In recent years, large language models (LLMs) have revolutionized many aspects of life and work, and their impact is expected to continue transforming professional practices in the near future. Artificial intelligence is poised to become a standard tool in our workflows. This paper investigates the comprehension and reasoning capabilities of LLMs for boulder detection from high-density Lidar data (20 points/m²) and its derivatives, such as DEM, DSM, slope, and roughness, evaluating their potential to achieve reliable results. Three LLMs with notable reasoning and coding capabilities—Claude 3.7 Sonnet, Gemini 2.5 Pro, and OpenAI o1—were selected for this study. Due to the complexity of working and availability with very high-resolution data for boulder detection, few studies have explored this area. As a result, this research highlights the potential of LLMs in innovative applications and underscores their role in advancing collaborative research efforts to enhance scientific capabilities.