Huawei Cloud retires openPangu-2.0-Flash and openPangu-2.0-Pro, recommends DeepSeek-V4.1-Flash and GLM-5.3

Huawei Cloud announced that its openPangu-2.0-Flash and openPangu-2.0-Pro model-as-a-service offerings will cease sales at 00:00 Beijing time on 6 November and be taken offline at 00:00 on 30 November, according to a retirement notice posted on the Huawei Cloud support site and reported by Pandaily. New users cannot purchase or activate the services after 6 November, while existing users may continue to call the models until the end of the month.

The notice directs openPangu-2.0-Flash (version 20260630) users to DeepSeek-V4.1-Flash-1017 and openPangu-2.0-Pro (version 20260730) users to Zhipu’s GLM-5.3. Huawei cautions that the replacements differ in capabilities, API parameters, pricing and task performance, and advises customers to review documentation and run evaluations. Migration guidance states that MaaS standard API V1 users must move to the OpenAI-compatible endpoint, while API V2 or OpenAI-compatible users only need to change the model parameter.

The two models were among Huawei’s largest open releases this year—openPangu-2.0-Flash with 92 billion parameters released in June, and openPangu-2.0-Pro, a 505 billion-parameter sparse model, released in July. The retirement applies only to the hosted MaaS services; the open-source code remains available. Huawei indicated a new generation of openPangu models is planned but gave no names or schedule.

Our read: The abrupt phase-out forces B2B teams to re-engineer pipelines before month-end, adding short-term migration overhead while Huawei signals a longer-term shift toward newer, possibly more proprietary, model families.