🌐 RENT TECH / RUNPOD CONTROL PANE

Your RunPods, one control pane

Rent, open, and manage only your pods - Virtual Desktops, web-app test remotes, and custom container images - all on real RunPod hardware, billed per second. Every pod below was created by you; nobody else's pods ever appear here, and every power button re-checks ownership server-side.

[i] Idle guard and billing policies

🔔 Idle guard (configurable per pod): 60 min with no input rings a warning chime in your open tab, 15 min later the pod stops (disk kept), and after 24h untended it terminates (disk lost). A server sweep enforces the same policy when your browser is closed. Coin figures are display equivalents only (100 🪙 = $1.00) - direct RunPod spend carries no Vibe cut and debits no coins. Mirror every dollar on /my/usage/ via POST /api/agents/runpod-sync.

STEP 1 · LAUNCH

Rent a desktop or custom image

CPU is preselected (cheapest). GPU adds hardware acceleration. Advanced dropdown rents any container image. Booking provisions a live pod and hands you a clickable proxy URL; nothing is ever faked - no stock, no key, or over budget comes back as an honest started: false state with no spend.

Loading launch options…

STEP 2 · MY PODS (ADMIN)

Control every pod you own

Desktops, test remotes, rental servers, render workers - each card shows its container image, live pod status, last activity, and idle guard, plus Stop / Start / Restart / Terminate / Delete. Stop ends compute billing (disk kept, storage still bills). Stop all pods pauses every pod at once. Start boots a stopped pod. Restart reboots in place. Terminate / Delete ends billing permanently (disk lost, asks for confirmation). The same dashboard lives at My RunPods.

Loading your pods…

STEP 3 · WEB-APP TESTING

Test 4weird games on a remote

Every test remote boots a Kasm Ubuntu desktop on port 6901 (Chromium inside), so the stream link always loads once the pod boots - the old remotes pointed at 6901 on images with no desktop server, which is why they showed “waiting” forever.

Open any game and find the 🎮 VibeCodeWorker autoplay panel under it (CPU preselected, cheapest).
Press Autoplay - a Kasm remote provisions and a stream link + one-time VNC password appear.
First boot pulls the desktop image (minutes): a 404/“waiting” page is normal - wait, then Reload.
Log in with the VNC password, open Chromium in the remote, and go to the locked game URL shown in the panel.
Play/test there. Keep the 4weird tab open: the idle guard chimes at 60 min, stops 15 min later, terminates after 24h untended.

Prefer agents or raw servers? Rent an AI agent → GPU play station: Xonotic autoplay → All your remotes: My RunPods →

💻 CPU Desktop - Ubuntu GUI (Kasm)

runpod/kasm-docker:cuda11 (template runpod-desktop) on a cheap CPU pod - full graphical desktop streamed on port 6901, software-rendered. JupyterLab + SSH on 8888 is one click away (official runpod/base:1.0.2-ubuntu2204).

Cheapest RunPod CPU with stock (cpu3/cpu5 family, 2 vCPU) - preselected default
20 GB disk max (RunPod CPU cap), one-time VNC password, SSH per RunPod console
Best for: browsing, files, notebooks, bots, cron helpers, light dev, cheap game testing

🖥️ GPU Desktop - RunPod Desktop (Kasm)

Official runpod/kasm-docker:cuda11 (template runpod-desktop). Full XFCE graphical desktop streamed on port 6901; open the link and you are logged into a GPU workstation. JupyterLab on a CUDA box is one click away (PyTorch image on 8888).

Cheapest Secure GPU at or under your $/hr max (set 0 for cheapest available) - live example shown at launch
60 GB disk, one-time VNC password (change it after login)
Best for: Blender, CUDA, ComfyUI sidecar, GPU testing

JupyterLab + SSH notebooks

CPU: official runpod/base:1.0.2-ubuntu2204 on 8888. GPU: PyTorch CUDA image on 8888. SSH per RunPod console - one click from either desktop plan above.

Best for: notebooks, light dev, bots, cron helpers
Stop ends compute billing (disk kept); terminate deletes the disk

How renting works

Sign in, pick CPU (default) or GPU, pick GUI (default) or Jupyter, optionally cap your $/hr max (GPU only) or paste a custom image.
POST /api/desktop/provision provisions a real RunPod pod - cheapest fitting stock, no theater. No stock, no key, or over budget comes back as an honest started: false state.
Open the returned proxy URL: Kasm desktop stream (GUI) or JupyterLab (Jupyter). Save the one-time VNC password.
Manage it in My pods above (or My RunPods); stop ends billing, terminate deletes the disk.
Idle guard watches every pod: chime at 60 min, stop 15 min later, terminate after 24h untended (per-pod timers configurable at launch and on each card). Sync real spend to /my/usage/ with the Sync button (RunPod card spend, no Vibe cut).

Need an agent instead of a desktop? Rent an AI agent → Want a worker to play for you? Xonotic autoplay →

Loading 4weird…

← Home

🌐 RENT TECH / RUNPOD CONTROL PANE

Your RunPods, one control pane

[i] Idle guard and billing policies

STEP 1 · LAUNCH

Rent a desktop or custom image

Rent your desktop

CPU is preselected - the cheapest remote box. Step up to GPU only when you need hardware acceleration (Blender, CUDA, AI art, GPU play). Quotes are USD/hour maximums on real RunPod pods; billed per second by RunPod, never more than the quote. Coin figures (≈ 100 coins per $1) are display equivalents only: direct RunPod spend carries no Vibe cut and debits no coins.

💡 Live GPU price example unavailable (no stock data right now). Your pod is still quoted exactly at rent time.

🔔 After 60 min with no input: warning chime + banner. After 15 more min idle: pod stops (disk kept). After 24h untended: pod terminates (disk lost). Any input resets the clock.

💻 CPU Desktopcheapest · default

A remote Ubuntu desktop in your browser; cheapest way to compute.

runpod/kasm-docker:cuda11 · port 6901 · 20 GB disk

🖥️ GPU Desktop

A full graphical desktop on real GPUs; streamed to your browser.

runpod/kasm-docker:cuda11 · port 6901 · 60 GB disk

🖥️ Ubuntu GUI desktop (default)

Graphical desktop streamed in the browser (port 6901). Log in with the VNC password.

runpod/kasm-docker:cuda11 · port 6901 · 20 GB disk

📓 JupyterLab + SSH

JupyterLab + SSH box (port 8888) for code and notebooks.

runpod/base:1.0.2-ubuntu2204 · port 8888 · 20 GB disk

Max $/hour (0 = cheapest available with stock) GPU picks the cheapest Secure GPU at or under this max. CPU always takes the cheapest CPU.

Selected plan

Ubuntu graphical desktop (Kasm) on port 6901, opened through the RunPod default proxy endpoint: XFCE desktop, Chromium, VS Code, terminal. JupyterLab + SSH is one click away (pick Jupyter below).

Browsing + files
Coding + notebooks
Always-on helper box

⚙️ Advanced: custom container image + idle timers

Custom image (Docker ref, e.g. runpod/pytorch:2.4.0-py3.11-cuda12.4.1-devel-ubuntu22.04; blank = plan default) Ports stay the interface defaults so the stream link keeps working (port 6901). Anything that is not a Docker ref is refused.

Idle timers for this pod (blank = default 60/15/24h)

Warn chime after (min, 5-240)Stop after warn (+min, 1-120)Terminate untended after (h, 1-168)

Track spend on /my/usage/