You need to understand how random forest works.
Its output is expected to be different every time (randomness). If they are very different every time, it means the observations based on your data are not robust – most likely due to small sample sizes (or outliers)