Service Status
Live operational status of every PixelPrism component. Updated every 5 minutes by an external monitor running on a different machine than the API itself, plus our daily 6 AM UTC drift-monitor canary run.
All systems operational
Detection API
/api/detect (paid) and /api/scan-public (free) — RTX 5090 inference backend on lilliemae behind SSH tunnel
App shell (auth, dashboard, billing)
/dashboard, /signup, /login, /billing, /scan, /scan-free — FastAPI on server1
Static frontend
Marketing pages, leaderboard, share cards — nginx + cached HTML
Webhook delivery (Pro+)
Outbound HMAC-signed POSTs after every Pro+ scan
Stripe billing
Checkout, subscriptions, metered overage events
Email (Postmark)
Verification, password reset, alerts, invoices
API uptime — last 90 days
Each segment = 1 day. Hover for details. Source: external uptime monitor pinging /api/health from lilliemae every 5 min.
Detector accuracy — daily canary
Pulled from /leaderboard. Live numbers update on each retrain (currently V9 / v0.10.0).
Real photo
96.1%
Fresh AI avg
96.6%
Per-gen min
91.0%
Drift gap
−2.7%
Recent incidents + maintenance
2026-05-04 23:09 UTC → 2026-05-05 22:35 UTC · ~23.5h
Scheduled maintenance V9 retrain pipeline
Backend uvicorn was intentionally taken offline to free GPU for the V9 retrain pipeline (Midjourney + PixArt + HunyuanDiT generation, then 16-detector chunked scoring + HGBM fit).
/api/health and /api/scan-public/quota remained 200; only POST endpoints (/api/detect, /api/scan-public) returned 502 during the window.2026-05-06 02:48 UTC
Resolved V9 LIVE — backend fully restored
V9 meta-classifier deployed. Holdout 96.7%, fresh-AI avg 96.6%, per-gen min 91% (Grok). Drift gap flipped from +6.1pp (V8) to −2.7pp.