Overview
- Anthropic opened an incident after detecting elevated errors on August 24 and began investigating at about 5:06 a.m. UTC.
- The company said the errors affected multiple models, specifically Claude Mythos 5, Claude Fable 5, Claude Opus 5 and Claude Opus 4.8.
- The outage hit several services used by customers and developers, including Claude chat, the Claude API, Claude Code and Claude for Workspaces (Cowork).
- Downdetector and user reports showed a clear spike in complaints during the disruption, with more than 1,300 reports in the U.S. and over 180 in India describing failed requests, error messages and slow or interrupted responses.
- Anthropic identified the cause, implemented a fix and moved its status page to operational, though some users may still see intermittent problems as recovery rolls out and the incident is monitored following earlier August disruptions on the 18th and 20th.