Large language models (LLMs) have made waves in various fields, but can they truly understand the nuances of legal reasoning? New research explored how well open-source LLMs perform in analyzing German legal texts, particularly within an educational context. The results are a mixed bag. While LLMs demonstrated some proficiency with basic legal analysis tasks, like identifying components of legal arguments, they struggled with more complex challenges, such as grading full legal opinions. This difficulty arose even when provided with extended context and clever prompting strategies. Interestingly, researchers found that an AI's ability to comprehend language played a more crucial role than its pure reasoning power. This highlights the importance of natural language proficiency for AI to grasp the intricacies of legal text. The research also showed the potential of 'Retrieval Augmented Generation,' a technique where the AI accesses a knowledge base to improve its understanding. This method significantly boosted performance, suggesting a promising avenue for future development. While LLMs show promise in legal education, they still have a long way to go before matching human expertise. This research illuminates both the capabilities and limitations of current AI in tackling the complexities of legal analysis, setting the stage for further advancements in making AI a powerful tool for legal professionals and students alike.