Text summarization is a type of semantic compression that involves constructing a reduced version of a text while preserving the most important components of its content. Named entity recognition is a task that helps to extract terms denoting persons, organizations, locations, notions specific for the field of knowledge in question. This paper examines the effect of multi-word named entity markup on the quality and semantic accuracy of Russian medical text abstractive summarization. Pretrained transformers T5 and GPT3 were used to test the hypothesis, and the results were evaluated using metrics BERTScore, BLEU, ROUGE-1, ROUGE-2 and ROUGE-L. The experiments were carried out for two datasets: a corpus of medical articles from the Russian Medical Journal and a dictionary of multi-word medical terms. We evaluated results of summarization obtained in two modes: with and without NER. The findings suggest that the markup has no significant impact on the results, with the summarization output showing better performance without markup. In addition, an application for summarizing medical texts in Russian, English and German was developed in the course of this work. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.