Aspen: Own your intelligence

🌿 Aspen is in Beta. We're actively improving it and would love your feedback — share your thoughts.

Own your<br>intelligence.

Private AI that runs on your machine. No subscriptions. No cloud. No one reading your conversations. No terminal, no configuration. Just download, open, and ask.

Download Free for Mac<br>See the Aspen device

Also available for Windows &middot; iPhone app on the App Store &middot; Free &middot; No account required &middot; Previous versions

The app is free forever on the devices you already own. The Aspen device is a separate, optional machine for running the biggest models, always on. You never need it to use Aspen.

Not sure yet? Try it free in your browser →

Optional hardware

For when you want the most.

Aspen runs free on the Mac, PC, or phone you already own, and the better your machine, the better it runs. The Aspen device is the top of that range: a dedicated machine that runs the largest models around the clock without touching your computer. It is completely optional. You never need it to use Aspen.

Product rendering. Actual device may vary.

1 Petaflop<br>AI performance

128 GB<br>Unified memory

200B<br>Parameter models

Silent<br>Always on

5.9″ &times; 5.9″ &times; 2″

Set it and forget it<br>It gets better while you sleep.

Aspen isn't static hardware. It's a living system that takes care of itself, so you never have to.

Automatic updates

Software updates install themselves. Security patches apply overnight. You never lift a finger.

Best model, always

When a better open model drops (Llama, Qwen, DeepSeek, Mistral), Aspen switches automatically. You always run the best available.

Instant rollback

If an update doesn't feel right, roll back to any previous version with one tap. Your device, your choice.

Air-gapped by design

Your conversations, documents, and data never leave the box. Not for updates. Not for analytics. Not ever.

Philosophy<br>I don't think any single company should own the most intimate details of a person's life: your medical questions, your financial plans, your private thoughts at 2am. Leadership changes. Cultures shift. Data gets licensed. Terms get rewritten. I think people should at least have the option not to make that trade.

Aspen brings the full power of AI into your home. Your data never leaves. It's never used to train anyone's model. It's yours.

Software<br>Beautiful on every screen.

Chat. Code. Vision. The latest open models, always updated. One app for Mac, Windows, iPhone, iPad, Android, and web.

Live code artifacts<br>Ask for a webpage and watch it render in a preview panel right beside the chat. Build, see it, copy it. Like a local coding assistant.

See your images<br>Drop in a screenshot or photo and a local vision model reads it. One tap installs a vision model if you don't have one. It never leaves your machine.

Connect your tools<br>Connect GitHub and more through open MCP connectors, so your AI can work with your real projects. Tokens stay encrypted on your device.

Reach it anywhere<br>Aspen gives you a secure private URL, so you can chat with the AI on your home machine from your phone, wherever you are.

Qwen 2.5

Llama 3.2

DeepSeek R1

Mistral

Help me write a bedtime story about a fox who learns to fly

Qwen 2.5 &middot; Running locally on your machine

Once, in a forest where the trees grew so tall they tickled the clouds, there lived a fox named Sable who was absolutely certain she could fly. Every evening she climbed the highest hill, spread her russet arms wide, and leapt. Every evening, she tumbled. But one night, a barn owl named Orla landed beside her and said, &ldquo;You're not failing, dear. You're practicing.&rdquo; And that night, for the first time, the wind caught her.

Privacy<br>We didn't add privacy.<br>We removed the server.

Your data never leaves your machine. There is no data to encrypt in transit because there is no transit. There is only Aspen, on your hardware, running on localhost, under your control.

01<br>Zero data collection<br>No telemetry. No analytics. No tracking. We literally cannot see what you do with your device.

02<br>Never used for training<br>Your conversations are never fed into any model. Your data exists to serve you and no one else.

03<br>Works without internet<br>Airplane mode. Off-grid. Air-gapped. Your AI works wherever you go, with or without a connection.

For developersOpenAI-compatible.<br>Two lines to switch.

Aspen exposes a standard OpenAI-compatible API. Change your base_url and api_key, two lines of code, and your entire stack runs privately on your own hardware. No data leaves, because everything runs on localhost. Works with ChatGPT and Claude SDKs out of the box.

CursorVS CodeLangChainContinue.devn8nZapierAny OpenAI SDK

// Two lines. That's it.<br>import OpenAI from 'openai';

const ai = new OpenAI({<br>baseURL: 'https://my.runonaspen.com/v1',<br>apiKey: 'sk-aspen-...',<br>});

const r = await ai.chat.completions.create({<br>model: 'qwen2.5:7b',<br>messages: [{ role: 'user',<br>content: 'Hello'...