Coffee quality recognition through brand and product assurance is critical for producers and consumers. However, classification based on the processing and farming region can be complex and time-consuming. Therefore, a SIMCA (soft independent modeling of class analogies) model was built using a portable near-infrared (NIR) spectrometer and 182 samples to differentiate among coffee qualities. The samples comprised (i) specialty coffees from two species (C. arabica and C. canephora), (ii) Arabica coffees from regions of geographical indication (GI), and (iii) commodity coffee blends. The developed SIMCA model obtained a lower, but good classification accuracy, between 76 and 90%, for classifying the individual Arabica coffees from regions of GI. The model classified specialty Arabica and Conilon coffees with 98 and 95% accuracy, respectively. The built model could distinguish the specialty of the Arabica coffees from regions of GI. Furthermore, the models rule out the specialty coffee samples from the commodity coffee blends class with 100% accuracy. Therefore, this study indicated that using NIR can offer a rapid and non-destructive analytical method to ensure the authenticity of specialty coffees.