Anthropic logged one incident on 5 October, elevated errors on requests to Claude Mythos 5.1 and Claude Fable 5.1. It lasted 17 minutes and the fault sat on Anthropic's side, not yours.
Started
13:05 UTC
Duration
Lasted 17m
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 shows a single incident starting at 13:05 UTC, affecting claude.ai and api.anthropic.com. The title names two specific models, Claude Mythos 5.1 and Claude Fable 5.1, so this was not a blanket platform failure but a model-level error rate problem. The first update says Anthropic was investigating elevated errors on requests to those two models. Sixteen minutes later, at 13:21 UTC, Anthropic marked it resolved, and the resolution note gives a different window: impact ran from 05:40 PT to 06:10 PT, which is 12:40 to 13:10 UTC, so the actual disruption predates the public "investigating" post by 25 minutes. That gap between when impact began and when Anthropic first told anyone is common with status pages and is worth noting on its own. There are only two updates in total on this incident, investigating and resolved, with nothing in between describing cause. No root cause is given anywhere in either update. No user reports were filed during this window on record. Anthropic's own 90-day history shows 82 incidents with a median duration of 1 hour 9 minutes, so this one, at 17 minutes, resolved faster than most. Across all time Anthropic has logged 835 incidents, averaging 29.8 a month, which puts a single short model-specific error spike in a fairly busy pattern for this provider.
Learning
This incident only affected requests to two named model versions, Claude Mythos 5.1 and Claude Fable 5.1, which is the kind of failure that will not show up if you are only pinging a generic health endpoint or checking that claude.ai loads. A basic uptime check against the homepage or a simple API ping can stay green while a specific model behind it throws elevated errors to real requests, because the front door opens fine even when one room behind it is broken. That is exactly the shape of this outage: narrow, model-specific, and resolved before most synthetic checks would even register a second failed poll. Anthropic's own record shows this pattern repeats often, 82 incidents in 90 days, so treating any single green check as proof the whole platform is healthy will eventually bite you. Our 30-day data for Anthropic shows 9 incidents with a median duration of 1h 39m, a different and heavier pattern than this short 17-minute spike, which tells you duration varies a lot incident to incident and a single sample tells you little. Your monitoring covers your half, UptimeRobot now watches the provider's half too, so next time you know which side broke without guessing.
