Huawei releases PanGu large-scale model 5.5 with breakthrough technology to accelerate the reshaping of the artificial intelligence industry.

On the afternoon of June 20th, at the Huawei Developer Conference 2025, Huawei's Executive Director and CEO of Huawei Cloud Computing, Zhang Ping'an, shared the rich innovative applications and practical landing of the Pangu large model in the fields of industry, agriculture, and scientific research. He also released the Pangu large model 5.5, with comprehensive upgrades in five basic models including natural language processing and multimodal, accelerating the reshaping of various industries. In the field of natural language processing, the Pangu NLP large model released the 718B MoE hybrid expert model, reaching the industry's top tier in knowledge reasoning, tool calling, and other areas. It has also been upgraded in terms of efficient long sequences, low hallucination, fast-slow thinking fusion, Agent, and other characteristics to enhance user experience. The Pangu large model is based on the full-stack software and hardware training of the Ascend Cloud, marking the ability to create world-class large models based on Ascend.