Frequently Asked Questions

PixelForge rents you a cloud machine with one dedicated NVIDIA GPU. The machine is a Linux container with PyTorch and CUDA preinstalled. You open it in your browser through Studio (JupyterLab) or connect over SSH, and pay only for the time it runs. No hardware to buy or maintain.

You get a whole GPU to yourself, root access to install what you need, and a simple hourly price, charged for the exact time it runs, with no subscription. The machines have no graphical desktop, screen sharing or game streaming: they are for work you run from a notebook or a terminal, such as rendering, AI and batch jobs.

A computer with a current web browser, and a PixelForge account with credit on your balance (a top-up is any amount from €1 to €20,000, or its equivalent in the currency you pick at the day's rate (US dollars included)). An SSH client is optional — it's built into Mac, Linux and modern Windows.

No. When you deploy, we rent a verified host that has your tier's GPU free, wherever it is.

PixelForge uses a pay-as-you-go credit system. Add credits in any amount from €1 to €20,000, or its equivalent in the currency you pick at the day's rate (US dollars included), then choose a GPU tier. Running time is charged at the tier's hourly rate for exactly as long as your workstation runs. Stop anytime and the charges stop. No subscriptions, contracts, or hidden fees.

Only while your workstation is in the Running state. Stopped workstations do not incur charges. Credits are taken from your balance every 5 minutes while it runs, and for the last few minutes when you stop it. You can monitor your spending from the Billing page in your creative studio.

We accept Visa and Mastercard credit and debit cards. Card details are sent over an encrypted connection to our card payment partner.

Unused credits can be refunded within 14 days of purchase. If a running workstation stops working because of a fault on our platform or on the GPU host it runs on, we credit the time you lost on request. Contact support@pixelforge.academy for refund and credit requests.

If your credit balance reaches zero while a workstation is running, it will be stopped automatically. Your creative work is preserved for 24 hours, giving you time to add more credits and resume. Once a day, if a machine is running and your balance is below $2.00 we email you a low-balance warning.

We offer four tiers: Sketch (RTX A4000 or RTX 4060 Ti, 16 GB VRAM) $0.62/hr, Canvas (RTX 4090, 24 GB VRAM) $1.22/hr, Studio (L40S, 48 GB VRAM) $2.47/hr, and Masterwork (A100 80GB, 80 GB VRAM) $4.97/hr. The hourly price includes the machine's disk, vCPUs and RAM, and you get root access.

Usually within a few minutes; it depends on the GPU host the machine is placed on. If a machine has not started within 10 minutes we cancel the deploy and you are not charged. Once it is running, Open Studio and the SSH details appear on its page in your dashboard.

Yes, for 24 hours. When you stop a machine its disk — your files and installed packages — is kept for 24 hours. Start it again within that window and everything is as you left it. A restart needs the same host's GPU to be free, so it can be delayed. Back up anything important.

No. Machines have no display, graphical desktop or remote-desktop access. You work in JupyterLab in your browser — notebooks, a terminal and a file browser — or over SSH.

A Linux container with root access. NVIDIA drivers, CUDA and PyTorch are preinstalled, and you can add other software with pip or apt.

Click Open Studio on the machine's page to open JupyterLab in your browser — no install needed. For SSH, add your public key from a Studio terminal, then use the SSH command shown on the machine page.

Any stable connection works: Studio and SSH send text and small updates, not a video stream. A faster connection mainly helps when you upload or download large files.

Check your own connection, close other bandwidth-heavy apps, and reload the Studio tab. Start long jobs from a terminal with nohup so they keep running if your browser disconnects.

Studio opens in a mobile browser, so you can check on a job or a notebook from a phone or tablet. For real work a laptop or desktop is much more comfortable.

Anything that runs on Linux without a screen. Common choices include Blender (command-line rendering), FFmpeg, PyTorch, Stable Diffusion, Ollama and llama.cpp. Windows-only and macOS-only apps, such as Adobe Creative Cloud or Cinema 4D, don't run on PixelForge.

Yes. You can run Stable Diffusion models with PyTorch, fine-tune LoRAs and run open language models from notebooks or scripts. Web interfaces such as ComfyUI or Automatic1111 can be installed and reached through an SSH tunnel.

No. The machines are Linux containers with no display. Use them for work that runs without a screen, such as command-line renders, light baking or asset-processing scripts.

Check that you have enough credit. If a tier is out of stock, try another tier or check back later. A deploy that hasn't started within 10 minutes is cancelled and not charged. If it keeps failing, contact support@pixelforge.academy.

Make sure the machine shows Running in your dashboard. If Open Studio isn't there yet, the machine is still getting its address — refresh in a minute. If your browser warns about the certificate, choose Advanced, then Proceed. For SSH, check that you added your public key and used the port shown on the machine page.

Check your own connection and reload the Studio tab. Start long jobs with nohup or inside tmux so they keep running if you disconnect.

Run nvidia-smi in a Studio terminal. If it shows no GPU, delete the machine and deploy a new one, and tell support@pixelforge.academy the machine ID so we can check the host.