Detecting out-of-distribution (OOD) data is critical to building reliable machine learning systems in the open world. The previous methods either need to use additional data or use the information of training data. The method of using only the parameter information of the model is relatively poor. We propose an efficient method for OOD detection using only model parameter information. To verify the effectiveness of our method, we conduct experiments on four benchmark datasets. Experimental results demonstrate that our RG outperforms existing state-of-the-art approaches by 4.57\% in average AUROC. Meanwhile, our method is easy to implement and does not require additional OOD data or fine-tuning process. We can realize OOD detection in only one forward pass of any pretrained model.