AI models
xTrimoPGLM -100B
Training compute
6.2×10²³ FLOP
Parameters
100B
Published
Jul 6, 2023
xTrimoPGLM -100B is an AI model developed by Tsinghua University and BioMap Research (China), first published in July 2023. It works in the biology domain, on tasks such as proteins, protein or nucleotide language model (plm/nlm) and protein generation.
Training it took an estimated 6.2×10²³ FLOP of compute (estimation method: reported,operation counting,hardware). The model has 100,000,000,000 parameters. It was trained on roughly 275B datapoints. Training ran on 768 NVIDIA A100 SXM4 40 GB for about 3.9K hours. The compute alone is estimated at $2M in 2023 dollars.
Access: Unreleased. Its weights are not openly released. The reference paper has 135 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- Tsinghua University, BioMap Research
- Country of organization
- China
- Domain
- Biology
- Task
- Proteins, Protein or nucleotide language model (pLM/nLM), Protein generation
- Training compute
- 6.2×10²³ FLOP
- Compute estimation method
- Reported, Operation counting, Hardware
- Parameters
- 100,000,000,000
- Dataset size
- 275B
- Training hardware
- NVIDIA A100 SXM4 40 GB
- Chips used
- 768
- Training time
- 3,912 h
- Chip-hours
- 3M
- Training power draw
- 611.2 kW
- Training cost (2023 USD)
- $2M
- Numerical format
- FP16
- Model accessibility
- Unreleased
- Open weights
- No
- Citations
- 135
- Epoch confidence
- Confident
SourceEpoch AI, 'AI Models'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/ai-models. Licensed under CC BY 4.0.
← All ai models