✦ AI

Credits & Handoff

How credits work and what happens when they run out

Archon uses a credit system to manage AI usage across multiple providers. Credits are consumed per request based on the model and token count.

How credits work

→ Each AI request costs credits based on the model used
→ Faster, smaller models cost fewer credits
→ Your remaining credits are shown in the chat header
→ Credits refresh monthly on your billing cycle

Automatic handoff

When your credits for the active model run low, Archon automatically switches to the next available model. You don't need to do anything — the handoff is seamless.

Handoff priority:

  1. GPT-4o         → if credits remain
  2. Claude 3.5     → if GPT-4o credits exhausted
  3. Gemini 1.5     → if Claude credits exhausted
  4. Ollama (local) → always available, no credits needed

  Archon tries cloud models first (best quality).
  Falls back to local models if all cloud credits are spent.

What you'll see

When a handoff happens, a small notification appears in the chat: "Switching to Claude 3.5 Sonnet — GPT-4o credits exhausted." Your conversation context is preserved and the AI continues where it left off.

Using Ollama for free

If you run Ollama on a local machine or server, you can use Archon with zero credits. Connect via SSH and Archon will route all AI requests to your local hardware.

# In Archon Settings → AI Provider:
  Provider: Ollama
  Host: 192.168.1.100
  Port: 11434

# No credits consumed. Fully offline. Fully private.

Monitoring usage

→ Chat header — shows remaining credits and current model
→ Settings → Usage — detailed breakdown by model and date
→ Status page — relayapp.pro/status — real-time system health