Despite long-standing concerns about gender bias in science, there remains a lack of understanding regarding the performance of female scientists as team leaders compared to their male counterparts. This study explores differences between female-led and male-led teams in terms of scientific impact, novelty, and disruption, utilizing a comprehensive data set of journal articles spanning from 1980 to 2016 across STEM fields. We employed Coarsened Exact Matching (CEM) to match female and male scientists based on their characteristics. Subsequently, we applied multivariable logistic regression models to compare the outcomes of journal articles produced by female-led and male-led teams. Our analysis reveals that female-led teams generate more novel and disruptive ideas. However, they tend to produce articles with less scientific impact compared to their male-led counterparts. This suggests a systemic undervaluation of the contributions of female scientists. Further analysis indicates that this gender bias intensifies in later career stages and with larger team sizes. Additionally, significant field-specific heterogeneity is observed, with the most pronounced bias found in Biology and Medicine. These findings highlight the urgent need for policy adjustments to address these biases and promote a more equitable evaluation system in scientific research.