AI models
UnifiedQA
Training compute
1.6×10¹⁹ FLOP
Parameters
11B
Published
May 2, 2020
UnifiedQA is an AI model developed by Allen Institute for AI and University of Washington (United States), first published in May 2020. It works in the language domain, on tasks such as question answering.
Training it took an estimated 1.6×10¹⁹ FLOP of compute (estimation method: operation counting,hardware). The model has 11,000,000,000 parameters. Training ran on 8 Google TPU v3 for about 36 hours. The compute alone is estimated at $59 in 2023 dollars.
Access: Open weights (unrestricted). Its weights are openly available. It is built on top of T5-11B. The reference paper has 816 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- Allen Institute for AI, University of Washington
- Country of organization
- United States
- Domain
- Language
- Task
- Question answering
- Training compute
- 1.6×10¹⁹ FLOP
- Compute estimation method
- Operation counting, Hardware
- Parameters
- 11,000,000,000
- Training hardware
- Google TPU v3
- Chips used
- 8
- Training time
- 36 h
- Training power draw
- 7.3 kW
- Training cost (2023 USD)
- $59
- Model accessibility
- Open weights (unrestricted)
- Open weights
- Yes
- Base model
- T5-11B
- Citations
- 816
- Epoch confidence
- Confident
More from Allen Institute for AI,University of Washington
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