Home / Setup Guide

Up and running in a few minutes

CereVault Setup Guide

CereVault is the interface. You choose where the models run. Pick one of the three paths below, download a model, and start chatting. Everything stays on your device or your trusted local network.

Download CereVault Choose a setup
Step 1

Choose how your models run.

CereVault connects to a local model provider. Each option below is private and runs on your own hardware, so pick whichever fits you.

CereVault Serve Easiest on Apple

An all-Apple setup. A one-time Mac App Store purchase, a curated model catalog, and built-in sharing to your iPhone and iPad. Jump to the steps ↓

Ollama Free & popular

A free, widely used local model runner with a large library. Great if you want maximum model choice. Jump to the steps ↓

LM Studio Free & flexible

A free desktop app for downloading and running models, with a built-in server CereVault can detect on your machine and across your local network. Jump to the steps ↓

Option 1 · CereVault Serve

Run models on your Mac, share to your devices.

  1. Install CereVault Serve

    Get CereVault Serve from the Mac App Store and open it. It lives in your menu bar, with no Dock icon or windows in the way.

    Get CereVault Serve →

  2. Pick a model from the catalog

    Choose from the curated catalog of Llama 3.2, Gemma 2, Qwen 2.5, and Phi 3.5. It downloads directly to your Mac (about 1–3 GB each).

  3. Turn on sharing (optional)

    Want to use CereVault on your iPhone or iPad? Flip the device-sharing toggle in CereVault Serve. Keep your devices on the same Wi-Fi network.

  4. Open CereVault

    CereVault automatically discovers CereVault Serve on your machine and your local network, and prefers it as the default. Pick your model and start chatting.

Option 2 · Ollama

Run models with Ollama.

  1. Install Ollama

    Download Ollama for your platform and install it.

    Download Ollama →

  2. Download a model

    Pull a model from the Ollama library. For a small, fast starting point, run:

    $ ollama run llama3.2

    Browse more options in the Ollama model library.

  3. Keep Ollama running

    Leave Ollama running in the background. It serves models locally at http://localhost:11434.

  4. Open CereVault

    CereVault finds Ollama on your machine automatically. Choose a downloaded model from the picker and start chatting.

Option 3 · LM Studio

Run models with LM Studio.

  1. Install LM Studio

    Download LM Studio for your platform and install it.

    Get LM Studio →

  2. Download a model

    In LM Studio, search the built-in catalog and download a model to your machine.

  3. Start the local server

    Open LM Studio's Developer (Local Server) tab and start the server. To reach it from your other devices, turn on serving over your local network.

  4. Open CereVault

    CereVault automatically detects LM Studio servers on your machine and across your local network. Pick a model from the picker and start chatting.

Step 2

Pick a model and start chatting.

CereVault automatically finds AI servers on your network, including CereVault Serve, Ollama, and LM Studio, and connects in seconds. Switch models at any time, even mid-conversation, without losing your flow. Attach screenshots or photos and a vision-capable model can read them.

Your chats stay on your device. No accounts, no tracking, no cloud required.

Learn more about CereVault →

Troubleshooting

  • No models listed? Make sure your server (CereVault Serve, Ollama, or LM Studio) is running and at least one model is downloaded.
  • Using another device? Keep it on the same Wi-Fi and turn on sharing in CereVault Serve.
  • Slow first reply? Models can take a moment to load on first use, then it speeds up after that.
  • Still stuck? Get in touch.

Ready to go

Download CereVault and connect to your local model in minutes.

Download on the Mac App Store