PANews reported on June 12 that Huawei released openPangu 2.0, an open-source large-scale model, at HDC 2026. Two versions, Pro and Flash, are available, both supporting 512K context. The 2.0 Pro version has a total of 505 billion parameters and 18 billion activated parameters, while the 2.0 Flash version has 92 billion parameters and 6 billion activated parameters. The official claim is that the model achieves twice the single-card throughput of other mainstream open-source models on Ascend computing power and is better adapted to HarmonyOS, improving agent task speed and accuracy. Starting June 30, openPangu 2.0 plans to gradually open-source seven major components, including newly added open-source pre-training code, post-training code, and training operators. Yu Chengdong stated that due to the large amount of computing power supporting the needs of other domestic enterprises, Huawei's self-reserved computing power is limited, and the company will focus more on latency and throughput optimization.