Now days due to exponential use of internet and advancement in technologies, social media usage is increased exponentially. It is becoming a main source of information. Huge amount of information is getting generated in online platform. This online user generated content gives a valuable contribution in public mental health. Identifying suicidal ideation from social media text is more difficult task due to several reasons. It gives implicit information which can be used for mental health diagnosis of particular user or in preventing suicides. In this research work we represent numerical analysis of different deep learning techniques like Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM), Convolution Neural Networks (CNN) also hybrid approaches of deep learning techniques such as CNN-LSTM, LSTM-CNN, CNN-LSTM-CNN on annotated text available on online social media platforms for the detection of suicidal ideation. Here we have evaluated performance of All model's performance are evaluated based on higher training, testing and validation accuracy with different epoch and hyper parameter turning by considering minimum loss.