Restormer is an efficient Transformer architecture designed for high-resolution image restoration. It leverages channel and spatial attention mechanisms to process features at different scales. The model achieves state-of-the-art performance on multiple low-level vision tasks including motion deblurring, image deraining, Gaussian and real-world denoising, and defocus deblurring. It is implemented in PyTorch and released with training codes and pre-trained weights.