STNWeb is a modern web-based tool developed in my group, which was designed to visualize the behavior of optimization algorithms such as metaheuristics. It enables graphical analysis of multiple applications of multiple algorithms to the same problem instance, providing insights into their behavior. This capability can aid in identifying an algorithm’s strengths and weaknesses, helping algorithm designers iteratively refine and enhance their approaches to improve performance. STNWeb is intuitive, user-friendly, and open source. In this talk, I will introduce STNWeb and discuss its latest improvements and extensions, including the integration of large language models (LLMs) for automated interpretation of the generated visualizations.