New B200 spot capacity is live in US East from $1.69 per GPU-hour. See availability
Partial outage

Elevated API error rates

About 9.5% of API requests failed for 22 minutes.

Reference 8b52c078c575 · Updated just now

Status
ResolvedResolved — last update 25 May 2026, 21:45 UTC
Started
25 May, 21:23Monday 25 May 2026 UTC
Duration
22 minEnded 25 May 2026, 21:45 UTC
Components
2API, Console
Timeline

What we posted, as we posted it.

Resolved4 updates · UTC
  1. ResolvedThis incident has been resolved. Requests that failed were not billed and can be retried safely.25 May, 21:45
  2. MonitoringError rates returned to normal at 21:39 UTC. We are monitoring while the node is rebuilt and put back behind the load balancer.25 May, 21:39
  3. IdentifiedThe errors come from one API node that kept serving after a routine deploy with an exhausted database connection pool. The node has been removed from rotation.25 May, 21:28
  4. InvestigatingWe are investigating elevated error rates on the API. About 9.5% of requests are receiving 5xx responses. The console is not affected.25 May, 21:23

Affected components

Details

Type
Partial outage
Started
25 May 2026, 21:23 UTC
Ended
25 May 2026, 21:45 UTC
Impact
About 9.5% of API requests failed for 22 minutes.
Reference
8b52c078c575

Follow the Atom feed