System Status
Honest availability reporting — refreshed live every 60 seconds from your browser.
Checking…
probing /api/health
Latency
—
How we run — the transparent version
- ✓Self-hosted, owned hardware. Inference runs on machines we own behind Cloudflare Tunnel — no rented GPU cloud, no per-request third-party dependency. This is how 20,000 images can cost $79.
- ✓CPU inference. Typical processing time is 1–3 seconds per image. We publish the real per-request number in the
X-Processing-Time-Msheader — verify it yourself in the Playground. - △No SLA contract. We do not promise an uptime percentage we can't contractually back. What we do instead: this page, honest incident notes below, and a service that auto-recovers within seconds of a crash.
- ✓Zero data retention. Images are processed in volatile memory only — no disk writes, no archives. Nothing to lose in an incident.
Incident history
tracking since 2026-09-21Public tracking begins 2026-09-21. The 30-day availability window below fills in as data accrues — we won't backfill invented history.
No incidents recorded
—
Monitored endpoints
POST /v1.0/removebgcore inference
GET /api/healththis page's probe
GET /v1.0/accountcredits lookup