Onboarding or a tab shows “No model available” and generation won’t start.
AIOS runs local inference through Ollama, which isn’t installed, isn’t running, or has no models pulled — and no BYOK cloud provider key is configured as an alternative.
You have two paths, and you don’t need a GPU to use either. Path A (local): install Ollama, make sure it’s running, and pull a model from the Models tab. Path B (no local model): skip Ollama entirely and connect a cloud provider key (Claude, OpenAI, etc.) in Settings, or use the Claude Code fallback path from onboarding. If you expected Ollama to be detected and it wasn’t, confirm the Ollama service is actually running before retrying — AIOS checks for it, it doesn’t start it for you.