Quick check: Go to *status.anthropic.com** right now. Active incident? It's them. Green across the board? It's probably you — and there's a fix.*
Claude not responding. The page just spins. Or you're getting a cryptic error message and no output.
Before you assume it's a widespread outage — or before you assume it's something on your end and spend 20 minutes clearing caches — there's one thing to do first. Check the status page.
Step 1: Check Anthropic's Official Status Page
Go to status.anthropic.com.
This is Anthropic's real-time status dashboard. It shows the operational status for all major components:
- claude.ai (the web app and chat interface)
- API (the developer API — claude-3-5-sonnet, claude-3-opus, etc.)
- Console (Anthropic's developer portal)
- API Training (relevant if you're doing fine-tuning work)
The color coding is standard: green means operational, yellow means degraded performance, red means major outage. If there's an active incident, you'll see a banner at the top with a description and a running timeline of updates.
If it shows "All Systems Operational" and Claude still isn't working for you — skip ahead to the local troubleshooting section.
Third-Party Status Checkers
The official status page is controlled by Anthropic, which means they decide what gets reported and when. It's generally accurate and updated quickly, but if you want a second data point:
Downdetector.com aggregates user-submitted reports. When a lot of people start reporting problems at the same time, a spike shows up on the chart — sometimes before Anthropic has officially acknowledged an incident. Not definitive, but useful context.
Searching "Claude down" or "Anthropic outage" on X/Twitter is also surprisingly useful. When Claude goes down, the developer community notices almost immediately. Real-time user reports often surface before the official status page is updated.
Is Claude Down — or Is It Just You?
This is the question that matters most. Some quick diagnostic checks:
Signs it's an Anthropic outage:
- status.anthropic.com shows an active incident
- Downdetector shows a sudden spike in reports
- Multiple people on your team can't access it
- You're seeing HTTP 500 or 503 errors
- The claude.ai page loads but chat won't send
Signs it's a local issue:
- status.anthropic.com shows all green
- Your teammates can access Claude fine
- Problem started after a browser update or extension install
- You're getting login errors or authentication failures
- Clearing cache fixes it temporarily
If the status page is clean and it's just you, the troubleshooting is different. Read on.
Common Claude Error Messages During Outages
A few messages that point specifically toward server-side issues:
"Claude is currently unavailable. Please try again later." — Classic outage message. Check the status page.
"An error occurred. Please try again." (with no other explanation) — Could be a temporary API hiccup or a real incident. Status page check required.
"Overloaded" or "Too many requests" — Claude's API has rate limits per model tier. During high-demand periods, this is throttling rather than an outage. Free-tier users hit this first. Wait 5-10 minutes and retry.
The page loads but messages never send — Usually means the claude.ai web app loaded but the API it depends on is having trouble. Often a partial outage — the front end is up, but model responses are failing.
"Your account is not authorized" — This is account-level, not an outage. Check your subscription status in account settings.
Anthropic's Maintenance Windows
Anthropic doesn't publish a fixed weekly maintenance window like some older infrastructure providers do. Maintenance typically shows up as a scheduled notification on status.anthropic.com a few hours or days in advance. If you're using Claude through the API in a production workflow, it's worth subscribing to status updates so you get advance notice.
Historically, Anthropic tends to schedule any significant maintenance during off-peak hours — late nights Pacific time. But nothing is guaranteed, and they'll announce it on the status page rather than anywhere else reliably.
Anthropic Outage Patterns (What We've Seen)
Complete platform-wide outages where claude.ai is fully inaccessible have been relatively rare. More common patterns:
API degradation without web app impact. The API gets slow or times out for developers while claude.ai continues working fine for regular users. If you're integrating Claude into your own app, this is the scenario to watch for — your users see errors even though "Claude" appears operational.
Model-specific degradations. Sometimes Claude 3 Opus has issues while Sonnet runs fine, or vice versa. The status page shows individual model status, which is more granular than the top-level summary.
Gradual rollout issues. Anthropic ships updates to Claude fairly regularly. Sometimes a rollout causes problems that aren't obvious immediately — response quality degrades, context handling breaks in specific cases. These can show up as "investigating" notices even when the service technically appears available.
Anthropic's incident reports (published after major issues) are actually worth reading if you're depending on Claude professionally. They include honest root cause analysis and what changed.
What to Do While Claude Is Down
If it's genuinely down and you need to get work done:
For writing and analysis: ChatGPT (chat.openai.com) handles most tasks Claude is used for. If you already have an account, it's the fastest fallback. Gemini is good for users in the Google Workspace ecosystem.
For long-context tasks: Gemini 1.5 Pro has a very large context window that's competitive with Claude's. If you're working with long documents, it's a viable temporary substitute.
For coding: For simple code tasks, ChatGPT works well as a fallback. For complex multi-file development work, it's worth waiting for Claude to come back — the alternatives have different strengths and weaknesses.
The practical reality: most outages resolve within an hour or two. If you have a hard deadline, use an alternative for the immediate task. If you can wait, wait.
How to Get Notified of Future Outages
Takes about 60 seconds to set up and saves a lot of "is this me or them?" confusion.
- Go to status.anthropic.com
- Click "Subscribe to Updates" (usually top right or bottom of page)
- Enter your email address
- Select which components matter to you — the API, claude.ai, or both
You'll get an email the moment Anthropic opens an incident and another when it's resolved. If you're using Claude in a workflow where downtime matters, this is just a sensible precaution.
Quick Fixes If It's a Local Issue
Status page is green but Claude's not cooperating? Work through these in order:
1. Hard refresh the page
Cmd+Shift+R (Mac) or Ctrl+Shift+R (Windows). This clears cached JavaScript that can cause the app to behave incorrectly even when the server is fine.
2. Try Incognito/Private browsing mode
Opens a clean session without extensions or cached state. If Claude works in Incognito but not your normal browser, extensions are the culprit.
3. Disable browser extensions
Ad blockers and privacy extensions are the most common cause of claude.ai loading failures when the service itself is operational. Disable everything, test, then re-enable one by one.
4. Clear site cookies
In Chrome: Settings → Privacy and security → Site data → clear data for claude.ai. Then log back in.
5. Check VPN settings
Anthropic's API endpoints can occasionally get rate-limited or blocked from certain VPN exit nodes. If you're using a VPN, try temporarily disabling it and testing directly.
6. Check your subscription/account status
A failed payment or subscription lapse causes account-level failures that look like outages. Check account settings before assuming it's Anthropic's problem.
For deeper troubleshooting — specific error messages, account issues, integration problems — see the full Claude AI common errors guide. And if you want a broader orientation to Claude's capabilities and how it compares to alternatives, the guide to using Claude AI covers the full picture.
