AI models
BlenderBot 3
Training compute
4.3×10²³ FLOP
Parameters
175B
Published
Aug 10, 2022
BlenderBot 3 is an AI model developed by McGill University, Meta AI and Mila - Quebec AI (originally Montreal Institute for Learning Algorithms) (Canada and United States), first published in August 2022. It works in the language domain, on tasks such as chat.
Training it took an estimated 4.3×10²³ FLOP of compute (estimation method: operation counting). The model has 175,000,000,000 parameters. It was trained on roughly 1.3B datapoints. Training ran on 128 NVIDIA A100 SXM4 40 GB.
Access: Open weights (non-commercial). Its weights are openly available. It is built on top of OPT-175B. The reference paper has 290 citations. Epoch AI rates the confidence of this record as likely.
Full record
- Organization
- McGill University, Meta AI, Mila - Quebec AI (originally Montreal Institute for Learning Algorithms)
- Country of organization
- Canada, United States
- Domain
- Language
- Task
- Chat
- Training compute
- 4.3×10²³ FLOP
- Compute estimation method
- Operation counting
- Parameters
- 175,000,000,000
- Dataset size
- 1.3B
- Training hardware
- NVIDIA A100 SXM4 40 GB
- Chips used
- 128
- Training power draw
- 102.6 kW
- Model accessibility
- Open weights (non-commercial)
- Open weights
- Yes
- Base model
- OPT-175B
- Citations
- 290
- 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