AI models
Eagle 2
Training compute
4.7×10²² FLOP
Parameters
8.9B
Published
Jan 20, 2025
Eagle 2 is an AI model developed by NVIDIA, Nanjing University, Tsinghua University, Hong Kong Polytechnic University, Johns Hopkins University and New York University (NYU) (United States, China and Hong Kong), first published in January 2025. It works in the vision, robotics and language domain.
Training it took an estimated 4.7×10²² FLOP of compute. The model has 8,930,000,000 parameters. Training ran on 256 NVIDIA H100 SXM5 80GB for about 131 hours.
Access: Open weights (non-commercial). Its weights are openly available. It is built on top of Qwen2.5-7B. The reference paper has 57 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- NVIDIA, Nanjing University, Tsinghua University, Hong Kong Polytechnic University, Johns Hopkins University, New York University (NYU)
- Country of organization
- United States, China, Hong Kong
- Domain
- Vision, Robotics, Language
- Training compute
- 4.7×10²² FLOP
- Parameters
- 8,930,000,000
- Training hardware
- NVIDIA H100 SXM5 80GB
- Chips used
- 256
- Training time
- 131 h
- Training power draw
- 352.1 kW
- Model accessibility
- Open weights (non-commercial)
- Open weights
- Yes
- Base model
- Qwen2.5-7B
- Citations
- 57
- 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