Sea breezes are one of the important dynamic coastal processes that introduce a source of uncertainty in wind forecasting, both onshore and offshore. Different types of sea breezes (pure, corkscrew, and backdoor) have different coastal impacts (i.e., location and offshore extension). We characterize the behavior of sea breezes along the northeastern coast of the U.S. through seven case studies, leveraging the observational dataset collected during the third Wind Forecast Improvement Project (WFIP3). The WFIP3 field campaign (2024-2025) deployed onshore and offshore measurements, providing detailed observations of the horizontal and vertical structure of the marine boundary layer. We assess the capability of the operational High-Resolution Rapid Refresh (HRRR) version 4 (HRRR v4, implemented in Dec 2020, based on WRF-ARW v3.9) to represent the observed sea breeze characteristics. We also investigate the performance of WRF-ARW (v4.5.1) with an updated version of the Mellor–Yamada–Nakanishi–Niino (MYNN) eddy-diffusivity / mass-flux (EDMF) planetary boundary layer (PBL) scheme (EXP_NewMYNN) in simulating sea breezes and exploring the effects of model physics changes through sensitivity experiments (vs. EXP_OldMYNN with the version of MYNN-EDMF used in HRRRv4 integrated into the WRF-ARW v4.5.1 codebase). Composite analysis of seven sea breeze events, including three sea breeze types, shows that all three model simulations tend to delay the onset of sea breeze, with EXP_NewMYNN having a slightly earlier sea breeze initiation than EXP_OldMYNN. HRRR tends to generate the highest sea breeze height, sometimes leading to an overestimation. EXP_NewMYNN diagnoses a similar but slightly higher sea breeze height than EXP_OldMYNN. The mean offshore and onshore extents of the seven sea breeze cases within the WFIP3 region are about 160 km and 60 km, respectively. Case studies show that at 20 UTC, the corkscrew sea breeze has the largest sea breeze offshore extent, while the backdoor has the smallest offshore extent. The coastal site (RHOD) has a larger cross-shore wind speed compared to the offshore site (NANT) during the pure sea breeze case, in contrast to the other two sea breeze types. Scaling analysis shows that existing empirical scaling laws fail in predicting sea breeze horizontal extent using WRF-ARW model simulations, suggesting the importance of other factors that are not included in the existing governing equations, such as the presence of non-negligible ambient winds. Scaling laws apply successfully for sea breeze height and sea breeze strength (defined by the cross-shore wind component) using both observations and model simulations at the three sites, RHOD,