LumenComing soon
LOCAL AI FOR MAC

Your AI.
On your Mac.

Private, offline models with room to think. Run several at once and see how their answers compare.

Coming soon macOS 13+  ·  No account needed
◉   Stays on your Mac⌘   Works offline
Lumen●  Local
Chat ☷
◈   qwen3.5-4b◈   gemma3-1b
What makes a good idea
worth pursuing?
qwen3.5-4b

A good idea connects a real need with a clear path to make things better.

gemma3-1b

Look for curiosity, usefulness, and the energy to keep exploring it.

Ask your models anything… ↑
✳   100% on-device
A QUIETER KIND OF AI

Powerful models.
Nothing leaves home.

Your prompts, files and conversations stay on your Mac. Lumen has no account, no tracking, and no cloud in the middle. Go offline whenever you like.

01Your data stays yoursConversations live only while the app is open.
02Private by designNo telemetry, analytics, or ad cookies.
03Ready for offlineYour models are there when the internet isn't.
MORE THAN ONE POINT OF VIEW

One question.
Different minds.

Load the models that fit your Mac, ask once, and compare their answers side by side. Lumen manages memory as you switch between them.

◈   2 models selectedMemory managed automatically   ●
“Explain local AI to someone who’s never used it.”
qwen3.5-4bThinking · 2.4 GB

It’s like having a helpful assistant built into your computer. You ask a question, and your Mac works out the answer right there—no internet trip required.

ANSWERED IN 2.1s ↗
gemma3-1bFast · 768 MB

Imagine a library that lives on your desk. You can ask it things, and everything stays right there with you—private, and ready anytime.

ANSWERED IN 1.4s ↗
＋Add another perspectiveChoose a loaded model
⌘   Load, unload, and switch in a clickMEMORY   6.4 / 16 GB
BRING YOUR OWN WORKFLOW

Your tools.
Your local models.

Point your favorite coding tools at models on your Mac. One click sets up Claude Code, Codex CLI, or OpenCode. No cloud model bill required for local work.

Claude CodeCodex CLIOpenCode
See how it works →
LOCAL APIRUNNING ON THIS MAC
http://127.0.0.1:11535
POST/v1/chat/completionsOpenAI-compatible
POST/v1/messagesAnthropic-compatible

Works with apps on your Mac. Refuses requests from websites and other computers.

GET STARTED, STAY LOCAL

Three steps.
Then it’s yours.

Everything happens on your Mac, from the first model download to your next big idea.

01
◉

Get Lumen

Install the Mac app. No account or setup wizard to get through.

02
⇩

Choose a model

Browse Hugging Face or import a model from Ollama.

03
✳

Start chatting

Ask a question, compare answers, and find your favorite.

MADE FOR MAC

Starts on macOS 13.
Thrives with memory.

Lumen runs on Macs with macOS Ventura or later. Apple Silicon and more unified memory let you load larger models or keep more models ready at once.

macOS 13 or laterApple Silicon recommendedMore memory, more choice
GOOD QUESTIONS

Before you
get started.

Is my data sent anywhere?＋

Your prompts and conversations stay on your Mac. If you download a model, Lumen contacts Hugging Face to fetch it. You choose if you connect to another server.

Which models can I use?＋

Download compatible models from Hugging Face or import models you already have in Ollama. Availability depends on your Mac's memory and model format.

How much memory do I need?＋

Lumen runs on Macs with macOS 13 or later. Smaller models need less memory; Apple Silicon and more memory make it possible to run larger models and more of them at once.

Does Lumen work offline?＋

Yes. Once your models are on your Mac, you can chat without an internet connection.

Are the coding tools part of Lumen?＋

No. Claude Code, Codex CLI and OpenCode are separate products. Lumen provides a local API connection; each tool has its own policies and may use its own online services.

YOUR MAC. YOUR MODELS.

A little more room
to think for yourself.

Coming soon ↗

For Mac  ·  macOS 13 and later