Anthropic's claude.ai and api.anthropic.com showed elevated errors for 40 minutes on 21 October 2026. The fault sat on Anthropic's side, tied to Claude Fable 5, and was resolved before any user reports were logged during the window.
Started
19:49 UTC
Duration
Lasted 40m
Source
IsDown
Next time
Anthropic is degraded. Is your product working?
An incident on their side does not always mean an outage on yours, and the only way to know is to already be watching your own endpoints. Set that up in under a minute and you will be notified the next time your own site is affected. And there will be a next time.
50 monitors for your own websites. No card.
Affected components: claude.ai, api.anthropic.com
What happened?
Anthropic's status page opened one incident on 21 October 2026, starting at 19:49 UTC. The title named Claude Fable 5, with the first update describing "elevated errors on requests to Claude Fable 5." Two components were listed as affected: claude.ai and api.anthropic.com. Only two updates were filed in total, the opening note at 19:49 and a resolution at 20:29, giving a total span of 40 minutes. The resolution text reads "The issue affecting other Claude models has been resolved," a slightly broader claim than the opening note, which named only Claude Fable 5. No further detail was given on what the elevated errors looked like in practice, how many requests failed, or why the fix took the time it did. No user reports were filed during the window, so there is no outside account of what you would have seen if you were calling the API during those 40 minutes. For context, Anthropic has logged 53 incidents in the past 90 days with a median duration of 1 hour 4 minutes, so this one resolved faster than typical. Over all time, Anthropic has filed 838 incidents, averaging 29.9 a month. This incident sits within that pattern as a short, self-resolved entry with no post-mortem published and no cause given beyond "elevated errors."
Learning
Anthropic's own status page is the only record of this incident, and it names a single model line, Claude Fable 5, with a resolution note that then broadens the scope to "other Claude models" without explanation. That gap matters if you run a product on top of Claude: a vendor's status page can report errors against one model while your own calls, running against a different model, log nothing unusual at all, and both of those things can be true at once. The failure here sits in request handling on Anthropic's side, not in your integration, so checking your own endpoints or client code would have told you nothing about what was happening upstream. UptimeRobot's own 30 day stats on Anthropic show 8 incidents, all resolved, with a median duration of 1h 27m, a different cadence than this 40 minute event, which is a reminder that any single incident's length can differ sharply from the provider's own baseline. Your monitoring covers your half, UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.
