Based on observations and the 24 h mid-low level wind speed and temperature forecast data from the ECMWF deterministic model in the period of January to December 2021, a support vector machine regression method was employed to develop a gust forecast model for the offshore areas of China so as to enhance the capability of predicting gusts at sea. Independent sample verification was conducted using data from January to September 2022, and a comparative analysis was performed against the gust factor method. The following conclusions were drawn. Changes in wind speed and temperature at different heights or vertical wind speed and temperature variations can all have an impact on gust forecasts. Consequently, relying solely on the 10 m wind speed forecast from the model, as done in the gust factor method, may lead to overestimation or underestimation of gusts in certain situations. The forecast model that incorporates upper-level meteorological elements based on the gust factor method can achieve better forecast performance. For gust of scale 9, the accuracy of this model is 50%, significantly higher than the 30% accuracy of the gust factor method. It also demonstrates good performance for gusts of large scales in different sea areas. When there is a certain deviation between the 10 m wind speed forecast from the ECMWF deterministic model and the observed wind speed, the gust forecast results of the support vector machine regression model, which considers upper-level element information, are closer to the observation compared to the result by the gust factor method.