MMPose is an open-source pose estimation toolbox built on PyTorch as part of the OpenMMLab ecosystem. It provides a comprehensive framework for 2D and 3D human pose estimation, hand pose detection, facial landmark identification, whole-body 133-keypoint estimation, animal pose analysis, and fashion landmark detection. The toolbox implements state-of-the-art algorithms including SimCC, RLE, and the RTMPose series (RTMO, RTMW, RTMPose3D) with various backbone architectures from ResNet to Swin Transformers. MMPose achieves faster training speed and higher accuracy than other popular codebases, with extensive benchmarks across COCO, MPII, AIC, OCHuman, and many other datasets. The modular architecture allows users to customize frameworks by combining different components while maintaining comprehensive documentation and unit tests. Version 1.3.0 introduced RTMO for real-time multi-person pose estimation and support for new datasets. With 7,400+ stars and active community contributions, MMPose remains the essential toolkit for computer vision researchers working on pose estimation tasks.