Tomato leaf diseases significantly impact global agricultural productivity, requiring precise and efficient detection systems for management. While several studies have employed YOLO models for tomato leaf disease detection, thorough comparisons across the latest YOLO versions remain limited. This study evaluates four state-of-the-art YOLO models to determine the most effective model for tomato leaf disease detection. A dataset of 10,853 labeled tomato leaf images covering ten disease classes was utilized. Data augmentation techniques, including random flips, rotations, and scaling, were applied to enhance model generalization. The models were trained under identical conditions and assessed using key performance metrics: Precision, Recall, F1-score, mAP@0.5, mAP@0.5:0.95, and inference time. Results demonstrate that YOLOv11 achieved superior detection accuracy with an mAP@0.5 of 0.991 and an F1-score of 0.985, positioning it as the preferred model for applications prioritizing detection accuracy. Conversely, YOLOv8 recorded the fastest inference speed (0.045 seconds), positioning it as the most suitable model for real-time detection tasks. YOLOv10 delivered balanced performance, excelling in mAP@0.5:0.95 (0.921), showcasing its robustness across varying IoU thresholds. These insights offer valuable guidance for choosing YOLO models to optimize tomato leaf disease detection in diverse agricultural environments.