This study systematically evaluated the gender bias performance of three English high-resource LLM (GPT-4o, Claude, and Gemini) and three Chinese high-resource LLM (Model 1, Model 2, and Model 3) in two language environments: English and Chinese, using four evaluation methods: Occupation Description, Simulated Workplace Recruitment, Occupational Story Generation, and Pronoun Coreference Resolution. The results showed that the English High-Resource LLM was biased toward males in Chinese responses and more gender-balanced or biased toward females in English responses, while the Chinese High-Resource LLM was significantly biased toward males in both linguistic contexts. The chi-square test results further confirmed the statistical significance of these differences, revealing the profound influence of model training data and cultural context on gender bias. To facilitate gender bias research and fairness improvement, this study open-sources all data and code to promote further research ( https://github.com/GZHU-Innovation-Intersection-Lab/LLMGenderBias ).