Anthropic identified the cause of elevated errors hitting Claude Mythos 5, Claude Fable 5, Claude Opus 5 and Claude Opus 4.8 on August 24, 2026, and said a fix is underway. It had flagged the errors on its status page just after 5 a.m. UTC.
Key Takeaways
- Anthropic identified the cause of elevated errors across four Claude models and says a fix is in progress.
- Claude Mythos 5, Claude Fable 5, Claude Opus 5 and Claude Opus 4.8 are the models named in the update.
- The status moved from Investigating to Identified in 21 minutes, between 5:06 a.m. and 5:27 a.m. UTC.
- Anthropic’s status page has logged 11 other incidents since August 12.
What We Know?
Anthropic titled the incident “Elevated errors for multiple models” on its status page. At 5:27 a.m. UTC, the company said it had “identified the cause of elevated errors on requests to Claude Mythos 5, Claude Fable 5, Claude Opus 5, and other Claude models and are working on a fix.” An earlier update, posted at 5:06 a.m. UTC, had named Claude Opus 4.8 among the affected models too.
On X, the account Downdetector posted that user reports of Claude AI problems began around 1:08 a.m. EDT, tagged #ClaudeAiDown, a timestamp that lines up closely with when Anthropic’s own investigation began.
User reports indicate problems with Claude AI since 1:08 AM EDT.
— Downdetector (@downdetector) August 24, 2026
How is it affecting you? #ClaudeAiDownhttps://t.co/EbXfoHRBJc
What We Don’t Know?
Anthropic hasn’t said what caused the errors or when a fix will ship. The company also hasn’t detailed how many requests failed, or whether claude.ai, the Claude API, Claude Code and Cowork saw the same error rates. As of this writing, the incident remained open at the Identified stage.
SQ Magazine’s Takeaway
Four models failing in that same short window points to a shared dependency, maybe a routing layer or a shared inference cluster, rather than a bug in one model’s weights. That’s harder to route around than a single model failure, especially as Claude usage grows across more products.
Teams running production traffic on any of the four named models should check Claude Status Page before retrying failed requests, since hammering a still degraded backend just adds load.
Anthropic hasn’t given a timeline for a fix, and this incident adds to a status page that has looked unusually busy for most of August.
