Every model, one picker
Switch between models from every major lab (Claude, GPT, Gemini, Llama and the rest) from a single menu. No separate logins to juggle.
A small native app for talking to every big model: served for you, wired to your own keys, or running fully offline on your machine. We had no idea what we were doing when we started. It works now.
Every model, one app
ChatGPTOpenAI
GrokxAI
MistralMistral AI
QwenAlibaba
GLMZ.AI
KimiMoonshot AIPlus anything you connect with your own key or run locally
Eaon is open source, and anyone can contribute. Come hang out, send a pull request, or help shape what ships next.
Join the DiscordEverything a proper desktop app should do, plus a few things no browser tab can.
Switch between models from every major lab (Claude, GPT, Gemini, Llama and the rest) from a single menu. No separate logins to juggle.
Add your OpenAI, Anthropic or Google key and talk to them directly. Keys live in your Keychain, never on a server.
Pull open models with Ollama, llama.cpp or MLX and chat completely offline. Nothing leaves the machine.
Point Eaon at Ollama, llama.cpp or MLX and it runs the model right on your laptop: no key, no account, no network. Same clean window, working on a plane. It even keeps the model warm and unloads it when you're done.
Not a subscription per lab. Start in Claude, switch to GPT halfway, run both on the same prompt. One window, every model a click apart.
Run a local model and the conversation never touches the internet. Use your own keys and they stay in the Keychain. Export or delete everything whenever. It's just files.
⌘K opens a real command palette. Pin the chats you return to, file the rest into projects, set custom instructions once, and never reach for the mouse.
A native, monochrome window that follows your system theme. No ads, no feed, no upsell. It looks like it belongs on a Mac because it was built as one.
Eaon is open source, and anyone can contribute: star the repo, send a pull request, or open an issue to help decide what ships. The download count below reads straight from GitHub releases.
Free forever, and open source. Your keys, your models, your history, kept on your machine.
The first 150 downloads get a free week of Plus, including Claude Opus 4.7 and other premium models. Email support@eaon.dev to extend yours.
Eaon for Windows isn't code-signed yet, so SmartScreen will show a blue "Windows protected your PC" box on first run: click More info, then Run anyway. That's expected, not a sign anything's wrong. Release notes ↗
Universal · macOS 14 Sonoma or later · v2026.4.0 · 0 downloads
Eaon is signed with an Apple Developer ID certificate and notarized by Apple, and the notarization ticket ships inside the download. Drag it to Applications and double-click. No right-click, no “unidentified developer”, no Terminal command, and nothing to approve in System Settings — it also works on a Mac that has never been online.
One thing to expect if you are updating. macOS ties the permissions you granted to an app’s signature, and Eaon’s has changed now that it is signed. The first time each one is needed you will be asked again for Accessibility, Automation and microphone access. Nothing is broken — approve them once more and they stay approved.
Eaon CLI is an agentic coding agent that runs in your shell, with plan mode, sub-agents and real tools. It works with any model, whether that's a local one through Ollama or a hosted one. One command, and you're in.
npx eaon-cli
Yes. The app is free forever, and open source. If you use the built-in hosted models or your own provider keys you pay whatever those cost, and local models cost nothing to run, but the app itself never costs a thing.
Not necessarily. Chat through Eaon's built-in connection, bring your own key from OpenAI, Anthropic or Google, or skip keys entirely and run a model locally. Mix and match: Eaon routes each chat to the right place.
Only when it has to. Hosted models run in the cloud, like any chat app. Local models never touch the network. Your keys sit in the macOS Keychain, and your history is files on your disk you can export or delete anytime.
Every major hosted model (Claude, GPT, Gemini, Llama, Mistral, DeepSeek, Grok, Qwen and more), plus anything you install locally with Ollama, llama.cpp or MLX, and any provider you connect with your own key.
Yes. It's an open project and anyone can pitch in. Star the repo, open an issue, or send a pull request. The download count on this page reads straight from GitHub releases, so the number you see is the real one.
A Mac on macOS 14 Sonoma or later. Local models run best on Apple silicon with enough memory for the model you pick, but the app itself is light.
Email support@eaon.dev and we'll sort it out. The Discord is good for quick questions too. If you'd rather reach the person who makes Eaon directly, for a bug, an idea, or just to say hi, that's sanscreates@eaon.dev.
This started as a request for help covering Apple’s $99/year developer account. That is paid for, and as of 2026.4.0 every Mac download is signed and notarized — no Terminal step, for anybody. The people below are why.
The fee is covered for this year. Anything further goes toward the next one, and toward code-signing the Windows build — that warning is still there.