Key takeaways

  • Claude returned elevated errors from 07:32 to 08:45 UTC on November 16, according to Anthropic's status page.
  • The incident hit Claude Sonnet 5.5 and Claude Fable 5.2 on claude.ai, the Claude API, Claude Code and Claude Cowork.
  • It was the fourth Claude incident on the status page since November 13, and Anthropic has not given a cause.

Claude was down for about 73 minutes on Monday, November 16. Anthropic's status page says requests to Claude Sonnet 5.5 and Claude Fable 5.2 returned elevated errors from 07:32 to 08:45 UTC (23:32 to 00:45 Pacific), across claude.ai, the Claude API, Claude Code and Claude Cowork. The incident is marked resolved.

The timing hit Europe's Monday morning and late Sunday night in the US. Anyone who saw failed replies, stalled Claude Code sessions or API errors in that window was seeing this incident, not a problem on their own machine.

What the status page says happened

The incident was titled "Elevated errors for Claude Sonnet 5.5 and Claude Fable 5.2". Anthropic posted three updates, all in UTC:

  • 07:49: investigating the issue, 17 minutes after errors began.
  • 08:02: a fix deployed "for Claude Sonnet 5.5 and other Claude models", with the team watching for recovery.
  • 09:04: resolved, with the impact window given as 07:32 to 08:45 UTC.

Anthropic has not said what caused it. The status page lists the affected surfaces but not what share of requests failed. Sonnet 5.5 is the default model for many Claude users, and Fable 5.2 is the flagship Anthropic released on November 9 (our report), so most chat and coding traffic runs through one of the two.

The fourth Claude incident in four days

This was not an isolated blip. The same status page lists three other incidents since Friday:

  • November 13, 09:58 to 11:58 UTC: sign-in errors on claude.ai and Claude Code.
  • November 13, resolved at 15:38 UTC: elevated errors on claude.ai, the API and Claude Code, with connectors, Claude Meetings and Managed Agents degraded and some scheduled tasks skipped.
  • November 14, 02:34 to 04:03 UTC: purchases and plan changes failing on claude.ai.

A Sign in with Apple problem on the iOS app, mainly affecting new accounts, was still listed as under investigation in Anthropic's last update on November 14.

What to do if Claude is down again

Check status.claude.com first. It is the only source that says whether a problem is on Anthropic's side, and you can subscribe there for email or text alerts. During an incident, a request that fails can usually be retried once the status page moves to "Monitoring". API users who need to stay up can retry with backoff, or fall back to a different Claude model while one is degraded.

The next thing to watch is whether Anthropic publishes a cause. Its status updates for this incident gave none.

Sources

  1. Elevated errors for Claude Sonnet 5.5 and Claude Fable 5.2Anthropic
  2. Claude status incident historyAnthropic

Add ClaudeGPT Si as a preferred source on Google