On July 31, Huawei officially open-sourced the large model openPangu-2.0-Pro under the openPangu AI model brand, along with model weights, basic inference code, and technical reports. The model is a large-scale Mixture of Experts (MoE) language model trained on Ascend NPU, with a total parameter scale of approximately 505B (505 billion), about 18B parameters activated per token, a context length of 512K, and a training data size of about 34T Tokens. In the post-training phase, the model completed supervised fine-tuning (SFT), multiple specialized reinforcement learning (RL), and achieved capability fusion through online distillation (OPD). This release is part of Huawei's openPangu 2.0 open-source plan, announced at the HDC 2026 Developer Conference in June by Huawei Executive Director Yu Chengdong, which includes gradually open-sourcing seven key components starting from June 30. The 92B-parameter openPangu-2.0-Flash model was released first on June 30. Yu Chengdong stated that Huawei will allocate a large amount of AI computing resources to support the domestic industry ecosystem, while focusing on optimizing the model's efficiency in latency and throughput. The Ascend-native AI development ecosystem is expected to be further improved.