Recently, the author selected real questions from the 2023 first-half teacher certification exam to test the response capabilities of three domestic large models: iFlytek's Xunfei Spark, Baidu's Ernie, and Alibaba's Tongyi Qianwen. The results indicate that these large models perform well in areas such as subject knowledge and logical reasoning. However, their performance varies significantly in questions that test regulations and comprehensive quality, particularly in open-ended questions where their responses are still not ideal. Among them, Ernie's answers are more standardized, demonstrating certain answering skills. Overall, large models could potentially be used for preparing for teacher certification exams, but secondary human review is still necessary. Whether large models can replace the assessment of teachers' professional qualities remains to be seen.