ensure consistency, a customized volatile organic compounds emission inventory was developed for RACM2. The results show pronounced inter-mechanistic differences in simulated O₃, radical chemistry, and precursor sensitivities under distinct NO_x conditions. SAPRC07 and RACM2 produced higher and comparable O₃ levels (~95 ppbv) with better performance in urban areas, whereas CB06 simulated lower O₃ (~80 ppbv) and performed better under low-NOₓ conditions. These differences are linked to variations in radical concentrations and reaction pathways, with RACM2 yielding the highest OH levels, SAPRC07 better capturing HO₂, and CB06 simulating lower RO₂. Despite these discrepancies, all mechanisms consistently reproduced the dominant O₃ formation pathway, with RO₂+NO contributing over 50 % of net production. However, radical sources and termination pathways differed substantially between urban and suburban environments, reflecting regime-dependent chemistry. Sensitivity analysis further shows that anthropogenic aromatics and alkenes dominate urban O₃ formation, whereas biogenic emissions dominate in suburban areas. Notably, O₃ sensitivity to biogenic emissions is higher in urban high-NO_x conditions, highlighting an amplified biogenic contribution. These findings demonstrate that mechanism choice fundamentally affects O₃ formation and precursor sensitivities, emphasizing the need for multi-mechanism approaches to improve model reliability and support effective emission control strategies.