There is a considerable gap between the translation performance of current mainstream large language models on high-resource languages such as English, Chinese, and Spanish, and their performance on low-resource languages such as Swahili and Burmese.