xTrimoPGLM -100B
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
- Tsinghua University,BioMap Research
- Organisation type
- Academia,Industry
- Country
- China
- Published
- 6 July 2023
- Authors
- Bo Chen, Xingyi Cheng, Yangli-ao Geng, Shen Li, Xin Zeng, Boyan Wang, Jing Gong, Chiming Liu, Aohan Zeng, Yuxiao Dong, Jie Tang, Le Song
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Biology
- Task
- Proteins, Protein or nucleotide language model (pLM/nLM), Protein generation
- Approach
- Self-supervised learning
- Numerical format
- FP16
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
- 100B
- Training data
- 275,000,000,000 tokens
- Batch size
- 2,097,152
Abstract: "training xTrimoPGLM at an unprecedented scale of 100 billion parameters and 1 trillion training tokens"
~24M protein sequences
"We employ batches of 2,048 sequences, each 1,024 tokens in length"
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
- 6.2 × 10²³ FLOP
- How it was established
- Reported,Operation counting,Hardware
"xTrimoPGLM-100B is trained on a cluster of 96 DGX-A100 GPU (8×40G) servers in FP16 precision from January 18 to June 30, 2023. During this time, xTrimoPGLM-100B has consumed 1 trillion tokens from the dataset consisting of Uniref90 and ColAbFoldDB. As of the current date, xTrimoPGLM-100B continues its pre-training process to pass through as many tokens as possible" 6 * 100 billion params * 1T tokens = 6e23 8*96 * 312 trillion * 163 days * 24 * 3600 * 0.3 ~= 1e24 directly given in the paper (…
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
- NVIDIA A100 SXM4 40 GB
- Chips used
- 768
- Chip-hours
- 3,004,416
- Wall-clock time
- 3,912 hours (163 days)
- Power draw
- 611.2 kW
- Compute cost
- $1,823,415
163 days
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
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
- Why it is tracked
- SOTA improvement,Training cost
- Record confidence
- Confident
- Citations
- 135
"Our extensive experiments reveal that xTrimoPGLM significantly outperforms other advanced baselines in diverse protein understanding tasks (13 out of 15 tasks across four categories)" "we propose xT-Fold, where building on the xTrimoPLGM-100B framework, marks a significant advancement by achieving SOTA results for the PLM-based structure prediction model on benchmarks such as CAMEO and the latest CASP15" (!) SOTA among PLM-based models
Sources
Where this record came from and when it was last checked.
- Reference
- xTrimoPGLM: Unified 100B-Scale Pre-trained Transformer for Deciphering the Language of Protein
- Last updated
- 1 December 2025
What the numbers mean
What this model is
xTrimoPGLM -100B was published by Tsinghua University,BioMap Research, in China, in July 2023. academia,Industry is the category the publisher falls under.
It works in Biology, and is recorded as doing proteins, Protein or nucleotide language model (pLM/nLM), Protein generation.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
What went into building it
The training run consumed about 6.2 × 10²³ FLOP, on NVIDIA A100 SXM4 40 GB. That figure describes the cost of creating it and has no bearing on how quickly it generates text.
Around 275,000,000,000 tokens went into training it.
The reason it appears in this catalogue at all is sOTA improvement,Training cost.
Answers
xTrimoPGLM -100B — common questions
Who created xTrimoPGLM -100B?
xTrimoPGLM -100B was published by Tsinghua University,BioMap Research, based in China, categorised as academia,Industry.
When was xTrimoPGLM -100B released?
xTrimoPGLM -100B was published in July 2023. 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 xTrimoPGLM -100B used for?
xTrimoPGLM -100B works in Biology, and is recorded as handling proteins, Protein or nucleotide language model (pLM/nLM), Protein generation. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.
How much compute was used to train xTrimoPGLM -100B?
Around 6.2 × 10²³ FLOP, on NVIDIA A100 SXM4 40 GB. 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 xTrimoPGLM -100B?
None. xTrimoPGLM -100B 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 xTrimoPGLM -100B open source?
No. xTrimoPGLM -100B has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does xTrimoPGLM -100B have?
xTrimoPGLM -100B has 100B parameters. Abstract: "training xTrimoPGLM at an unprecedented scale of 100 billion parameters and 1 trillion training tokens". 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.
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.