Huawei open-sources the 9.2B-parameter openPangu-2.0-Flash model

Chainthink

Release Time:

Summary

On June 30, 2026, Huawei released on-chain news open-sourcing the 9.2B-parameter openPangu-2.0-Flash model. Built on a MoE architecture, it features 600 million active parameters and supports 512K ultra-long context. The model is now available on GitCode’s Ascend Tribe community, including weights, inference code, and operators. A more powerful openPangu-2.0-Pro version is set to launch in July, with additional components expected later this year. Crypto news highlights this as a major AI advancement for developers and researchers.

ChainThink reports that on June 30, according to official announcements, Huawei has officially open-sourced the Pangu openPangu-2.0-Flash model and will launch it on the Ascend Tribe community on GitCode.

This model features a MoE architecture with 92 billion parameters and 6 billion activated parameters, supporting a 512K ultra-long context. According to the open-source roadmap, the initial releases will include model weights, basic inference code, and training and inference operators.

The enhanced openPangu-2.0-Pro weights and inference code will be released in July, with additional components gradually open-sourced throughout the second half of the year.

Source:Show original

Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information. Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.