Nowadays, the data over the internet are immense and are growing rapidly day by day. As the amount of online data from diffrent sources has expanded exponentially, it makes it very time-consuming and challenging for people to access relevant information promptly. Also, looking for specific information will lead to retrieving vast amounts of online data, which is difficult for people to read thoroughly. There is a need for an Automatic Text Summarization model to gather concise and quality knowledge from these data. The Automatic Text Summarization model provides a summary from a text, which this summary includes the most critical content in a compressed form of the input text. In this study, we propose a hybred extractive-abstractive NLP predictive model that generates summary text based on the input text using two diffrent Deep Learning Artificial Neural Network (ANN), The Bidirectional Gated Recurrent Units for Extractive Text Summarization Model and Bidirectional Long Short-Term Memory for Abstractive Text Summarization Model . The proposed model is capable get the most important information using the Attention Mechanism, and capable of understanding the word relations, using the Contextual Pre-trained models to provide a good paraphrased summary in a readable format.