Installing Ollama

How to install Ollama, the local AI runtime that powers every AI feature in Time Garden Eternal. ~10 minutes, one-time setup.

Ollama is the local AI runtime that powers everything in Time Garden Eternal. Think of it as a small server that runs language models on your own machine — no cloud, no subscription, no data leaving your computer.

Setup is a one-time thing. Maybe 10 minutes total.

1. Download Ollama

Head to ollama.com/download and grab the installer for your OS:

Mac: download the .dmg, drag Ollama to Applications, run it once. A llama icon appears in your menu bar.
Windows: download the .exe installer, run it. Ollama becomes a background service.
Linux: a one-line install script — curl -fsSL https://ollama.com/install.sh | sh

After install, Ollama runs quietly in the background, listening at http://localhost:11434. You don't need to interact with it directly — Time Garden talks to it automatically.

2. Verify It's Running

Open your terminal (Mac: Terminal app; Windows: Command Prompt or PowerShell) and run:

ollama --version

You should see a version number. If you see "command not found", restart your terminal (or your computer) and try again.

3. Pull Your First Model

Now you need to download a model — the actual AI brain Ollama runs.

The recommended starter for Time Garden is phi4-mini (Microsoft's small-and-fast model). Run:

ollama pull phi4-mini

You'll see a download bar. The model is about 2.5 GB. Wait until it finishes.

While that runs, you can read Choosing Your AI Models to decide if you want others.

4. Test It

In your terminal, run:

ollama run phi4-mini

You'll get an interactive prompt. Type "Hello!" and watch the model respond. Type /bye to exit.

If that worked: Ollama is installed and working. 🎉

5. Plug Into Time Garden

Open Time Garden
Open Obsidian Settings (gear icon, bottom-left)
Scroll to Time Garden Plugin in the left sidebar
Verify the connection — most users see a green status indicator confirming Ollama is reachable
Make sure phi4-mini shows up in your installed-models list (use the Add Model button if it doesn't)

If Time Garden doesn't see Ollama, see the AI not responding / Ollama issues page for the diagnostic flow.

6. (Optional) Upgrade To A Smarter Model

phi4-mini is fast but light. For more nuanced summaries and Q&A, pull a heavier model:

ollama pull phi4

That's roughly 9 GB. Or:

ollama pull mistral:instruct

About 4 GB.

Then assign it to specific operations in Choosing Your AI Models.

A Few Things Worth Knowing

Ollama needs to be *running* when you use Time Garden's AI buttons.

On Mac and Windows, the installer sets it up to launch on startup automatically. On Linux you may need to start the ollama serve daemon yourself (or set up a systemd service). If your AI buttons stop working, the very first thing to check is "is Ollama running?"

Disk space.

Models are big. phi4-mini is 2.5 GB; phi4 is 9 GB; some "uncensored" / "chat" / "code" models can be 30+ GB. Pull only what you'll use. You can always ollama rm <model-name> to remove one.

Resource use.

While a model is running, it uses a chunk of RAM (and on Apple Silicon / Nvidia GPUs, a chunk of VRAM). If your computer feels slow during AI operations, that's normal — pick a smaller model in Choosing Your AI Models or close other apps.

Updating Ollama

Ollama auto-updates on Mac/Windows. On Linux, re-run the install script:

curl -fsSL https://ollama.com/install.sh | sh

To update a model: ollama pull <model-name> again — it'll fetch the latest version.

Up Next

Choosing Your AI Models — pick the right model for each operation
Customizing AI Prompts — tune what each AI button asks for
AI not responding / Ollama issues — troubleshooting if something doesn't work

On this page