Relation extraction is a key task in NLP, aimed at identifying semantic relationships between entities in a text. This study evaluates the contribution of LLMs to relation extraction in the economic domain, comparing them to a domain-specific BERT model. Four LLMs were tested: FinGPT, XLNet, ChatGLM2, and Llama3, using techniques such as few-shot learning and fine-tuning. The results show that Llama3, fine-tuned for the task, achieves the best performance in terms of F-score, surpassing other LLMs and BERT, highlighting the potential of LLMs for specialized tasks.