<p>This study presents a hybrid classification framework that leverages both classical and quantum-inspired features to predict the complexity levels of MAX-3-SAT problem instances using supervised and unsupervised machine learning. The MAX-3-SAT problem, a canonical NP-complete optimization challenge, serves as a key benchmark for evaluating the effectiveness of optimization algorithms. The proposed approach utilizes classical structural features, such as the clause-to-variable ratio, alongside features derived from quantum-native representations of the problem, including interaction means and Hamiltonian coefficient sums. Although these Hamiltonian-based features can be computed classically for small instances using simulators, they are based on quantum encodings employed by variational quantum algorithms. The variational quantum eigensolver (VQE) is employed not to solve the optimization problem directly but to extract quantum-derived features such as ground-state energy and interaction metrics from Hamiltonians. By integrating these feature sets, the framework achieves high classification accuracy across four difficulty levels. The results indicate that combining classical and quantum-inspired features provides deeper insights into problem complexity, supports adaptive decision-making, and enhances the prioritization of quantum resources.</p>