As a result of a global push towards renewable energy, including wind and solar, the interest in wind derivatives which are financial contracts devised to hedge the uncertainty of wind power generation has grown significantly. An adequate wind speed model is essential for accurate pricing of wind derivatives. We examine how the choice of wind speed model affects the prices of wind speed derivatives. We show that generalized hyperbolic distribution, which can capture the skewness and heavy tail in wind speed data, provides significantly better goodness-of-fit than normal distribution. We obtain the risk neutral prices of wind speed derivatives with the Esscher transform. When different distributions are assumed for wind speed, the derivative prices vary substantially, thereby indicating a high model risk. In addition, our analysis demonstrates that the extent of model risk differs with the choice of the underlying index and the payment structure of the wind speed derivative. We conclude that model risk cannot be ignored in pricing wind speed derivatives but can be reduced through careful structuring of wind speed derivatives.