Live check
Is Anthropic down right now?
Observed from Anthropic’s public feed by our independent monitor. Always confirm operational decisions on the official status page.
While you wait: the Anthropic incident checklist
- Check the right page. Anthropic's status lives at status.claude.com (status.anthropic.com redirects there). Components split the API, claude.ai and the Console — your product cares about api.anthropic.com.
- Read the error semantics. 429 means rate limiting; overloaded errors and 5xx mean service trouble. Overload responses during an incident deserve slower, capped retries with jitter — not tight loops.
- Use your second serving path if you have one. Claude on AWS Bedrock or Google Vertex AI is served by those platforms; if you're multi-homed, shift traffic there and watch that provider's status instead.
- Degrade the feature, not the product. Fallback model or provider, cached responses, or a clear 'AI temporarily unavailable' state — switched by feature flag, not discovered by timeout.
- Throttle async consumers. Pause queue workers and batch jobs calling the API; drain after resolution rather than building a retry backlog against a degraded service.
What breaks when Anthropic has issues
Anthropic incidents hit whatever rides the Claude API: chat features, summarization and extraction pipelines, agentic workflows, and coding assistants like Claude Code. Latency spikes upstream of generous timeouts can cascade into worker-pool exhaustion inside your own services. Teams using claude.ai interactively lose a productivity tool; products calling the API lose features — design your degraded mode around the latter.
FAQ
Are claude.ai and the Anthropic API the same thing?
They're separate components on the same status page and can fail independently. Check the API component for product impact, not the consumer app.
Is Claude on AWS Bedrock or Vertex AI down too?
Those are separate serving paths operated by AWS and Google; an Anthropic API incident doesn't automatically affect them, and vice versa. Check the respective cloud provider's status page.
What's the retry etiquette during an incident?
Respect any retry-after header, use exponential backoff with jitter and a cap, and keep requests idempotent. Overloaded errors are a signal to slow down, not to hammer.
Why does status.anthropic.com redirect somewhere else?
Anthropic serves its public status page at status.claude.com; the old hostname redirects there. This page reads the resolved host directly.
Put Anthropic’s status in your toolbar
Is it you — or your vendor? You checked once. The extension watches Anthropic and the rest of your stack continuously — badge, popup and Side Panel, free for 5 vendors, no account.
Add to Chrome — free, no accountNot an outage? Track API changes before they become incidents.
VendorStatus shows when a dependency is failing now. Breakwater watches official developer changelogs and migration deadlines locally in your browser, so your team can plan the change before the next deploy.
Also on your stack?
Not affiliated with Anthropic. Status data comes from their public status page.