Workflows/Run open-source AI apps with Pinokio

Compute / WORKFLOW

Run open-source AI apps with Pinokio

Install Pinokio and launch one-click open-source AI apps such as Maestro, either on a rented RunPod GPU in your browser or on your own Windows, Mac or Linux computer.

Verified sourceSource checked 10/21/2026

LOCALRENTED GPU

Before you begin

Hardware

Requirements come from each app’s Pinokio listing. Maestro is shown as the worked example; lighter apps need less.

Machine	Running Maestro through Pinokio
RunPod · NVIDIA RTX 4090 · 24 GB	Recommended rented starting point: NVIDIA, well above Maestro’s 6 GB minimum. Filter for CUDA 12.8+ and use a 150 GB+ volume.
RunPod · NVIDIA A100 · 80 GB	Runs comfortably; more VRAM than Maestro needs. Choose it for heavier apps or larger models.
Windows PC · NVIDIA GPU 6 GB+	Supported: meets Maestro’s listed minimum. Update the NVIDIA driver so it reports CUDA 12.8 or newer.
Windows PC · AMD Radeon GPU	Not supported for Maestro because it is NVIDIA-only. Pick an app with an AMD version in the catalog, or rent an NVIDIA GPU on RunPod.
Mac · Apple Silicon 16 GB+	Not supported for Maestro because it needs an NVIDIA GPU. Install apps marked for Apple, or use the RunPod route.
Linux PC · NVIDIA GPU 6 GB+	Supported on Ubuntu 22.04+ with pciutils and dmidecode installed so Pinokio detects the GPU.

These are app requirements from Pinokio listings, not speed measurements. Disk space matters as much as GPU memory: AI apps and their models can use tens to hundreds of gigabytes.

AI APP LAUNCHER / RUNPOD · WINDOWS · MAC · LINUXSOURCE-LINKED SETUP

Run open-source AI apps with Pinokio

Pinokio is a free desktop app that installs and launches open-source AI apps in one click: image and video studios, voice tools, chat apps and more. Run it on your own computer, or on a rented RunPod GPU in your browser when your machine is not powerful enough. The worked example installs Maestro, an all-in-one local AI creative studio.

Choose software Run its commands Check the result

BEFORE YOU BEGIN

The software

Install the software for your operating system, then follow the steps to start your local server.

These requirements describe the GPU you rent. Use Windows, Mac or Linux as the browser client; your saved computer’s VRAM does not limit this setup.

CURRENT SETUP

Pinokio · official

Check the app’s listing in the Pinokio catalog for its GPU and platform support before installing. NVIDIA-only apps, including Maestro, need the Windows/Linux NVIDIA route or a rented GPU.

View software source

THE WORKFLOW

Set up Pinokio on your rented GPU

The steps below run on the rented Linux GPU; connect from your own browser. The instructions below update to match your selection.

CHOOSE YOUR SETUPRented GPU

You rent an NVIDIA GPU by the hour and open a full Linux desktop in a browser tab. Your own computer only needs a modern browser; Pinokio and its apps run on the rented Linux machine, so files and prompts you use there live on that machine. RunPod bills while the pod runs and charges for storage while it is stopped. These steps were run on a RunPod PyTorch pod (Ubuntu 24.04) on October 20, 2026.

Deploy RunPod’s official PyTorch template

In RunPod, go to Pods → Deploy and choose the official Runpod PyTorch template on an Ubuntu 22.04 or 24.04 image. Click Edit Template and apply the settings below, then pick a GPU and deploy. Maestro lists an NVIDIA GPU with 6 GB+ of VRAM; a 24 GB card such as an RTX 4090 leaves room for most Pinokio apps. Maestro’s environment uses PyTorch built for CUDA 12.8, so filter for CUDA 12.8 or newer (13.x for RTX 50-series), and it asks for 150 GB of free disk (500 GB recommended).

Setting	Value
Template	Runpod PyTorch (Ubuntu 22.04 or 24.04)
Expose HTTP ports	6080 (keep the template’s existing ports)
Volume disk	150 GB or more, mounted at /workspace
CUDA version filter	12.8 or newer

Open RunPod

Open a terminal in the pod

When the pod is running, click Connect → Jupyter Notebook, then New → Terminal. Copy and paste work normally here, and processes started here keep running after you close the tab. Paste each of the next three commands on its own and wait for its done message.

Install the desktop (paste 1 of 3)

Paste this single line and press Enter. It takes a few minutes. Wait for “STEP DONE: desktop installed” before the next step. It also installs pciutils and dmidecode, which Pinokio needs to detect the GPU.

export DEBIAN_FRONTEND=noninteractive; apt-get update </dev/null && apt-get install -y xfce4 xfce4-terminal dbus-x11 tigervnc-standalone-server tigervnc-tools novnc websockify wget curl pciutils dmidecode </dev/null && echo "STEP DONE: desktop installed"

Install Pinokio (paste 2 of 3)

Paste the next line. It downloads the latest x64 Pinokio package from the official GitHub releases. Wait for “STEP DONE: /usr/bin/pinokio”.

Pinokio releases on GitHub (x64 .deb)

URL=$(curl -s https://api.github.com/repos/pinokiocomputer/pinokio/releases/latest | grep browser_download_url | grep amd64.deb | cut -d '"' -f 4); wget -O /tmp/pinokio.deb "$URL" </dev/null && DEBIAN_FRONTEND=noninteractive apt-get install -y /tmp/pinokio.deb </dev/null && echo "STEP DONE: $(which pinokio)"

Start the desktop (paste 3 of 3)

Paste the last line. It should print “HTTP/1.1 200 OK” and “Desktop password:” followed by a password. Write the password down. Run this line again any time to restart the desktop with a new password.

mkdir -p ~/.vnc /workspace/pinokio; PW=$(tr -dc 'A-Za-z0-9' </dev/urandom | head -c 8); printf '%s\n' "$PW" | vncpasswd -f > ~/.vnc/passwd; chmod 600 ~/.vnc/passwd; printf '#!/bin/sh\nunset SESSION_MANAGER DBUS_SESSION_BUS_ADDRESS\nexec dbus-launch startxfce4\n' > ~/.vnc/xstartup; chmod +x ~/.vnc/xstartup; vncserver -kill :1 >/dev/null 2>&1; vncserver :1 -geometry 1920x1080 -localhost yes -xstartup ~/.vnc/xstartup </dev/null; pkill -f websockify; (nohup websockify --web=/usr/share/novnc 6080 localhost:5901 >/tmp/novnc.log 2>&1 &); sleep 3; curl -sI localhost:6080/vnc.html | head -1; echo "Desktop password: $PW"

Open the desktop in your browser

The desktop address is made from your pod ID, the short code under the pod’s name on its RunPod page: https://YOUR-POD-ID-6080.proxy.runpod.net/vnc.html. For example, pod ap1ebtoub9hylb opens at https://ap1ebtoub9hylb-6080.proxy.runpod.net/vnc.html. Clicking Connect → HTTP Service [Port 6080] opens the same address without vnc.html; add it to the end in the address bar. Click Connect and enter the desktop password.

Start Pinokio on the desktop

In the desktop, open Applications → Terminal Emulator and type the command below. Keep that terminal open while you use Pinokio. When Pinokio asks for its home folder, choose /workspace/pinokio so your apps survive stopping the pod. To paste into the remote desktop, open the noVNC side menu (the small tab on the left edge of the browser window), put the text in its Clipboard box, then right-click → Paste in the desktop terminal.

pinokio --no-sandbox

Let Pinokio finish its one-time setup

Pinokio downloads its own tools, such as Python, Node.js, Git, Conda and FFmpeg, into the home folder. This can take 5–15 minutes and the window shows little progress while it runs. Leave it open until the “Setting up…” message disappears.

Install an app from Explore

Open Explore, search for the app you want (for example Maestro, an all-in-one local AI creative studio) and check the platform and GPU requirements on its listing. Click Install. Large apps can take from a few minutes to more than half an hour and use tens of gigabytes.

Pinokio app catalog Maestro listing

Start the app and create one result

When the install finishes, click Start and wait for the app’s Open Web UI button before opening it; opening it early shows “Disconnected”. Create one output, such as an image, a clip or a reply, and confirm it appears. That result is the success check.

Stop the pod, don’t terminate it

When you finish, click Stop on the pod. The volume keeps Pinokio’s apps in /workspace/pinokio for a small storage fee; Terminate deletes everything. The desktop software lives on the container disk, so after the pod restarts, run pastes 1–3 again, then start Pinokio as before.

SUCCESS CHECK · Pinokio

An app you installed with Pinokio produces a result

After clicking Start, wait for the app’s Open Web UI button, open it, and create one output, such as an image, a clip or a reply. Confirm it appears. To use it again later, open Pinokio, select the app under My Apps and click Start; use the app’s Update item for new versions.

The RunPod route installs a browser-accessible XFCE desktop on RunPod’s PyTorch template because Pinokio is a desktop app. On RunPod, Pinokio and its apps run on the rented Linux machine; your computer is only the browser client.

Back to setup steps

When it doesn’t go to plan

An app says it needs an NVIDIA GPU (or shows “Disconnected”) on RunPod or Linux even though you have one

Run nvidia-smi to confirm the GPU and a CUDA version of 12.8 or newer. Install pciutils and dmidecode, check that lspci | grep -i nvidia prints a line, then fully restart Pinokio (close it or run pkill -f pinokio and start it again), click the app’s Update, then Start, and wait for Open Web UI.

Pinokio fails with “GLIBC_2.34 not found”

The operating system is too old for current Pinokio. Use Ubuntu 22.04 or newer. On RunPod, the older “RunPod Desktop” (Kasm) templates run Ubuntu 20.04; use the PyTorch-template route in this guide instead.

Installing the .deb reports “unmet dependencies … not installable”

You downloaded the ARM64 package for an x64 machine, or the reverse. Run uname -m and download the matching file: amd64 for x86_64. RunPod GPU machines are x64.

A pasted RunPod command only partly runs

Paste one command at a time and wait for its STEP DONE message. Each of the three setup commands is a single line; running them separately in the Jupyter terminal avoids lost input.

RunPod shows “Waiting for service to respond” or a 404 on port 6080

Nothing is serving the desktop yet. Confirm paste 3 printed HTTP/1.1 200 OK and that port 6080 is exposed under HTTP ports. After a pod restart, run pastes 1–3 again; if only the desktop stopped, run paste 3 again.

The desktop asks for a password you do not know

Run paste 3 again in the Jupyter terminal. It restarts the desktop and prints a new Desktop password; then reload the 6080 page and enter it.

You cannot paste into the remote desktop

Open the noVNC side menu from the small tab on the left edge of the browser window, put the text in its Clipboard box, then right-click → Paste in the desktop terminal. Short commands can also be typed directly.

Pinokio crashes with “Segmentation fault” right after its window opens on a RunPod desktop

Start it with software graphics: run export __GLX_VENDOR_LIBRARY_NAME=mesa, then pinokio --no-sandbox --disable-dev-shm-usage in the desktop terminal. Apps you install still use the GPU.

Pinokio closes after saving its first-run settings and does not come back

Pinokio restarts itself once after you save its settings, and on a Linux root desktop the restarted copy can exit. Your settings are saved; run pinokio --no-sandbox again from the desktop terminal and keep that terminal open.

“Failed to execute default Web Browser” on the RunPod desktop

The desktop has no browser for Pinokio to open links in. Install Firefox from Mozilla’s APT repository (Ubuntu’s own Firefox package is a snap and does not run in containers), then set WebBrowser=firefox in ~/.config/xfce4/helpers.rc.

An app install stops partway or the disk fills up

Free space on the drive that holds Pinokio’s home folder. On RunPod, stop the pod, increase the volume under Edit Pod, start it again, then use the app’s Install or Reinstall item.

An app says it needs CUDA or an NVIDIA GPU on a Mac or AMD machine

That app does not support your hardware. Check its listing for an AMD or Apple version, or run it on RunPod with an NVIDIA GPU.

REFERENCE LIBRARY

Sources and files

Official software, installation instructions, and support resources behind this setup.

9 SOURCES

01Pinokio desktop appdesktop.pinokio.co 02Official Pinokio download pagedesktop.pinokio.co 03Pinokio manualdesktop.pinokio.co 04Pinokio releases on GitHub (Linux .deb, .rpm, AppImage)github.com 05Pinokio app catalogpinokio.co 06Maestro listing in the Pinokio catalogpinokio.co 07Maestro source repositorygithub.com 08Install Firefox on Linux from Mozillasupport.mozilla.org 09RunPod (affiliate link)runpod.io

Keep exploring

Text

SETUP GUIDE

Qwen3.5 · 9B private chat setup

Source-backed package and platform-route plan for private local chat with Qwen3.5 · 9B. Preserves the official Ollama 9B package, a smaller 4B family fallback, and a clearly labeled community uncensored GGUF option.

Freshness 40 out of 100. Release age: 238 days old, Feb 27, 2026. Measures age, not quality.

Build a private coding assistant with Qwen3.8 27B

Run Qwen3.8 27B with Ollama, connect it to Cline in VS Code, and make one reviewable code change in a tiny local project.

Freshness 74 out of 100. Repository age: 79 days old, Aug 5, 2026. Measures age, not quality.

Arena Open Source Web coding: rank 11 of 60. Ranked model: Qwen3.8-27B. Original model ranking, not a benchmark of this package. Export dated Oct 22, 2026. Source checked Oct 23, 2026. Human preference ranking, not hardware suitability.+3

SOFTWAREOllama + Visual Studio Code + Cline + Python

Create images locally with FLUX.2 Klein

Generate images from text prompts in ComfyUI with the fast distilled or more adjustable base 4B model. Save your results and reuse the visual workflow.

Freshness 34 out of 100. Release age: 281 days old, Jan 15, 2026. Measures age, not quality.

SOFTWAREComfyUI

LOCALRENTED GPU

Verified source