Anthropic statusLiveFetched seconds agoSource: status.claude.com
Is Anthropic down?
Live status check · All systems operational
Live status pulled from status.claude.com, cached and revalidated every 60 seconds. Below: the components that can break, what to do during an incident, and how to keep your scheduled jobs running independently.
About Anthropic status
Anthropic runs Claude.ai and the Claude API. Incidents are usually scoped to one surface — the Claude.ai web app, the API, the Workbench, or a specific model family — and elevated latency is more common than a hard outage.
Components and surfaces that can break
- Surface 01
Claude.ai (web and mobile).
- Surface 02
Claude API (Messages, batches, file uploads).
- Surface 03
Workbench, Claude Code, the developer console.
Recent Anthropic incidents
Anthropic maintains a public incident history. Read the latest entries to see whether you're hitting a known issue, and look at the post-mortems for context on how long similar incidents typically take to clear.
What to do if Anthropic is actually down
Check whether it's API or Claude.ai
Claude.ai going down doesn't mean the API is down (and vice versa). The status page has both as separate components — confirm which one affects you before you rip out your integration.
Fall back to a different model family
If Claude Opus is throwing 5xx, the Sonnet and Haiku tiers may still be healthy. Provider-agnostic clients (OpenRouter, your own wrapper) make this swap trivial during an incident.
Cache long prompts client-side
Anthropic's prompt caching has a 5-minute TTL by default. If you hit a transient error, the retry from your client benefits from the cache as long as you re-send within the window.
Keep firing through the incident
Schedule your own Anthropic health check with Crontap
Scheduled Claude jobs (daily research agents, weekly summarizations, AI eval pipelines) are HTTP cron under the hood. Crontap calls your handler at the cadence you set, retries on 5xx and 429, and stores the response — so a transient Anthropic blip doesn't kill the whole nightly run.
External HTTP cron hits your endpoint on the cadence you pick, stores every response, and emails you the moment a run fails. Pro schedules down to 1 minute; $3.25/mo annual flat for unlimited jobs.
Go deeper
Anthropic status: FAQ
- Where do I check Anthropic status?
- status.anthropic.com publishes component status for Claude.ai, the API, the Workbench, and Claude Code. The history page has incident records back several months.
- Why is the Claude API slower than usual?
- Elevated latency without a full outage usually means the inference fleet is under load. The status page flags 'degraded performance' separately from outright outages. Backoff + retry is the right behavior; failing over to another provider is overkill for a 30-second latency spike.
- How do I schedule a daily Claude job that doesn't break on hiccups?
- Wrap the Claude API call in your own HTTP endpoint, set Crontap to fire it daily, and let Crontap retry on 5xx. The retry catches the transient blip; the alert email catches the persistent failures.