While Huawei entered the large language model arena early with its original Pangu release in 2021, it has since been ...

Huawei has denied claims that its Pangu Pro MoE model copied Alibaba’s Qwen 2.5 14B, after an anonymous paper alleged high ...

Huawei released the source code for its PanGu 7B dense parameter model and the larger PanGuPro mixture-of-experts (MoE) 72B ...

Huawei argues that its model was independently trained and developed with its own Ascend Hardware platform refuting the claims made online that it used code from another source.

AI-native cloud solutions will help drive South Africa’s intelligent transformation and unlock economic growth.

Huawei's Noah Ark Lab has denied claims that its Pangu Pro Moe model copied elements from Alibaba's Qwen 2.5 14B, asserting independent development and training.

Pangu 5.5 was debuted globally for the first time at Huawei’s developer conference last month, so it has taken only a few weeks for the solutions to be rolled out to local customers.

Huawei releases its Pangu AI models as open source to counter US sanctions, a strategic move to drive sales of its Ascend hardware and win China's AI tech war.

Huawei has open-sourced two Pangu AI models and related inference tools to expand its Ascend chip ecosystem and global reach. Analysts say the move reflects a hardware-first strategy similar to Google ...

Huawei announced on Monday the open-sourcing of two of its AI models under its Pangu series, as well as some of its model reasoning technology. According to AI and tech experts, the move is ...