← All posts · News
Claude API had elevated errors for 2.5 hours on November 13
Anthropic rated it a major outage across the Claude API, claude.ai and Claude Code.
Anthropic's Claude API had elevated errors for about two and a half hours on November 13, 2026. The incident on Anthropic's status page ran from 13:09 to 15:38 UTC and was rated a major outage. It also hit claude.ai, Claude Code and Claude Cowork.
What changed
- 13:09 UTC: Anthropic began investigating degraded performance on claude.ai, the Claude API and Claude Code. By 14:03 the incident also covered connectors (including Claude Docs and Claude Sheets), Claude Security and Claude Managed Agents.
- 14:36 UTC: claude.ai and the Claude API began to recover, with a warning that errors could return.
- 15:33 UTC: errors on the Claude API, Claude Code and claude.ai were back to normal levels. Scheduled tasks and routines could run more than two hours late, and adding a new connector on claude.ai was temporarily unavailable.
- 15:38 UTC: Anthropic marked the incident resolved.
- Earlier that day, from 09:58 to 11:58 UTC, a separate major incident affected sign-in to claude.ai and Claude Code and listed the Claude Console (platform.claude.com) as affected.
Anthropic's status updates give no cause for either incident.
What to do
- Check API logs from 13:09 to 15:38 UTC on November 13 for failed requests and replay any jobs that did not complete.
- Treat 5xx errors as retryable, with exponential backoff and a cap, so a short incident does not drop work.
- Keep a tested fallback model from another lab for traffic that cannot wait. With RouterPlus, Claude models and models from other labs sit behind one key at
https://api.routerplus.com/v1, so switching is a change to themodelfield. - Subscribe to updates on status.claude.com.
