AI models
Gopher (280B)
Training compute
6.3×10²³ FLOP
Parameters
280B
Published
Dec 8, 2021
Gopher (280B) is an AI model developed by DeepMind (United Kingdom), first published in December 2021. It works in the language domain, on tasks such as language modeling and question answering. It counts among the frontier models: the systems trained with the most compute of their moment.
Training it took an estimated 6.3×10²³ FLOP of compute (estimation method: reported). The model has 280,000,000,000 parameters. It was trained on roughly 300B datapoints. Training ran on 4,096 Google TPU v3 for about 920 hours. The compute alone is estimated at $641K in 2023 dollars.
Access: Unreleased. Its weights are not openly released. The reference paper has 1,605 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- DeepMind
- Country of organization
- United Kingdom
- Domain
- Language
- Task
- Language modeling, Question answering
- Training compute
- 6.3×10²³ FLOP
- Compute estimation method
- Reported
- Parameters
- 280,000,000,000
- Dataset size
- 300B
- Training hardware
- Google TPU v3
- Chips used
- 4,096
- Training time
- 920 h
- Chip-hours
- 3.8M
- Training power draw
- 3.7 MW
- Training cost (2023 USD)
- $641K
- Numerical format
- BF16
- Model accessibility
- Unreleased
- Open weights
- No
- Citations
- 1,605
- Epoch confidence
- Confident
Benchmark results
More from DeepMind
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