Modern distance education requires rapid development of unique educational materials that are of high quality from a methodological perspective. This task creates a significant workload for teachers and methodologists. The aim of the article is to evaluate the effectiveness and practical applicability of various artificial intelligence language models (GPT-4, Claude 3.5 Sonnet and Gemini) for creating educational content using B2 level English language materials as an example. Research methods include comparative analysis of the effectiveness of language models, development of a structured methodology for creating educational materials, and evaluation of generated content according to four key criteria: compliance with B2 level, methodological value, uniqueness of material, and educational potential. Within the framework of the research, a detailed course structure was developed covering all aspects of language learning: grammar, vocabulary, speaking, and speech perception. The research confirms the high efficiency of using artificial intelligence in creating educational materials. Key advantages include significant reduction in content development time, the ability to quickly generate unique materials, and flexibility in adapting to individual student needs. Of particular importance is the methodological variability of generated materials, allowing for the creation of various task formats for developing different language competencies. At the same time, the necessity of professional moderation and refinement of automatically generated materials is emphasized to ensure their maximum educational effectiveness.