Skip to content
API Docs

Use i14 Models in Pi

Pi is a terminal AI assistant. Because the i14 API is OpenAI-compatible, you can point Pi at the i14 inference endpoint and use Sugar Glider (i14-sugarglider-nemotron3-super-120B-v1) without changing how you work. Follow the steps below to install Pi, create an API key, and wire everything together.

  1. Install Pi

    On macOS and Linux, open a terminal and run the install command from the Pi homepage:

    curl -fsSL https://pi.dev/install.sh | sh

    The installer puts the pi command in a folder on your PATH, such as ~/.local/bin. On a Mac with Homebrew, it may link pi into Homebrew’s bin folder (/opt/homebrew/bin, or /usr/local/bin on an Intel Mac) without asking. To remove the link, delete pi from that folder.

    If no such folder is on your PATH, the installer puts pi in ~/.pi/agent/bin and asks if it should add that folder to your PATH. Press Enter to accept. It then adds # Pi and an export PATH=... line to your shell startup file, such as ~/.zshrc. To undo it, delete the two lines under # Pi. If you type n, run Pi by its full path, ~/.pi/agent/bin/pi.

    On Windows, see Pi on Windows.

  2. Create an API key

    Sign in to platform.i14.ai and create an API key with access to the model you want to use. Keep the key private and never commit it to source control.

  3. Add the i14 provider to Pi

    Pi reads custom providers and models from ~/.pi/agent/models.json (on Windows, ~\.pi\agent\models.json). Create the folder and the file if they do not exist yet. Then open ~/.pi/agent/models.json in any text editor.

    Add the provider configuration below, replacing YOUR_I14_API_KEY with the key you created:

    {
      "providers": {
        "i14": {
          "baseUrl": "https://api.i14.ai/v1",
          "api": "openai-completions",
          "apiKey": "YOUR_I14_API_KEY",
          "models": [
            { "id": "i14-sugarglider-nemotron3-super-120B-v1" }
          ]
        }
      }
    }

  4. Start Pi

    Start Pi from your terminal:

    pi

    Select the i14 provider and the model you configured. If Pi was already running while you edited models.json, type /reload or restart Pi to pick up the changes.

Troubleshooting

Section titled “Troubleshooting”