Tea and herbal infusions are significant to human life because they possess many medical benefits such as antioxidant and anti-inflammatory properties, among others; for that reason, the identification of the tea class is crucial to fulfill this task. Some analytical methods are used to describe the chemical composition, nonetheless, they are costly, time-consuming and require sophisticated laboratory training. Novel devices like electronic noses (e-noses) offer real-time and objective odor monitoring to identify tea types. The aim of this work is to determine which kind of tea (among 9 known classes) a sample belongs based on volatile organic compounds (VOCs) e-nose detection using two models of artificial intelligence: support vector machine (SVM) and convolutional neural network (CNN). A total of 59 tea samples were analyzed by a lab-made e-nose consisting of an olfactometer, seven metal-oxide semiconductor (MOS) gas sensors, and a 12-bit analog-to-digital converter. Each tea was sampled ten times to ensure repeatability, obtaining a database of 590 tea measures with 2499 samples per sensor. The data was modelled using machine learning (ML) and deep learning (DL) techniques where DL can classify without a previous feature extraction. The SVM model employed principal component analysis as a feature extraction method. Using 10-fold cross-validation, the capability of both models was assessed. The classification rate was ca. 96.4% in both artificial intelligence (AI) methods, demonstrating that DL models are cost-effective and efficient.