AI models
BloombergGPT
Training compute
2.4×10²³ FLOP
Parameters
50.6B
Published
Mar 30, 2023
BloombergGPT is an AI model developed by Bloomberg and Johns Hopkins University (United States), first published in March 2023. It works in the language domain, on tasks such as language modeling, language modeling/generation, question answering and 2 more.
Training it took an estimated 2.4×10²³ FLOP of compute (estimation method: reported,hardware). The model has 50,558,868,480 parameters. It was trained on roughly 569B datapoints. Training ran on 512 NVIDIA A100 for about 1.3K hours. The compute alone is estimated at $370K in 2023 dollars.
Access: Unreleased. Its weights are not openly released. The reference paper has 1,299 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- Bloomberg, Johns Hopkins University
- Country of organization
- United States
- Domain
- Language
- Task
- Language modeling, Language modeling/generation, Question answering, Financial management, Text classification
- Training compute
- 2.4×10²³ FLOP
- Compute estimation method
- Reported, Hardware
- Parameters
- 50,558,868,480
- Dataset size
- 569B
- Training hardware
- NVIDIA A100
- Chips used
- 512
- Training time
- 1,270 h
- Chip-hours
- 650.2K
- Training power draw
- 408.3 kW
- Training cost (2023 USD)
- $370K
- Numerical format
- BF16
- Model accessibility
- Unreleased
- Open weights
- No
- Citations
- 1,299
- 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