Claude (Anthropic) outage on September 22, 2026: Elevated errors for multiple models
Claude reported an incident affecting Claude (Anthropic) for 1 hour 38 minutes on 22 September 2026, from 00:57 to 02:35 UTC.
Resolved
Official incident reported by Claude
- Started
- 00:57
- 22 Sep 2026 UTC
- Resolved
- 02:35
- 22 Sep 2026 UTC
- Duration
- 1h 38m
- 1 hour 38 minutes
- Impact
- Major
- as rated by Claude
- Updates
- 5
- posted by Claude
Timeline
All times UTC.
Official incident from Claude
Elevated errors for multiple models
-
Official
Investigating
We are investigating elevated errors on requests to Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5. We will provide an update as soon as possible.
-
Official
Identified
We have identified the cause of elevated errors on requests to Claude Mythos 5.1, Claude Fable 5.1, and Claude Opus 5 and are working on a fix. We will provide an update as soon as possible.
-
Official
Identified
We are continuing to work to resolve errors affecting some models. At this time, requests to Claude Fable 5 and 5.1 and Mythos 5 and 5.1 have returned to normal success rates. We are working to resolve remaining errors affecting Claude Opus 5, and will provide an additional update shortly.
-
Official
Monitoring
We have seen success rates return to normal across affected models, and are monitoring closely to ensure no further issues.
-
Official
Resolved
This issue has been resolved. Impact occurred from 5:50pm PT / 00:50 UTC to 7:10pm PT / 02:10 UTC.
"Official" rows are quoted from Claude's public status page, status.claude.com. Very long posts are shortened.
What our checkpoints saw
HostTracker's checkpoints, which test claude.ai from several locations, recorded no outage of 5 minutes or longer during this incident. A provider can report a problem that affects only some features, regions or accounts, which a check of the main address does not see. The live Claude (Anthropic) status shows what our checks see right now.
Does your product depend on Claude (Anthropic)?
An outage like this one reaches your users through your product, and most teams hear about it from a customer rather than from a dashboard. HostTracker watches any endpoint you name - your own site, an API you call, a vendor you resell - from 300+ locations, confirms a failure from several of them before it says anything, and then tells you by email, SMS, voice call or your chat tool.
You get the same record you are reading here for your own services: every outage timestamped, with the checks that observed it.
Independent measurement by HostTracker. Not affiliated with Claude (Anthropic). Official updates are quoted from Claude's public status page and remain Claude's own words.