Video compression is essential for modern multimedia applications, enabling efficient storage and transmission of video content. However, lossy compression algorithms like H.264, HEVC, and AV1 often introduce artifacts such as blockiness, blurring, and ringing. Recent advances in deep learning, particularly with Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs), have shown strong potential for reducing these video compression artifacts. In this paper, we introduce a transformer-based model for HEVC video compression artifact reduction, leveraging a recurrent video restoration transformer. Our model utilizes a recurrent architecture with guided deformable attention to effectively capture spatial and temporal dependencies, reducing artifacts while preserving fine details. Experimental results on a benchmark dataset demonstrate that our approach significantly enhances visual quality, achieving an average PSNR improvement of 0.51 dB. These findings underscore the effectiveness of transformer-based architectures in reducing video compression artifacts.