Qwen
Qwen3.5-122B-A10B
image text to textapache-2.0904.2K downloadsHugging Face
No server in stock right now has the 280 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)
280 GB
Inference (8-bit)
140 GB
Inference (4-bit)
70 GB
LoRA fine-tuning
303 GB
QLoRA (8-bit base)
151 GB
Full fine-tuning
1118 GB