As a carcinogenic and highly neurotoxic hazardous gas, benzene vapor is particularly difficult to be distinguished in BTEX (benzene, toluene, ethylbenzene, xylene) atmosphere and be detected in low concentrations due to its chemical inertness. Herein, we develop a depth-related pore structure in Cu− TCPP−Cu to thermodynamically and kinetically enhance the adsorption of benzene vapor and realize the detection of ultralow-temperature benzene gas. We find that the in-plane π electronic nature and proper pore sizes in Cu−TCPP−Cu can selectively induce the adsorption and diffusion of BTEX. Interestingly, the theoretical calculations (including density functional theory (DFT) and grand canonical Monte Carlo (GCMC) simulations) exhibit that benzene molecules are preferred to adsorb and array as a consecutive arrangement mode in the Cu− TCPP−Cu pore, while the TEX (toluene, ethylbenzene, xylene) dominate the jumping arrangement model. The differences in distribution behaviors can allow adsorption and diffusion of more benzene molecules within limited room. Furthermore, the optimal pore-depth range (60−65 nm) of Cu−TCPP−Cu allows more exposure of active sites and hinders the gas-blocking process. The optimized sensor exhibits ultrahigh sensitivity to benzene vapor (155 Hz/μg@1 ppm), fast response time (less than 10 s), extremely low limit of detection (65 ppb), and excellent selectivity (83%). Our research thus provides a fundamental understanding to design and optimize two-dimensional metal−organic framework (MOF)-based gas sensors.