EXAMS-V introduces a challenging multi-discipline exam benchmark for evaluating vision language models, emphasizing diverse languages and complex reasoning.
