Gas leak accidents can easily lead to fires and explosions, and result in conflagration. A gas leak monitoring system, that can detect gas leaks in advance, is therefore needed. In general, gas leak detection at home depends on a gas sensor. However, the gas sensor only detects the gas leak after it passes a certain threshold, which makes it difficult to take precautionary action. In this study, a system for predicting the amount of gas leaked and monitoring the gas leakage situation through a deep learning model, based on gas data collected in real life, is proposed. To determine the risk level when a gas leakage occurs, the explosion risk level is divided into five stages. The current risk, predicted risk, and absolute leakage amount can be checked through the monitoring platform. It is hoped that this model will help in establishing a gas safety management system for implementation.