GPU for Stable Diffusion and Flux from €0.12 an hour
SDXL runs on a 12 GB card, Flux in fp8 needs 16 GB, and Qwen-Image or HiDream 24 GB. An RTX 4090 makes a 1024 × 1024 Flux schnell image in about four seconds. On VRAMfarm ComfyUI opens in your browser with the models already downloaded, from €0.12 an hour.
Which image and video app runs on which card
| GPU | Memory | SDXL | Flux | Qwen-Image | Wan 2.2 5B video | Wan 2.2 14B video | Price |
|---|---|---|---|---|---|---|---|
| RTX 3060 | 12 GB | ✓ yes | — | — | — | — | €0.12/h |
| RTX 4060 Ti | 16 GB | ✓ yes | ✓ yes | — | — | — | €0.16/h |
| RTX 3090 | 24 GB | ✓ yes | ✓ yes | ✓ yes | ✓ yes | — | €0.30/h |
| RTX 4090 | 24 GB | ✓ yes | ✓ yes | ✓ yes | ✓ yes | — | €0.59/h |
| RTX 5090 | 32 GB | ✓ yes | ✓ yes | ✓ yes | ✓ yes | — | €0.76/h |
| RTX A6000 | 48 GB | ✓ yes | ✓ yes | ✓ yes | ✓ yes | ✓ yes | €0.72/h |
| L40S | 48 GB | ✓ yes | ✓ yes | ✓ yes | ✓ yes | ✓ yes | €1.06/h |
Whether our ready-made app runs on the card: SDXL needs 12 GB, Flux 16 GB, Qwen-Image and Wan 2.2 5B 24 GB, Wan 2.2 14B 48 GB. With your own setup, offloading and smaller formats go lower, but slower. Prices are for one card per hour and change with the market; the price is fixed when you launch. Billing is per minute, minimum 15 minutes.
Ready to use, or your own setup
ComfyUI with models
Choose Flux, SDXL, Qwen-Image, Qwen Image Edit or HiDream: the server installs ComfyUI, downloads the model and opens the interface in your browser.
Your workflows and LoRAs
Upload workflows, LoRAs and checkpoints through the browser or SSH. Save /workspace on stop and they are there next time.
Automatic1111 and Forge
Take a clean server with PyTorch and install the interface you prefer, or bring a Docker image with everything inside.
LoRA training
Train a Flux or SDXL LoRA on your own photos on a 24 GB card with kohya or ai-toolkit.
Memory first, then speed
If the model does not fit, ComfyUI offloads part of it to system memory and every image gets several times slower. So first pick a card where your model fits with room to spare, then a faster one if you generate a lot. Between cards with the same memory the newer ones are faster: an RTX 4090 is roughly 1.6–2.5 times faster than a 3090 for images.
How long the start takes
The server is ready in 2–5 minutes; after that the app downloads its model: SDXL about 7 GB, Flux about 17 GB, Qwen-Image about 32 GB. The dashboard shows when the app is ready. Download minutes are billed, so a saved folder pays off if you start often.
Commercial use
The images are yours, but the model licence decides commercial use. SDXL and Flux.1 schnell (the version in our app) allow it; check the licence of any other model or LoRA you add.
What 1,000 images cost
An RTX 4090 makes about 15 Flux schnell images a minute at 1024 × 1024, so a thousand images take a little over an hour: roughly €0.65. An RTX 3090 is cheaper per hour but about half as fast at Flux: about €0.62 for the same thousand.
Ready-made apps for this
The app installs itself with its models and opens in your browser. You pay only for the GPU time.
Guides on this
11 November 2026 Best GPU for Flux, SDXL and ComfyUI: 16 GB vs RTX 3090, 4090, 5090 and 48 GB cards
SDXL runs on 12 GB, Flux on 16 GB, Qwen-Image and HiDream need 24 GB, Wan 2.2 14B 32–48 GB. Which card to rent for ComfyUI and when the faster one is cheaper.
11 November 2026 ComfyUI on a rented GPU: your own models, LoRAs, custom nodes and the 502 page
Run ComfyUI on a rented GPU from €0.59 an hour: when it is ready, where models and LoRAs go, how to add custom nodes and keep them for next time.
12 November 2026 Train a Flux LoRA on your own photos: 20 images, 24 GB and about two hours
How many photos, which captions, ai-toolkit or FluxGym, 24 or 16 GB, 2,000 steps in 1–2.5 h on an RTX 4090, and what happens to your photos afterwards.
Frequently asked questions
What is the minimum GPU for Stable Diffusion?
SD 1.5 runs on 6–8 GB, SDXL comfortably on 12 GB, Flux in fp8 on 16 GB. Our cheapest card for images is the RTX 3060 with 12 GB, from €0.12 an hour.
Which GPU is best for Flux?
The RTX 4090 is the best value: 24 GB holds Flux with ControlNet and LoRAs, and it is fast. The RTX 5090 with 32 GB is about 30–50% faster still.
Can I use my own models and LoRAs?
Yes. Upload them to ComfyUI through the browser, or copy them over SSH. Save the folder when you stop, and they stay for the next session.
Is there a queue or a limit on images?
No. The card is yours while the server runs: no queue, no credits per image, no content filter from us beyond the law and our terms.
Open ComfyUI in a few minutes
Top up from €10, choose the app and the card, and start generating.