PanGu-α
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 Noah's Ark Lab
- Organisation type
- Industry
- Country
- China
- Published
- 25 April 2021
- Authors
- Wei Zeng, Xiaozhe Ren, Teng Su, Hui Wang, Yi LiaoZhiwei WangXin JiangZhenzhang YangKaisheng WangXiaoda ZhangChen LiZiyan GongYifan YaoXinjing HuangJun WangJianfeng YuQi GuoYue YuYan ZhangJin WangHengtao TaoDasen YanZexuan YiFang PengFangqing JiangHan ZhangLingfeng DengYehong ZhangZhe LinChao ZhangShaojie ZhangMingyue GuoShanzhi GuGaojun FanYaowei WangXuefeng JinQun LiuYonghong Tian
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Language
- Task
- Language modeling/generation, Text summarization, Question answering
- Approach
- Self-supervised learning
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
- 207B
- Training data
- tokens
- Epochs
- 1
Table 1
Column 1 in Table 4 sums to 258.5B tokens. Note however that "Epochs elapsed when training" from this table appears to conflict with other statements about training, so my confidence in the info here is somewhat low.
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
- 5.1 × 10²² FLOP
- How it was established
- Operation counting,Third-party estimation
Figure 8 digitized with https://automeris.io/wpd/?v=5_2, training the 200B model runs for 42.6B tokens. 6 * 200B * 42.6B = 5.112e22 Elsewhere, the akronomicon leaderboard indicates 538 PF-days, or 5.037e22 FLOPs. https://web.archive.org/web/20220211115721/https://lair.lighton.ai/akronomicon/ Note however, that this seems to contradict Table 4, which appears to indicate that PanGu-α 200B saw 317.569B tokens. If this were the case, training compute would be ~ 6 * 200B * 317.569B = 3.811e23
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 910
- Chips used
- 2,048
- Power draw
- 1.3 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
- Unreleased
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Likely
- Citations
- 234
Sources
Where this record came from and when it was last checked.
- Reference
- PanGu-α: Large-scale Autoregressive Pretrained Chinese Language Models with Auto-parallel Computation
- Last updated
- 1 December 2025
What the numbers mean
Background
PanGu-α was published by Huawei Noah's Ark Lab, in China, in April 2021. It comes out of industry.
It works in Language, and is recorded as doing language modeling/generation, Text summarization, Question answering.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
Training and provenance
The training run consumed about 5.1 × 10²² FLOP, on Huawei Ascend 910. That figure describes the cost of creating it and has no bearing on how quickly it generates text.
Answers
PanGu-α — common questions
Is PanGu-α open source?
No. PanGu-α has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does PanGu-α have?
PanGu-α has 207B parameters. Table 1. 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.
Who created PanGu-α?
PanGu-α was published by Huawei Noah's Ark Lab, based in China, categorised as industry.
When was PanGu-α released?
PanGu-α was published in April 2021. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.
What is PanGu-α used for?
PanGu-α works in Language, and is recorded as handling language modeling/generation, Text summarization, Question answering. These are the areas it was designed around; they describe intent rather than a hard boundary.
How much compute was used to train PanGu-α?
Around 5.1 × 10²² FLOP, on Huawei Ascend 910. 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.
What GPU do I need to run PanGu-α?
None. PanGu-α 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.
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.