AI models
CogView
Training compute
2.7×10²² FLOP
Parameters
4B
Published
May 26, 2021
CogView is an AI model developed by Tsinghua University and Alibaba DAMO Academy (China), first published in May 2021. It works in the image generation domain, on tasks such as text-to-image and image generation.
Training it took an estimated 2.7×10²² FLOP of compute (estimation method: third-party estimation). The model has 4,000,000,000 parameters. It was trained on roughly 964.8B datapoints. Training ran on 512 NVIDIA Tesla V100 DGXS 16 GB. The compute alone is estimated at $60K in 2023 dollars.
Access: Open weights (unrestricted). Its weights are openly available. The reference paper has 979 citations. Epoch AI rates the confidence of this record as likely.
Full record
- Organization
- Tsinghua University, Alibaba DAMO Academy
- Country of organization
- China
- Domain
- Image generation
- Task
- Text-to-image, Image generation
- Training compute
- 2.7×10²² FLOP
- Compute estimation method
- Third-party estimation
- Parameters
- 4,000,000,000
- Dataset size
- 964.8B
- Training hardware
- NVIDIA Tesla V100 DGXS 16 GB
- Chips used
- 512
- Training power draw
- 259.1 kW
- Training cost (2023 USD)
- $60K
- Numerical format
- FP16
- Model accessibility
- Open weights (unrestricted)
- Open weights
- Yes
- Citations
- 979
- Epoch confidence
- Likely
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