Objective Counties are key spatial units for implementing the new-type urbanization and the "dual carbon" strategies， and their spatiotemporal evolution and driving mechanisms of carbon emission efficiency are crucial for addressing the green transformation challenges in resource-based regions. Methods Based on multi-source heterogeneous data from 102 counties in Shanxi Province from 2012 to 2023， this study employed a Super-SBM model with undesirable outputs to measure carbon emission efficiency. It then integrated the Dagum Gini coefficient， exploratory spatiotemporal data analysis （ESTDA）， the XGBoost-SHAP interpretable machine learning model， and panel quantile regression to systematically analyze the spatiotemporal patterns， regional disparities， dynamic transitions， and nonlinear driving mechanisms of carbon emission efficiency. Results 1） Spatiotem-porally， carbon emission efficiency exhibited a stable distribution pattern of "high in the north and south， low in the central" with phased increases， with high-efficiency areas gradually agglomerating in the core zones of urban agglomerations and areas along the Yellow River. 2） Regional disparities primarily originated from inter-group differences and super-variable density， exhibiting a fluctuating process of "divergence-convergence-re-divergence"， with persistent spatial imbalance. 3） Global and local spatial analyses indicated that carbon emission efficiency had significant positive spatial autocorrelation and path dependence. LISA time paths and spatiotemporal transitions revealed that southern and southwestern Shanxi were active zones of local structural changes， showing evident signals of state transitions and restructuring. 4） The driving mechanisms exhibited significant nonlinearity and quantile heterogeneity. Economic density and land use efficiency had positive promoting effects on carbon emission efficiency with threshold effects， while population density and fiscal expenditure intensity generally acted as negative constraints. The positive effects of new-type urbanization and industrial structure upgrading were more pronounced in medium- and high-efficiency counties. Conclusion The study reveals the complex adaptive mechanisms of county-level low-carbon development in resource-based regions. The spatiotemporal heterogeneity and threshold effects of the driving factors indicate that future efforts require the formulation of differentiated and precise emission reduction policies to synergize high-quality county-level development with the "dual carbon" goals.