Against the backdrop of climate change， meteorological disasters such as high temperatures have become increasingly frequent， attracting great attention from all sectors of society. This paper constructs a Dynamic Stochastic General Equilibrium （DSGE） model to quantify the long-term economic impacts of climate risks and assess the economic losses caused by extreme high-temperature disasters with “fat-tailed” characteristics. By coupling four modules—“carbon emission module， temperature change module， economic loss module， and macroeconomic module”—the probability of disaster occurrence is endogenized into the discount rate equation， and different types of loss functions are designed to predict economic losses from 2025 to 2200 under various scenarios. The main conclusions are as follows： （1） The economic losses from “fat-tailed” high temperatures are significantly higher than those in the benchmark scenario， under the nonlinear scenario， the global loss in 2 200 could reach 11.06 trillion US dollars， which is 10.8 times higher than that in the linear model. Based on the temperature control targets of the Paris Agreement， simulations are conducted for three warming thresholds： 1.5 ℃， 2 ℃， and 2.3 ℃， and it is found that when the temperature rise exceeds 2.3 ℃， the proportional loss will increase sharply；（2） The disaster discount rate significantly increases the social cost of carbon emissions， the socially nonlinear growth cost of carbon emissions in 2 100 is 1 091.8 US dollars per ton， which is 5.2% higher than that in the traditional model， revealing that the current carbon pricing mechanism is seriously underestimated；（3） If temperature uncertainty and loss variance increase by 10%， the losses will increase by 30.1% and 21.0% respectively， verifying the uncertainty of the loss results. This study dynamically links “fat-tailed” risks with the discount rate， proposes an analytical framework of “risk premium carbon pricing”， and enriches the research content of climate change economics.