Pangu Ultra MoE

Closed weights Huawei 718B parameters May 2025

No estimate

No hardware requirements for this model

The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation
Huawei
Organisation type
Industry
Country
China
Published
7 May 2025
Authors
Yehui Tang, Yichun Yin, Yaoyuan Wang, Hang Zhou, Yu Pan, Wei Guo, Ziyang Zhang, Miao Rang, Fangcheng Liu, Naifu Zhang, Binghan Li, Yonghan Dong, Xiaojun Meng, Yasheng Wang, Dong Li, Yin Li, Dandan Tu, Can Chen, Youliang Yan, Fisher Yu, Ruiming Tang, Yunhe Wang, Botian Huang, Bo Wang, Boxiao Liu, Changzheng Zhang, Da Kuang, Fei Liu, Gang Huang, Jiansheng Wei, Jiarui Qin, Jie Ran, Jinpeng Li, Jun Zh…

What it does

The problem areas the model was built for. A model can carry several of each.

Domain
Language
Task
Language modeling/generation, Question answering

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters
718B

a sparse LLM with 718 billion parameters 39B activated parameters

Training data
tokens

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute
3.1 × 10²⁴ FLOP

Speculatively assuming they trained MoE model on the same amount of tokens as Pangu Ultra (13,2 T): 6 FLOP / parameter / token * 39 * 10^9 activated parameters * 13,2 * 10^12 tokens = 3.0888e+24 FLOP

How it was established
Operation counting

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Training hardware
Huawei Ascend 910B
Chips used
6,000
Hardware utilisation
MFU 30.0%

"Taking advantage of our model selection strategy and parallel computing system optimization, when training Pangu Ultra MoE, with performance comparable to DeepSeek R1 [10], we achieve a Model Flops Utilization (MFU) of 30.0% and Tokens Per Second (TPS) of 1.46M on 6K Ascend NPUs, compared to the baseline MFU of 18.9% and TPS of 0.61M on 4K Ascend NPUs."

Power draw
4.7 MW

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights
Closed — provider access only
Model access
Hosted access (no API)
Training code
Unreleased

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Likely above 10²³ FLOP
Yes
Record confidence
Speculative

Sources

Where this record came from and when it was last checked.

Reference
Pangu Ultra MoE: How to Train Your Big MoE on Ascend NPUs
Last updated
28 November 2025

What the numbers mean

What this model is

Pangu Ultra MoE was published by Huawei, in the country recorded as China, during May 2025. The category the publisher falls under is industry.

It works in the domain of Language, and is recorded as performing the task of language modeling/generation, Question answering.

Its weights were never published, so it can only be reached through its provider. No graphics card changes that.

How it was trained

Producing it required arithmetic totalling around 3.1 × 10²⁴ FLOP, on hardware recorded as Huawei Ascend 910B. That figure measures what producing the model cost, and has no bearing on how fast it answers.

Answers

Pangu Ultra MoE — common questions

01

Pangu Ultra MoE— what GPU do I need to run it?

None. This is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.

02

Pangu Ultra MoE— is it open source?

No. Its weights have not been published, so it exists only as a service controlled by its owner.

03

Pangu Ultra MoE— how many parameters does it have?

It has a parameter count of 718B. a sparse LLM with 718 billion parameters 39B activated parameters. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

04

Pangu Ultra MoE— who created it?

It was published by Huawei, based in China, an organisation categorised as industry.

05

Pangu Ultra MoE— when was it released?

It was published in May 2025.

06

Pangu Ultra MoE— what is it used for?

It works in the domain of Language, and is recorded as handling the task of language modeling/generation, Question answering. These are the areas it was designed around; they describe intent rather than a hard boundary.

07

Pangu Ultra MoE— how much compute was used to train it?

Training consumed around 3.1 × 10²⁴ FLOP, on hardware recorded as Huawei Ascend 910B. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

Source

Original publication

Record last updated 28 November 2025

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.