Skip to main content

Configuration quick-start

LLMAI uses the OpenAI Realtime API over WebSocket. OpenAI, Grok (xAI), and LocalAI are three provider profiles on the same session model—you configure one place in the editor, then connect with ConnectToAI.

Enable the plugin first: Installation on your projects. Compare provider capabilities (not setup steps) on the Providers comparison table.

Open Project Settings​

All provider defaults live under Edit → Project Settings → Plugins → LLMAI.

Set Default AI Provider to the profile you use most often. You can override provider, model, voice, and instructions per session via ConnectToAI or the Terminal demo connection dialog.

Choose a provider​

You only need one provider configured to get started. Fill in the subsection that matches your choice.

OpenAI (cloud)​

  1. Open OpenAI Provider under LLMAI Project Settings
  2. Enter your API Key (platform.openai.com)
  3. Set Model to gpt-realtime-2 (default for LLMAI v2.2)
  4. Pick a Realtime Voice (for example alloy)

Grok (cloud)​

  1. Open Grok Provider under LLMAI Project Settings
  2. Enter your API Key from console.x.ai
  3. Set Realtime Voice — Ara, Rex, Sal, Eve, or Leo

LocalAI (self-hosted)​

  1. Start a LocalAI Realtime server — see LocalAI Realtime Endpoint
  2. Set Default AI Provider to LocalAI
  3. Under LocalAI Provider, set:
    • Realtime Base URL — e.g. ws://localhost:8080/v1/realtime (use ws://, not http://)
    • Model — e.g. gpt-realtime
    • Realtime Voice — voice ID your TTS backend supports

LocalAI does not require an API key unless your server is configured for authentication.

Verify your setup​

Plugin installed​

  1. Create an Actor Blueprint
  2. Add Component → search for LLMAI Client Component
  3. If it appears, installation succeeded — if not, see Installation troubleshooting

Realtime connection​

  1. Add LLMAI Client Component to an Actor
  2. On Begin Play, call Connect to AI
  3. Use your Default AI Provider, or pass the provider explicitly
  4. For Grok, pass Voice (e.g. Ara)
  5. For LocalAI, ensure the server is running first
  6. Play the level and check Output Log for AI Session Ready

Try a demo (optional)​

If you have LLMAI demo content, open Content/LLMAITerminal/Levels/LVL_LLMAITerminal.umap, play in editor, and connect via the Terminal dialog. See the Demo guide.

Troubleshooting​

  • Upgrading the plugin? See Changelog — Upgrading from a previous version.
  • Confirm the correct provider is selected and API Key (or LocalAI Realtime Base URL) is set.
  • Check Output Log for errors; run llmai.debug.EnableAll in the console for more detail.
  • Cloud (OpenAI, Grok): verify key, credits, network, and WebSocket access.
  • LocalAI: verify the server is running and the URL uses ws://.

For API key priority, environment variables, and advanced settings, see the Configuration reference. Also see the FAQ, Debug and logging, and Installation troubleshooting.