Is Suede Agent Studio up right now?
This page reads live checks of the Studio API, the model gateway, and x402 settlement, and shows availability measured from real data — not a marketing number. Every figure below is paired with the window it was measured over.
Availability counts every check that was not a major outage. Checks run hourly, so a shorter outage can hide inside one recorded hour — no percentage is published until a window is meaningfully filled.
What “production-grade” actually means here.
- Served from Vercel's global edge network.
- Every flow runs in a separate Test environment before a saved checkpoint is promoted to Live.
- Saved versions are immutable; promotion is release-style.
- Settlement is dry-run by default.
- Every run enforces a per-run USDC cost ceiling and a per-agent daily cap.
- A failing node halts its branch instead of charging for downstream work.
- Every run writes a per-step USDC cost ledger you can audit.
More on what these checks probe and why we publish no uptime number yet: the reliability explainer.
Shown as throughput, not a graded success rate — a run can fail on a bad prompt or a broken draft, which is a user outcome, not a platform outage.
Monitors and agents can poll /api/health (200 when up, 503 on a major outage) or /status.json for the same figures shown above.