Forest ecosystems play a critical role in regulating the global carbon cycle through carbon uptake. Flash droughts, characterized by rapid declines in soil moisture, may occur with or without extreme heat and are classified as compound heat flash droughts (CHFDs) and non-heat flash droughts (NHFDs). However, the role of extreme heat in shaping flash drought characteristics and their ecological impacts remains insufficiently understood. In this study, we identified CHFDs and NHFDs across China during 2001-2023 based on soil moisture and temperature data. Using remotely sensed gross primary productivity (GPP) data, we quantitatively assessed differences in forest GPP responses to CHFDs and NHFDs and evaluated the contributions of their driving factors. The results show that CHFDs occur more frequently and last longer than NHFDs in China. Forest GPP responses differ significantly between drought types. Specifically, the response magnitude to CHFDs is 30.5%, exceeding the 26.9% observed for NHFDs, with the largest difference occurring in arid regions. The response intensity to CHFDs (16.0th percentile; where lower values indicate stronger responses) is also greater than that to NHFDs (18.9th percentile). In addition, the mean response time under CHFDs (41.0 days) is longer than that under NHFDs (29.8 days). Extreme Gradient Boosting (XGBoost) models combined with Shapley Additive Explanation (SHAP) analysis were used to quantify the contributions of key driving factors. The results show that differences in forest GPP response characteristics between drought types are largely explained by the varying contributions of drought duration, soil moisture, and underlying water-use efficiency (uWUE). These findings enhance understanding of how different flash droughts impact forest carbon dynamics and provide a scientific basis for ecosystem management and climate adaptation strategies.