AI models
DreamerV3
Training compute
2.2×10²⁰ FLOP
Parameters
200M
Published
Jan 10, 2023
DreamerV3 is an AI model developed by DeepMind and University of Toronto (United Kingdom and Canada), first published in January 2023. It works in the games domain, on tasks such as open ended play.
Training it took an estimated 2.2×10²⁰ FLOP of compute (estimation method: hardware). The model has 200,000,000 parameters. It was trained on roughly 1.6B datapoints. Training ran on 16 NVIDIA V100 for about 6.5K hours.
Access: Unreleased. Its weights are not openly released. The reference paper has 1,073 citations. Epoch AI rates the confidence of this record as likely.
Full record
- Organization
- DeepMind, University of Toronto
- Country of organization
- United Kingdom, Canada
- Domain
- Games
- Task
- Open ended play
- Training compute
- 2.2×10²⁰ FLOP
- Compute estimation method
- Hardware
- Parameters
- 200,000,000
- Dataset size
- 1.6B
- Training hardware
- NVIDIA V100
- Chips used
- 16
- Training time
- 6,528 h
- Training power draw
- 9.6 kW
- Numerical format
- BF16
- Model accessibility
- Unreleased
- Open weights
- No
- Citations
- 1,073
- Epoch confidence
- Likely
More from DeepMind,University of Toronto
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