<p>Every aspect of modern life has become dependent on the internet. The number of internet users is increasing rapidly. More people than ever before are using the Internet. Therefore, there is a growing risk of cybercrimes and cyberthreats. Illegal activities performed over the internet are referred to as cyber threats. Over time, Cybercriminals are developing new techniques to circumvent security measures. Conventional techniques are unable to identify advanced and zero-day attacks. To date, various machine learning algorithms have been developed to detect cybercrimes and decrease cyber risks. To address this challenge, we propose a novel Hybrid deep learning model called Dent-LSTM to detect the Cyber threats. This paper proposes a novel hybrid deep learning model, Dent-LSTM, for cyber threat detection. Initially, imputation techniques are used to handle missing values, and Z-score normalization for standardizing numerical variables performed on the dataset obtained. Moreover, we employ two different feature selection methods, Correlation analysis (CA) and Mutual information (MI) to select the significant attributes in the dataset that contribute to the prediction of threats. The proposed Dent-LSTM model combines the Improved Deep Dendritic Artificial Neural Network (IDD-ANN) and the Peephole-LSTM by leveraging the benefits of each to obtain temporal correlation and hierarchical attributes from network traffic data. In addition, the model’s parameters are optimized by the Enhanced Snow Leopard Optimization (ESLO), a nature-inspired approach to optimize performance. The proposed scheme is implemented in Python. The performance of the suggested method is assessed based on accuracy, recall, f-score, precision, and specificity.</p>