Wikipedia is a widely recognized, collaboratively edited encyclopedia and a key information source for journalists, investors, and consumers researching companies. A crucial quality criterion of Wikipedia articles is the Neutral Point of View (NPOV), but ensuring neutrality can be challenging due to editor biases. Therefore, it is essential to verify whether NPOV is met. One possible approach to achieve this is by assessing the sentiment of Wikipedia articles. This paper presents a comparative study of different approaches for sentiment analysis of Wikipedia articles about companies. We used lexicon-based (TextBlob, VADER) and transformer-based models (RoBERTa-based, DistilBERT-based models and PySentimiento). Our main objectives was to determine the differences between sentiment in various Wikipedia articles and how publicly available models apply to such texts. First, we described the data collection process, including article selection and preprocessing steps to ensure content relevancy. We also proposed three methods to calculate an aggregation score for each article. Next, we presented the overall results across all selected articles, showcasing how each model performs in sentiment analysis. Additionally, we defined subsets of articles belonging to different quality classes and separate industries and compare the sentiment for these groups. Results of this study shows that sentiment in Wikipedia articles varies depending on the industry and chosen model. Additionally it provides a practical framework for researchers and practitioners seeking to implement a sentiment analysis that may be applied in the case of longer texts. Our findings contribute to a better understanding of how each model evaluates sentiment in Wikipedia articles about companies.