Huawei Open-Sources 505B-Parameter MoE Model openPangu-2.0-Pro
langsfang · reddit · 2026-07-31
Huawei has officially open-sourced openPangu-2.0-Pro, the latest model in the Pangu series. Trained on Ascend compute, this Mixture-of-Experts (MoE) model features 505B total parameters with 18B activated parameters and a 512k context length.
The model was pre-trained on 34T tokens. For post-training, it utilizes unified SFT (combining slow and fast thinking), multiple specialist RL training, and on-policy distillation. A comprehensive technical report has also been released.
Related event: Huawei Open-Sources 505B Parameter Pangu MoE Model(3 posts)→
More from Models
- Reflection's Beam, a 501B-parameter open MoE model with 1M context, lands on Hugging Face under Apache 2.0 — QGallouedec · 2026-10-16
- Why are sub-15B open models so rare now? Labs seem to have settled on 24-27B minimum — Hot_Example_4456 · 2026-10-16
- User spots model openly generating "ideal responses" to train other models — liminal_bardo · 2026-10-16
- Reflection's 501B MoE Beam hits 463 tok/s on 8x GB200 with NVFP4 at day 0 — zhyncs42 · 2026-10-16
- Gemma 4 31B QAT on one 24GB card: 46 tok/s plain, ~80 with draft model on RTX 4090 — Subject_Poet_3861 · 2026-10-16
- JevBench v1.7.46 leaderboard: new d1-ultra tops 32 API systems at 76.1 composite — airesearch12 · 2026-10-16