Status

Some systems degradedDegraded

Live snapshot of the Perchy routing layer, console, payments, and the upstream model providers we route to. We post incidents within five minutes of detection. Subscribe for updates via the link below.

Subscribe to RSS Subscribe to email

Components

System health

Routing layer (api.perchy.ai)

Chat, embeddings, lane reservations, and models endpoints. · Operational — p95 routing add-latency 23 ms.

Operational

Customer console (perchy.ai)

Sign-in, billing, API keys, usage dashboard. · Operational.

Operational

Stripe payments

Checkout, usage extensions, payouts, webhook delivery. · Operational.

Operational

Anthropic upstream

Claude family routing. · Operational.

Operational

OpenAI upstream

GPT family routing. · Operational.

Operational

Google upstream

Gemini and Gemma routing. · Operational.

Operational

DeepSeek upstream

DeepSeek V4 / R2 routing. · Elevated cold-start latency on Asia-Pacific routes — investigating since 14:02 UTC.

Degraded

Marketplace hosts

Aggregated GPU host fleet across regions. · Operational — 412 hosts online, 81% lane utilization.

Operational

Recent incidents

Last 90 days

Brief 503s on api.perchy.ai during region failoverResolved

April 30, 2026 · 12 min · INC-2026-04-30-01

Automated failover from us-east to us-west returned 503 to a fraction of streaming requests. Mitigated by tightening the health-check threshold; full RCA published.

Stripe webhook delivery lagResolved

April 12, 2026 · 47 min · INC-2026-04-12-02

Usage extension confirmations took up to 47 minutes to apply during a Stripe-side queue backlog. No payments were lost; account state reconciled automatically.

Looking for SLA terms?

Premium and Business plans include a 99.9% routing-availability commitment and SLA credits documented in our refund policy.

Read SLA terms