FengWu-GHR is a data-driven global weather forecasting model operating at a 0.09° horizontal resolution, the first of its kind. It leverages a pretrained low-resolution model to inherit prior knowledge, enabling high-resolution forecasts. The model outperforms IFS-HRES in 2022 hindcast evaluations and demonstrates competitive operational forecasting skills through station observations and extreme event case studies. FengWu-GHR addresses challenges in high-resolution forecasting by introducing a novel approach that reduces computational costs and enhances forecast accuracy. The model's architecture includes a meta model based on attention-based vision transformers, spatial identical mapping extrapolation, decompositional and combinatorial transfer learning, and low-rank adaptation for long-lead forecasting. These innovations allow FengWu-GHR to achieve higher resolution, better station prediction, and more stable long-lead forecasts. The model also excels in predicting extreme weather events such as heat waves and winter storms, providing more accurate and detailed forecasts than existing models. FengWu-GHR represents a significant advancement in high-resolution global weather forecasting, demonstrating the potential of machine learning to improve weather prediction accuracy and operational efficiency.FengWu-GHR is a data-driven global weather forecasting model operating at a 0.09° horizontal resolution, the first of its kind. It leverages a pretrained low-resolution model to inherit prior knowledge, enabling high-resolution forecasts. The model outperforms IFS-HRES in 2022 hindcast evaluations and demonstrates competitive operational forecasting skills through station observations and extreme event case studies. FengWu-GHR addresses challenges in high-resolution forecasting by introducing a novel approach that reduces computational costs and enhances forecast accuracy. The model's architecture includes a meta model based on attention-based vision transformers, spatial identical mapping extrapolation, decompositional and combinatorial transfer learning, and low-rank adaptation for long-lead forecasting. These innovations allow FengWu-GHR to achieve higher resolution, better station prediction, and more stable long-lead forecasts. The model also excels in predicting extreme weather events such as heat waves and winter storms, providing more accurate and detailed forecasts than existing models. FengWu-GHR represents a significant advancement in high-resolution global weather forecasting, demonstrating the potential of machine learning to improve weather prediction accuracy and operational efficiency.