Ollama Setup

    Ollama lets you run open-source AI models locally. Combined with Clearminutes, you get fully offline meeting summaries — no API key, no cloud, no usage costs.

    Pro Feature

    Ollama integration requires Clearminutes Pro.

    How It Works

    Clearminutes auto-detects Ollama running on your machine. When you open the Summary settings, it queries Ollama for the list of models you have downloaded and lets you pick one. Summaries are generated entirely on your device.

    Step 1 — Install Ollama

    Download and install Ollama from ollama.com/download for your platform (macOS or Windows).

    After installing, Ollama runs automatically in the background on port 11434.

    Verify Ollama is running:

    ollama list

    You should see a list of installed models (empty if you haven't pulled any yet).

    Step 2 — Pull a Model

    Ollama needs at least one model downloaded before Clearminutes can use it. Open a terminal and run:

    # Recommended — good quality, reasonable size
ollama pull llama3.2

# Lighter option — faster on low-end hardware
ollama pull llama3.2:1b

# Higher quality — needs 16+ GB RAM
ollama pull llama3.1:70b

    Model recommendations by hardware:

    HardwareRecommended ModelWhy
    Apple Silicon (M1/M2/M3/M4), 16+ GBllama3.2 or llama3.1:8bRuns well on unified memory
    Apple Silicon, 8 GBllama3.2:1b or gemma2:2bFits in RAM with room for the OS
    Windows, NVIDIA GPU 8+ GB VRAMllama3.1:8bGPU inference is fast
    Windows, CPU onlyllama3.2:1b or gemma2:2bKeeps inference time reasonable

    You can also browse all available models at ollama.com/library.

    Step 3 — Configure in Clearminutes

    1. Open Clearminutes and go to Settings → Summary
    2. Select Ollama (Local) from the provider dropdown
    3. Clearminutes will automatically fetch models from http://localhost:11434
    4. Select your preferred model from the grid
    5. Click Save

    Using a Remote Ollama Instance

    If Ollama is running on a different machine (e.g. a home server or NAS), you can point Clearminutes at it.

    1. In Settings → Summary, select Ollama (Local)
    2. Expand Custom Endpoint
    3. Enter the URL of your Ollama instance:
      • Example: http://192.168.1.100:11434
      • Example: https://ollama.example.com
    4. Click Fetch Models to load the model list from that endpoint
    5. Select a model and click Save
    Warning

    After changing the endpoint URL, always click Fetch Models before saving — otherwise Clearminutes will try to use a model name that may not exist on the new server.

    The endpoint must start with http:// or https://. An empty field defaults to http://localhost:11434.

    Troubleshooting

    "Ollama is not installed or not running"

    Clearminutes couldn't reach Ollama on the configured endpoint. Check:

    • Ollama is installed and running — try ollama list in a terminal
    • On Windows, look for the Ollama icon in the system tray
    • On macOS, check the Ollama icon in the menu bar
    • Restart Ollama if it's been idle for a long time

    No models appear in the grid

    You haven't pulled any models yet. Run ollama pull <model-name> in a terminal, then click Fetch Models again.

    "Endpoint changed — click Fetch Models first"

    You edited the endpoint URL but haven't refreshed the model list yet. Click Fetch Models, then re-select your model before saving.

    Summaries are very slow

    • Large models (70B+) are slow on CPU — switch to a smaller model like llama3.2:1b or gemma2:2b
    • Enable GPU offloading in Ollama: set OLLAMA_GPU_LAYERS or ensure your GPU drivers are installed
    • On macOS with Apple Silicon, Ollama uses Metal automatically — no extra configuration needed

    Remote Ollama not reachable

    • Ensure Ollama is listening on all interfaces, not just localhost: 
      OLLAMA_HOST=0.0.0.0 ollama serve
    • Check your firewall allows port 11434 from the Clearminutes machine