Robust estimation is used in a wide range of applications. One of the most popular algorithms for robust estimation is the random sample consensus (RANSAC) that achieves a high degree of accuracy even with a significant amount of outliers. A major drawback of RANSAC is the fast increasing number of iterations caused by higher outlier ratios, resulting in increasing computational costs. In this paper FestGPU, a framework for Fast robust ESTimation on GPU, is presented which reaches a speedup of up to 135Â compared to a singlecore CPU. Together with a C?? and a Matlab interface the framework is made publicly available on the authors' website for the research community.