LamaDeck
Version 1.0 · free and open source (GPL-3.0)

A control panel for your local AI stack

LamaDeck is a desktop app for KDE that runs Ollama on every computer in your home network, joins all their models behind one address for VS Code and other tools, and has its own chat that can search the web and cite its sources. No terminal, no Docker, no cloud.

Install on Debian / Ubuntu User guide Source code on GitHub
LamaDeck dashboard: switches for Ollama, the LiteLLM router and Open WebUI, hosts online, loaded models and CPU, VRAM and RAM gauges
Dashboard: every service with its own switch, the models in memory, and the load on this computer.

Everything in one window

It started as a small switch for the Ollama service and grew into one place for the whole stack. Every button calls the same tools you would type in a terminal, and actions that need admin rights ask for your password in the usual KDE dialog.

Ollama on several hosts

This computer and any machine on your LAN: online or offline, version, response time, VRAM and RAM used, and which models sit on disk or in memory. Add a host by typing its address.

One address for all models

The LiteLLM router puts every host behind http://localhost:4000/v1 (OpenAI-compatible). Fixed aliases such as code or chat, and a ready config.yaml for Continue.dev.

Find and download models

One search for the Ollama library and Hugging Face GGUF files, with the real file size of every variant and a memory estimate. An "only MoE" filter for APUs and laptops.

Built-in chat

Talk to any model on any host, with tokens per second under every answer. Web search with sources, pictures, PDF and Office documents, links to web pages, and conversations saved on your disk.

Open WebUI with one click

Installed and run without Docker or admin rights. Start, stop, open it in the browser, change the port and back up all chats to a ZIP file.

Logs that explain themselves

Ollama, LiteLLM, Open WebUI and the app in one list. Errors get a plain explanation of what went wrong, and a link to the host they mention.

A chat that shows the numbers

  • Web search with sources: one click on the globe button, the model writes the queries, the app reads the pages and the answer cites them as [1], [2]. Results from SearXNG installed with one click, no account.
  • Speed of every answer, measured by Ollama: generation and prompt processing in tokens per second.
  • Context usage, e.g. 4,051 / 65,536 tokens, with a warning before the model starts forgetting the beginning.
  • Thinking on or off for reasoning models such as gemma4 or qwen3: a short answer in half a second instead of fifteen.
  • Large files up to 20 MB are split into excerpts and searched with an embedding model in Ollama. Nothing else to install.
  • Regenerate, edit the last question, copy, rename and export to Markdown.
LamaDeck chat: a conversation list on the left, a code answer in the middle, model details and speed statistics on the right

Or keep it in the tray

  • A small panel next to the clock: service switches, CPU, VRAM and RAM, and the models in memory.
  • Opens the chat or the main window with one click. Closing the window keeps the app running.
  • Optionally starts after login, already hidden in the tray.
  • Seven interface languages: English, Polish, German, Spanish, French, Portuguese and Italian.
LamaDeck tray panel with service switches and resource gauges

Install

Made for KDE Plasma (other desktops work too): a repository for Debian and Ubuntu, an AppImage for every other distribution. Ollama itself can be installed from inside the app.

Recommended

From the APT repository

Add the repository once. After that LamaDeck updates together with the rest of your system.

sudo curl -fsSLo /usr/share/keyrings/lamadeck-archive-keyring.gpg \
  https://cyryllo.github.io/LamaDeck/apt/lamadeck-archive-keyring.gpg
sudo curl -fsSLo /etc/apt/sources.list.d/lamadeck.sources \
  https://cyryllo.github.io/LamaDeck/apt/lamadeck.sources
sudo apt update
sudo apt install lamadeck

From a downloaded .deb

Download lamadeck_…_all.deb from the releases page and install it. The package adds the APT repository for you, so updates arrive the same way.

sudo apt install ./lamadeck_*_all.deb

Any distribution: AppImage

One file with everything inside, for Fedora, Arch, openSUSE and others. Download LamaDeck-…-x86_64.AppImage from the releases page and run it. On the first start it offers to add itself to the applications menu.

chmod +x LamaDeck-*-x86_64.AppImage
./LamaDeck-*-x86_64.AppImage

Without admin rights

Copies the app into your home folder and adds it to the menu. Run the same script again to update.

git clone https://github.com/cyryllo/LamaDeck.git
cd LamaDeck
./install.sh

The AppImage needs glibc 2.34 or newer (Ubuntu 22.04, Debian 12, Fedora 35 and later). LamaDeck was called "Ollama Manager" up to version 0.9; installing it replaces the old package and keeps your settings and conversations.

Roadmap

1.0 · complete

  • Done: hosts, models, LiteLLM router, Open WebUI, log, tray, seven languages
  • Done: chat with attachments, saved conversations, context usage, thinking on/off, Markdown export
  • Done: web search in the chat with sources (SearXNG or Ollama's web search)
  • Done: APT repository, .deb and AppImage

From now on · bug fixes

The app covers what it set out to do. Anything bigger, such as user accounts, knowledge bases or voice, is what Open WebUI is for, and it is built in.