As demand in renewable energy is increasing now a day’s which gain attention of wind power generation. It is mostly dependent on wind speed (WS) which is stochastic variable. WS depends on pressure, turbulence, temperature and topography which form non linear complex relation. Therefore Artificial Intelligence (AI) models prefer in comparison to conventional model. The accuracy of different AI models for WS prediction gets changes which require comparative analysis of different AI models. In this study Long Short-Term Memory (LSTM), Bidirectional LSTM (BLSTM), Recurrent Neural Network (RNN) and Gated Recurrent Unit (GRU) are compared for WS prediction. Its performance is evaluated by error metrics. Among all these models Bidirectional LSTM provide better performance with accuracy of mean square error, root mean square error, mean absolute error and R2 value of 1.36, 1.17. 0.92 and 0.49 respectively. This finding prove efficacy of BLSTM model for WS prediction.