Cursor ships a .deb, an RPM and an AppImage for Linux, and the download page says nothing about how any of them update. The .deb quietly registers an apt repository, and that repository is the best way to install the editor. This guide sets it up in three steps, compares the other install methods, and covers Grok Bot, which comes with paid plans and runs on Ubuntu.
Anthropic ships an official Claude desktop app for Ubuntu and Debian, installed from a signed apt repository rather than a downloaded file, so updates arrive with the rest of your system. This guide adds the repository, verifies Anthropic’s signing key before trusting it, installs the app, and gets Cowork’s virtual machine running when the tab comes up grey.
OpenAI ships an official ChatGPT desktop app for Linuxas a deb package that wires itself into apt so future updates arrive with everything else. This guide installs it, signs you in, gets Codex working on a local repository, and fixes the Wayland bug that makes typing do nothing.
LM Studio is a desktop app that downloads and runs AI models on your own machine, with a chat window, a model browser and a built-in check for whether a model will fit your hardware. This guide installs it from the official deb package, gets a model running, and fixes the two errors that stop the AppImage starting on recent Ubuntu.
Open WebUI puts a browser chat window in front of the models you already run with Ollama, with saved conversations, model switching and file uploads. This guide installs it with one Docker command, connects it to Ollama, and fixes the error that stops most first attempts.
Mistral is the open model family from France, and its small models are the whole point. This guide installs it with Ollama in two commands and explains why the obvious command gets you a two year old model when a better one is a letter away.
Qwen is Alibaba’s open model family, and it spans a wider range of sizes than any of its rivals. This guide installs it with Ollama and helps you pick the version your machine can actually run, which is usually not the newest one.
DeepSeek Harness is an open source agent that runs in your browser. It normally bills a cloud API, but it will talk to a model running in Ollama instead, so nothing is billed and nothing leaves your machine.
DeepSeek R1 runs on your own machine, for free, and shows its reasoning before it answers. Here are the install steps, the download sizes, and which version fits your hardware.
Ollama runs AI models on your own machine instead of someone else’s server. Here are the three ways to install it on Ubuntu, how to control the background service, and how to remove it cleanly.
