Abstract
The article discusses the possibilities of using large language models (LM), such as ChatGPT, Deep Seek and Grok, in the tasks of evaluating student papers. The author conducts a qualitative analysis of the results obtained using ChatGPT, in comparison with teaching assessments, with an emphasis on identifying the strengths and weaknesses of the automated approach. The potential advantages of using LLM are discussed – processing speed, compliance with criteria, scalability – as well as limitations associated with evaluating creativity and depth of analysis. Special attention is paid to the applicability of various models depending on the type of assignment (text, code) and the specifics of the discipline. The work is of a review and analytical nature and can serve as a starting point for further research in the field of digitalization of educational assessment and integration of LLM into the educational process.