AI models
CODEFUSION (Python)
Training compute
7.9×10¹⁸ FLOP
Parameters
75M
Published
Oct 26, 2023
CODEFUSION (Python) is an AI model developed by Microsoft and Microsoft Research (United States), first published in October 2023. It works in the language domain, on tasks such as code generation.
Training it took an estimated 7.9×10¹⁸ FLOP of compute (estimation method: hardware). The model has 75,000,000 parameters. It was trained on roughly 4.4M datapoints. Training ran on 4 NVIDIA V100 for about 11 hours. The compute alone is estimated at $9 in 2023 dollars.
Access: Unreleased. Its weights are not openly released. The reference paper has 57 citations. Epoch AI rates the confidence of this record as confident.
Full record
- Organization
- Microsoft, Microsoft Research
- Country of organization
- United States
- Domain
- Language
- Task
- Code generation
- Training compute
- 7.9×10¹⁸ FLOP
- Compute estimation method
- Hardware
- Parameters
- 75,000,000
- Dataset size
- 4.4M
- Training hardware
- NVIDIA V100
- Chips used
- 4
- Training time
- 11 h
- Training power draw
- 2.4 kW
- Training cost (2023 USD)
- $9
- Model accessibility
- Unreleased
- Open weights
- No
- 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