vLLM is not serving right now

The inference service is either starting or intentionally parked for the overnight GPU quiet window.

Active serving window starts at 05:00 Asia/Jakarta. If this appeared during the day, retry shortly.