The 5th Generation vehicular to everything (V2X) has been introduced to broaden the use cases and throughput to tackle the needs of the modern world which is full of automation and enhanced mobility. Predicting the throughput can aid the network in allocating more resources and transmitting power to tackle path loss and interference along with the retransmission rate. Here, a viable technique has been put forth to use a deep hybrid learning model trained on real-world 5G datasets to estimate the available throughput in the wireless environment, both in the uplink and downlink directions. The hybrid model has been designed by combining both long short-term memory (LSTM) and gated recurrent unit (GRU) and training the model with the necessary features to bring the best result out of the model. Its result shows that a hybrid model may be used for efficiently predicting throughput, which will facilitate the modeling of wireless scenarios for V2X applications.