Every notable desktop app for chatting, coding, running agents, and fine-tuning on your own machine, with live GitHub stars, release downloads, platform support, pricing, and the engines and models each one runs.
| Maintainer | Category | License | Pricing | Latest | Compare | ||||
|---|---|---|---|---|---|---|---|---|---|
| Anthropic | Chat GUI | Proprietary | Freemium | — | — | — | — | ||
| Nous Research | Agent Runner | MIT | Open Source | — | — | — | — | ||
| Element Labs, Inc. | Chat GUI | Proprietary | Free | — | — | — | — | ||
| PewDiePie (Felix Kjellberg) | Chat GUI | AGPL 3.0 | Open Source | — | — | — | — | ||
| Unsloth AI | Fine-Tuning | Other | Open Source | — | — | — | — |
A desktop AI app is the program you install on your own computer to chat with, code with, run agents on, or fine-tune AI models, sitting on top of the inference engines and models that do the real work.
Picking one shapes how easily you can get started, whether your data stays on your machine, and how much control you have over the models you run. The simplest app that covers your real need almost always beats the most powerful one.
Start with the computer you have (Mac, Windows, or Linux), then with what you are trying to do: chat, write code, run agents, or fine-tune your own model. Pin two or three to the comparator and read the trade-offs side by side before you commit.
Side-by-Side
Pick any two or three and see stars, downloads, platforms, capabilities, and the case for each one in a single matrix you can share.
A desktop AI app is a program you install on your computer to use AI directly: chat with a model, write code, run agents, or fine-tune your own. Some run models locally on your own hardware for privacy and zero per-token cost, some connect to cloud APIs for access to frontier models, and many do both. The app is the friendly interface on top of the engine that runs the model.
For most people, LM Studio and Jan are the easiest starting points: download, install, pick a model, and chat, with no command line. They run on Mac, Windows, and Linux and support NVIDIA, AMD, Apple Silicon, or CPU. If you want a code-focused workflow or agents, the right pick depends on your machine and goal. Use the filters and the Find My App quiz on this page to narrow it down.
An inference engine (like Ollama, llama.cpp, or vLLM) is the software that actually runs the model. A desktop app is the user-facing program you click on, and it usually wraps one or more of those engines behind a friendly interface. Many apps bundle an engine so you never see it. On each app page here we list which engines it runs, so you can trace the whole stack.
Many are. Open-source apps like Jan and Open WebUI are free to use and self-host. Others are free for personal use with a paid tier for teams, and a few are fully paid. The pricing filter on this page splits the field by open source, free, freemium, and paid so you can see the real cost before you download.
It depends on the model. Small models run fine on a recent laptop, even on CPU, while large models need a capable GPU or an Apple Silicon Mac with plenty of memory. Apps that connect to cloud APIs run on any machine because the model runs in the cloud. Use the capability filters to find apps that match your hardware, and check our hardware directory to see what a given model needs.
For open-source apps, GitHub stars, contributors, last commit, and the latest release (version, date, and download count) come from a scheduled sync against GitHub. Capability flags like platform support, local vs cloud models, MCP support, and agent mode are editorial, set by reading the app and its docs. Closed-source apps show editorial data only.
Stars, release downloads, version numbers, and commit timestamps refresh on a daily cron from GitHub. Capability flags are reviewed regularly. When an app ships a major release or gets archived, this directory updates within a day.
