The adsorption and diffusion behavior of shale oil in kerogen nanopores is a key issue for optimizing CO2 oil displacement efficiency. Especially temperature has differential regulation on the competitive adsorption of light/heavy components. The study used the molecular dynamics (MD) simulation method to investigate the adsorption and diffusion mechanisms of single-component alkanes (C7H16, C12H26, C18H38) and multi-component on the surface of kerogen under temperature gradient (303-403 K), the synergistic effects of multi-component systems under real reservoir conditions, and nuclear magnetic resonance (NMR) experiments to validate pore-scale displacement under temperature gradient. The results of MD showed that at low temperatures, long-chain alkanes formed a “crystal-like” ordered adsorption layer. The results of the multi-component system show that the “size complementary filling” of alkanes with different chain lengths increases the mass density of the mixed system by 20%-40% compared to the single-component, but intermolecular entanglement leads to a decrease in the diffusion coefficient of about 80%. NMR experiments confirmed that recovery shows a strong positive correlation with MD-derived diffusion coefficient. Moreover, high temperature (343 K) activates oil in small pores (less than 25 nm) previously constrained by strong adsorption-consistent with MD’s “temperature-dissociated adsorption layers” conclusion. This study established a quantitative correlation between temperature-chain length-displacement efficiency, differentiated the temperature sensitivity differences between light and heavy oil reservoirs and proposed a differentiated CO2 displacement temperature strategy for light oil reservoirs (383-403 K) and heavy oil reservoirs (363 K), providing theoretically (MD) and experimentally (NMR) consistent support for CO2 displacement in shale oil reservoirs.