As the digital landscape evolves, so does the complexity of cyber threats, making traditional cybersecurity methods, such as penetration testing, less effective against sophisticated attacks. This study introduces the NextGen-PenTest model, a pioneering approach that integrates machine learning techniques with traditional penetration testing tools to enhance the detection, analysis, and prioritization of network vulnerabilities. Using a Random Forest Classifier, the model processes network-scan data and known vulnerabilities to predict the likelihood of successful exploits, thereby enabling a more efficient and targeted approach to cybersecurity. Our comprehensive evaluation in a simulated network environment, mirroring real-world infrastructure, demonstrated significant improvements with an accuracy of 0.93, precision of 0.89, recall of 0.93, and F1-score of 0.91. The NextGen-PenTest model not only optimizes penetration testing processes, but also adapts to emerging threats, offering a robust framework for proactive cyber defence. This advancement represents a shift from reactive to predictive security measures, ensuring that cybersecurity practices keep pace with the evolving threat landscapes.