Air pollution is a severe danger that impacts the natural environment and humanity. When chemicals physical agents or biological agents contaminate the air and affect the natural characteristics of the surrounding atmosphere, this is known as air pollution. Dust, pollen mold sores, ozone, and other solid and gaseous particles can be included. With an emphasis on detecting levels of pollution, this paper presents thorough assessments of many machine learning and deep learning models for air quality categorizations. Based on the category efficiency and accuracy, we assessed models such as Support Vector Machine (SVM), Random forest, KNN, Recurrent Neural Networks, and LSTM networks. According to the results, RF demonstrated essential reliability in air quality prediction, achieving an ultimate accuracy of 99.8% followed by SVM at 97.5% and KNN at 97.2%. RNN and LSTM models, on the other hand, demonstrate just 42% accuracy, indicating difficulty in using these architectures for this purpose. These results illustrate the superiority of distance-based and ensemble models over recurrent models for classifying in air quality, influencing future approaches in environmental monitoring systems.