nvidia
NVIDIA-Nemotron-3-Super-120B-A12B-BF16
text generationother1M downloadsHugging Face
No server in stock right now has the 276 GB this model needs for inference.
GPU memory it needs (estimate)
Rule of thumb from the model's size. Real needs depend on batch size, context length and framework.
Inference (full precision)
276 GB
Inference (8-bit)
138 GB
Inference (4-bit)
69 GB
LoRA fine-tuning
299 GB
QLoRA (8-bit base)
150 GB
Full fine-tuning
1105 GB