Abstract: Accurate thermosphere density modeling is a prerequisite for reliable orbit prediction of satellites and space debris, particularly under the growing demands of modern space traffic management in low Earth orbit. This study systematically evaluates the performance of the High Accuracy Satellite Drag Model (HASDM) using thermosphere density data retrieved from SWARM-C satellite accelerometer measurements spanning the period 2014－2019. The analysis investigates model bias and variability in response to different solar and geomagnetic activity levels, as well as latitude and local time dependencies. Results indicate that solar activity exerts a marked influence on model performance: during moderate to high solar activity years, HASDM exhibits a mean bias of approximately 12.5% with a standard deviation near 0.2, whereas under low solar activity conditions, the bias increases to 18.7% and the standard deviation rises to 0.4. During geomagnetic disturbances, the model maintains an average bias about 17%, though with an elevated standard deviation, particularly during the main phase of storms. In terms of spatial distribution, polar regions demonstrate the lowest bias (5%～10%), with relatively larger variability in the southern hemisphere; conversely, equatorial regions present the highest biases, ranging between 20% and 30%. The diurnal pattern further reveals peak modeling errors during 03:00－06:00 LST and 18:00—24:00 LST, highlighting limitations in representing nighttime density variations. Additionally, during geomagnetic storms, HASDM tends to overestimate density in the initial phase, displays significant fluctuations in the main phase, and gradually stabilizes during recovery. These findings highlight systematic deficiencies in existing empirical parameterizations and suggest the necessity of incorporating enhanced solar-geophysical proxies and regionally adaptive corrections.