Claude is having trouble again. At 06:14 UTC on October 25, 2026, Anthropic posted an incident on its status page saying it is investigating elevated errors on requests to five of its Claude models. At last report the problem was not fixed, and no further updates had been posted.
What is happening
The incident is titled "Elevated errors for multiple models." Anthropic named Claude Mythos 5.1, Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5 and Claude Haiku 5.5. It said it would share an update as soon as possible. Seven minutes after posting, the status was still "Investigating." The cause has not been shared.
Who it affects
Four services are marked as a partial outage, meaning some requests fail while others may work. Two other services were listed as working normally.
- Partial outage: claude.ai, the Claude API, Claude Code and Claude Cowork
- Operational: Claude Console and Claude for Government
- The API is how other software and business tools connect to Claude
A rough few days
This is not an isolated event. The status page lists three incidents on October 24, all resolved. Each involved elevated errors on one or more Claude models. The stated impact windows ran from about 11 minutes to 35 minutes. Earlier incidents appear on October 22 (errors affecting Connectors) and October 21 (slower performance on Claude Fable 5).
Over the past 90 days, the page shows uptime of 99.58 percent for claude.ai, 99.64 percent for the API, 99.58 percent for Claude Code and 99.57 percent for Claude Cowork. Claude Console shows 99.84 percent. Claude for Government shows 100 percent.
What to do
If your team or your product relies on Claude, expect some failed requests today. A few simple habits help while this plays out.
- Retry a failed request after a short wait before assuming it is lost
- Have a backup plan for urgent work, such as another tool or a manual step
- Check the Claude Status page for updates before reporting a problem
What happens next
Anthropic has promised an update but has not said when. Earlier incidents on the status page were resolved within minutes to a few hours. Whether this one follows that pattern is not yet known.
The takeawayClaude is having errors across five models today, so retry failed requests, keep a backup option for urgent work and watch the status page.