M6-10T
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
- Alibaba
- Organisation type
- Industry
- Country
- China
- Published
- 8 October 2021
- Authors
- Junyang Lin, An Yang, Jinze Bai, Chang Zhou, Le Jiang, Xianyan Jia, Ang Wang, Jie Zhang, Yong Li, Wei Lin, Jingren Zhou, Hongxia Yang
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Multimodal, Language, Vision
- Task
- Language modeling/generation
- 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
- 10T
- Training data
- tokens
"We demonstrate a practice of pretraining unprecedented 10-trillion-parameter model, an order of magnitude larger than the state-of-the-art, on solely 512 GPUs within 10 days"
"We conduct experiments for pretraining and finetuning to analyze model competence in upstream and downstream tasks. Following the classical data setup for pretraining and finetuning, we pretrain the model on BookCorpus [52] and English Wikipedia [9], which are corpora with around 16GB of plain texts." I used http://extraconversion.com/data-storage/gigabits/gigabits-to-words.html for the conversion to number of words
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
- 5.5 × 10²¹ FLOP
- How it was established
- Hardware
512 GPUs in 10 days - using NVIDIA V100 GPUs Using the NVIDIA V100 Specifications this works out to be: 0.30 * 125E12 * 512 * 10 * 86400 = 1.66E22 (Assuming 30% utilisation, and 125 TFLOPS)
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
- Training code
- Unreleased
"This work is generally reproducible. Following the description in Section 3, researchers can easily implement the training strategy on the codebases for pretraining, including Huggingface Transformer 1, Fairseq 2."
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Confident
- Citations
- 47
Sources
Where this record came from and when it was last checked.
- Reference
- M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining
- Last updated
- 28 November 2025
What the numbers mean
Background
M6-10T was published by Alibaba, in China, in October 2021. It comes out of industry.
It works in Multimodal, Language, Vision, and is recorded as doing language modeling/generation.
Its weights were never published, so it can only be reached through its provider. No graphics card changes that.
What went into building it
The training run consumed about 5.5 × 10²¹ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.
Answers
M6-10T — common questions
How many parameters does M6-10T have?
M6-10T has 10T parameters. "We demonstrate a practice of pretraining unprecedented 10-trillion-parameter model, an order of magnitude larger than the state-of-the-art, on solely 512 GPUs within 10 days". 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 M6-10T?
M6-10T was published by Alibaba, based in China, categorised as industry.
When was M6-10T released?
M6-10T was published in October 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 M6-10T used for?
M6-10T works in Multimodal, Language, Vision, and is recorded as handling language modeling/generation. These are the areas it was designed around; they describe intent rather than a hard boundary.
How much compute was used to train M6-10T?
Around 5.5 × 10²¹ FLOP. 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 M6-10T?
None. M6-10T 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.
Is M6-10T open source?
No. M6-10T has not had its weights published, so it exists only as a service controlled by its owner.
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.