Abstract: High temperature gas-cooled reactors using helium-xenon gas mixtures as the reactor coolant are the optimal solution for reliable energy under extreme conditions where the compact small reactors are necessary. It is essential for the thermal-hydraulic and safety design of such reactors to fully understand the turbulent heat transfer mechanisms of helium-xenon gas mixtures in the reactor core. The Prandtl number of typical helium-xenon gas mixtures is around 0.2 under normal operating conditions. As low-Prandtl-number fluid, the turbulent heat transfer characteristics of helium-xenon gas mixtures differ significantly from those of the typical coolant of pressurized water reactors because Prandtl number reflects the relative relationship of the development of the velocity and thermal boundary layers. In this paper, a high-fidelity numerical method, namely direct numerical simulation (DNS), was employed to investigate the turbulent heat transfer characteristics of helium-xenon gas mixtures in a round pipe across a wide range of temperature (from 300 to 1 500 K). The DNS method was validated against limited experimental data from literature. High-resolution and high-fidelity numerical data of the flow field, temperature field, and turbulent structures were obtained by DNS and turbulent statistic methods. It is found that the turbulent Prandtl number (Pr_t) of helium-xenon gas mixtures varies significantly across the radial direction in the round pipe, which indicates that the constant Pr_t model assumption which is commonly used in the engineering simulations is invalid and novel Pr_t models need to be developed for better numerical predictions of turbulent heat transfer of helium-xenon gas mixtures. In addition, by comparing the heat transfer coefficient results of DNS and four different conventional empirical heat transfer correlations, it is proved that, under the high temperature conditions, all the empirical correlations have underestimated the heat transfer coefficients with relative deviations exceeding 20%. The main reason for the relative deviation could be that the parameters of the empirical correlations are calibrated based on the experimental data under the low temperature conditions while the physical properties of the helium-xenon gas mixtures vary differently under the high temperature conditions, which resulting in the underestimation of the heat transfer coefficients for the empirical correlations. Among the four empirical correlations, the Pickett correlation performs relatively better and it is selected to be furtherly calibrated based on the DNS data under the high temperature conditions. The modified Pickett correlation can accurately predict the turbulent heat transfer coefficients of helium-xenon gas mixtures under the high temperature conditions, achieving the relative deviations within 3% of DNS results. The present study provides essential theoretical support for thermal-hydraulic design and safety assessment of the high temperature helium-xenon gas-cooled reactor.