Huawei Open-Sources 505B Parameter Pangu MoE Model

Huawei has open-sourced openPangu-2.0-Pro, a 505B parameter MoE model trained on Ascend computing power. Now available on Hugging Face, it features 18B active parameters per token and supports a 512K context length, specifically optimized for long-context agents.

2026-07-31 ~ 2026-07-31 · 3 related posts

• Huawei Open-Sources 505B-Parameter MoE Model openPangu-2.0-Pro — langsfang · 2026-07-31
• Huawei Open-Sources 505B Parameter Pangu MoE Model Tailored for Long-Context Agents — 智东西 · 2026-07-31
• Huawei Releases openPangu-2.0-Pro: A 505B MoE Trained Entirely on Ascend NPUs — zephyr_z9 · 2026-07-31