Although large language models (LLMs) have the potential to transform biomedical research, their ability to reason accurately across complex, data-rich domains remains unproven. To address this ...