Anthropic Claude Down Three Hours Across Multiple Models

Anthropic's Claude platform went dark for about three hours in the early hours of August 24, with claude.ai, the Claude API, Claude Code, and Claude Cowork all reporting elevated errors, Android Authority reported. The status page initially named Opus 4.8 before updating to include Opus 5, Mythos 5, and Fable 5, and users hit "529 Overloaded" responses. Claude Console and Claude for Government stayed up.

Engineers began investigating at 05:06 UTC and said they had identified the cause roughly 20 minutes later at 05:27 UTC, though remediation dragged past 06:42 UTC. Anthropic said it was "continuing to work to resolve issues causing elevated requests on multiple models," and later confirmed users would "now be able to use Claude on the web, mobile, desktop, or through the API as usual" as of 8:30 AM UTC. The company did not disclose the underlying technical fault.

This one lands mid-string. The status page has logged disruptions on August 5, 12, 13, 16, 18, and 20, and one tracker flagged the August 5 incident as Anthropic's 164th of the year. It also comes a day after Ramp data we covered showed Opus 5 overtaking Fable 5 in enterprise spend — meaning the outage hit the tier customers are actively migrating onto. It is one more entry in the 334 Anthropic stories we have tracked over the last 90 days.