Skip to content
gysga docs

Getting started

Gysga is a marketplace of GPU servers. Owners of GPU machines (“hosts”) connect them to the platform; you rent a GPU by the hour, run a ready template or your own Docker image, and connect over SSH or through the API.

1. Create an account

Section titled “1. Create an account”

Open the dashboard, enter your email and the 6-digit code we send you. There is no password and no identity check. The code is valid for 10 minutes and can be used once.

2. Add an SSH key

Section titled “2. Add an SSH key”

Servers are reached through our SSH gateway, which authenticates you with your public key.

  1. If you don’t have a key yet, create one:
    Terminal window
    ssh-keygen -t ed25519 -C "my laptop"
  2. Copy the public part (~/.ssh/id_ed25519.pub) and paste it in Settings → SSH keys.

Supported keys: ed25519, ECDSA and RSA of at least 2048 bits.

3. Add funds

Section titled “3. Add funds”

Top up with USDT on the TRON network (TRC20) to your personal deposit address. The balance updates automatically about 1–2 minutes after confirmation. See Add funds with USDT.

To launch a server, your balance must cover at least one hour of its price.

4. Rent a GPU

Section titled “4. Rent a GPU”

Open Rent a GPU, then:

  1. Pick a GPU model and the number of GPUs.
  2. Pick a template — for example vLLM for an OpenAI-compatible LLM API, ComfyUI for images and video, Jupyter Lab for notebooks.
  3. If the template serves a model, pick the model. Only models that fit the selected GPUs are offered.
  4. Optionally change the disk size and add environment variables.
  5. Click Launch.

The server goes through creating → pulling → starting → running. Billing starts only when it is running. See Renting a GPU.

5. Connect

Section titled “5. Connect”

Open the server in My servers — the connection panel shows ready-to-copy commands:

Terminal window
ssh -p 2222 [email protected]

For web interfaces (Jupyter, ComfyUI, Open WebUI) forward the port and open it locally:

Terminal window
ssh -p 2222 -L 8188:localhost:8188 [email protected]
# then open http://localhost:8188

For model servers you can also call the API directly with your API key. See Connecting and API.

When you’re done

Section titled “When you’re done”
  • Stop the server to pause billing. Its data in /workspace is kept for 7 days.
  • Destroy the server to delete it and its data.