OPEN MODELS / Image editing and upscaling
Run flux-2-klein-base-4b-mflux-q8 on a cloud GPU
flux-2-klein-base-4b-mflux-q8 is a Image editing and upscaling model by mflux-community. See how much GPU memory it needs, which GPUs it fits on today and what it costs to run it yourself, billed per second.
GPUs it fits on today
Our estimate is about 11 GB for inference with BF16 weights, including working memory. Real use depends on resolution, batch size and context length.
| GPU | Memory | Per hour | 10 hours | |
|---|---|---|---|---|
| RTX 3060 | 12 GB | $0.16 | $1.60 | Rent |
| Titan Xp | 12 GB | $0.18 | $1.80 | Rent |
| GTX 1080 Ti | 11 GB | $0.20 | $2.00 | Rent |
| GTX TITAN X | 12 GB | $0.20 | $2.00 | Rent |
| RTX 2060 | 12 GB | $0.20 | $2.00 | Rent |
Live starting prices from our catalog. You pay per second while the machine runs.
How to run flux-2-klein-base-4b-mflux-q8
- 01
Rent a GPU
Pick one of the GPUs above. Billing is per second; stop the machine and you stop paying.
- 02
Open a terminal or Jupyter
Connect over SSH or open Jupyter in your browser. PyTorch and CUDA are ready to use.
- 03
Download and run the model
Run the commands below. The weights download straight from Hugging Face to your machine.
pip install -U diffusers transformers accelerate
huggingface-cli download mflux-community/flux-2-klein-base-4b-mflux-q8 --local-dir ./model
python -c "import torch; from diffusers import DiffusionPipeline; p = DiffusionPipeline.from_pretrained('./model', torch_dtype=torch.bfloat16).to('cuda'); print(p)"Why run it yourself?
Batch editing and upscaling are heavy jobs. A cloud GPU finishes them quickly and you stop paying when the job is done.
Questions
- How much GPU memory does flux-2-klein-base-4b-mflux-q8 need?
- By our estimate about 11 GB for inference (4B parameters). Higher resolution, longer context or bigger batches need more.
- How much does it cost to run flux-2-klein-base-4b-mflux-q8?
- From $0.16 per hour on RTX 3060, billed per second. Ten hours cost about $1.60.
- Do I need a subscription?
- No. You pay only for the time your machine runs, from your USD balance. Stop it whenever you are done.
Try it on your own GPU in minutes
Pick a GPU, open Jupyter or SSH and run any open model. You pay per second and can stop anytime; no subscription, no setup fee.
- Per-second billing in USD
- PyTorch and CUDA ready in minutes
- Rent from ChatGPT or Claude too