Elevated errors across many models

2 min read Original article ↗

Resolved

The incident has now been resolved. We had two phases:
* between 10:23 PT / 17:23 UTC to 11:00 PT / 18:00 UTC -- all Sonnet and Opus models were affected, reaching an error rate of around 10%
* after 11:00 PT / 18:00 UTC and until 12:20 PT / 19:20 UTC -- Opus 4.8 had an average error rate of 10%

The incident is now resolved.

Posted Jun 16, 2026 - 19:32 UTC

Monitoring

A fix has been implemented and we are monitoring the results.

Posted Jun 16, 2026 - 19:22 UTC

Update

We are continuing to work on a fix for this issue.

Posted Jun 16, 2026 - 19:08 UTC

Update

We are continuing to work to resolve an elevated rate of errors on requests to Claude Opus 4.8, and will provide an additional update as soon as possible.

Posted Jun 16, 2026 - 19:00 UTC

Identified

We are still seeing errors on Opus 4.8 and Haiku 4.5.

Posted Jun 16, 2026 - 18:15 UTC

Monitoring

A fix has been implemented and we are monitoring the results.

Posted Jun 16, 2026 - 18:00 UTC

Investigating

We are currently investigating this issue.

Posted Jun 16, 2026 - 17:29 UTC

This incident affected: claude.ai, Claude API (api.anthropic.com), Claude Code, and Claude Cowork.