A new efficient perceptual image quality index, Gradient Magnitude Similarity Deviation (GMSD), is proposed. GMSD uses gradient magnitude similarity between reference and distorted images, combined with a standard deviation pooling strategy, to predict image quality. It is faster than most state-of-the-art IQA methods and achieves competitive accuracy. The method computes gradient magnitude maps, then calculates pixel-wise gradient similarity (GMS) to generate a local quality map (LQM). The standard deviation of the GMS map is used as the final quality score. GMSD is efficient and effective, outperforming other models in accuracy and speed. It is suitable for real-time applications due to its low computational complexity. GMSD is tested on multiple databases and shows consistent performance across different distortion types. It is more accurate than other models in predicting image quality, especially for complex distortions. The method is simple and effective, making it a promising candidate for various image processing applications.A new efficient perceptual image quality index, Gradient Magnitude Similarity Deviation (GMSD), is proposed. GMSD uses gradient magnitude similarity between reference and distorted images, combined with a standard deviation pooling strategy, to predict image quality. It is faster than most state-of-the-art IQA methods and achieves competitive accuracy. The method computes gradient magnitude maps, then calculates pixel-wise gradient similarity (GMS) to generate a local quality map (LQM). The standard deviation of the GMS map is used as the final quality score. GMSD is efficient and effective, outperforming other models in accuracy and speed. It is suitable for real-time applications due to its low computational complexity. GMSD is tested on multiple databases and shows consistent performance across different distortion types. It is more accurate than other models in predicting image quality, especially for complex distortions. The method is simple and effective, making it a promising candidate for various image processing applications.