Is Claude Down? How to Check Claude's Status

Is Claude AI down? How to check Claude's official status page, what each component means, how to tell an outage from a local issue, and what to do meanwhile.

Josh Pocock6 min read

Is Claude down? cover

AI Coding

Is Claude down?

Stride AI Academy

To check if Claude AI is down, go to Anthropic's official status page at status.claude.com. It lists claude.ai, the Claude API, Claude Code and the other Claude services separately, each marked Operational, Degraded Performance, Partial Outage, Major Outage or Maintenance. If everything shows Operational but Claude still won't respond, the problem is more likely your plan's usage limit, a capacity message or your own network.

This guide covers what the status page shows, how to subscribe to alerts, how to tell an outage from a local problem in claude.ai, Claude Code and the API, and what to do while Claude is down. Everything here was checked on November 22, 2026 against Claude Status, the Help Center article on troubleshooting error messages, Claude Code's errors, commands, authentication and model configuration docs, and the API errors docs.

How to check Claude AI status

Open status.claude.com. On November 22, 2026, the page lists these components:

  • claude.ai
  • Claude Console (platform.claude.com)
  • Claude API (api.anthropic.com)
  • Claude Code
  • Claude Cowork
  • Claude for Government
  • Claude on Cloud Platforms, Claude Platform on AWS, Claude in Microsoft Foundry and Claude Models on Google Cloud (Anthropic managed)

So "Claude Code down" and "Claude AI down" can be different answers. Check the row for the thing you're actually using. The page also shows 90 days of uptime bars and past incidents.

Get alerts instead of refreshing

The status page lets you subscribe by email, SMS, Slack, Microsoft Teams, webhook, or an Atom or RSS feed. If you run anything important on the Claude API, a webhook or Slack alert means you hear about an incident before your users do.

Is it Claude, or is it you?

The Help Center separates two things that look similar:

  • A service incident. Claude is unavailable or badly degraded for all or most users. These show up on status.claude.com with their scope and progress.
  • A capacity message. "Due to unexpected capacity constraints, Claude is unable to respond to your message. Please try again soon." The Help Center calls this normal load management, not an outage, and says it won't appear on the status page. Wait a few minutes and try again.

Other messages are about your account, not an outage:

  • "5-hour limit reached" or "You've reached your weekly limit" means you've used your plan's allowance. See how Claude usage limits work.
  • "Your message will exceed the length limit for this chat" means the chat is too long. Split the content or start a new chat.
  • "There was an error logging you in" is often local. The Help Center suggests turning off any VPN, disabling browser extensions and clearing cache and cookies, then checking the status page if it persists.

Is Claude Code down?

Claude Code gives you more to work with, because its error messages tell you which kind of failure you have.

Check your own setup first

Run these inside Claude Code:

Terminal
/status
/doctor

Per the commands docs, /status shows your version, model, account and connectivity, and it works even while Claude is responding. /doctor runs a setup checkup that diagnoses installation and configuration problems and proposes fixes. If /status shows the wrong login method, that's your problem, not an outage. A common one is an ANTHROPIC_API_KEY in your environment taking over from your subscription. The authentication docs say to run unset ANTHROPIC_API_KEY and check /status again.

What the errors mean

From the Claude Code errors docs:

Message What it means
API Error: Repeated 529 Overloaded errors The API is at capacity. Claude Code points you to status.claude.com
Opus is experiencing high load Same, for one model
Server is temporarily limiting requests or Request rejected (429) Temporary throttling. Claude Code retries these
You've hit your session limit or weekly limit Your plan quota is used up. Wait for the reset
API Error: 500 Internal server error A server-side fault. Retry
Unable to connect to API, Connection refused Check your network, proxy and firewall
SSL certificate verification failed Usually a corporate proxy. Add your CA to NODE_EXTRA_CA_CERTS

So a 529 points at Anthropic, a 429 usually points at your limits, and a connection or certificate error points at your network. If Claude Code fails to start at all with an exit code, see our fix for Claude Code process exited with code 1.

Claude Code already retries

The errors docs say Claude Code retries transient failures (server errors, overloads, timeouts, dropped connections) up to 10 times with exponential backoff. You'll see Retrying in Ns · attempt x/y on the spinner. For unattended jobs such as CI, you can make it keep retrying 429 and 529 errors:

Terminal
export CLAUDE_CODE_RETRY_WATCHDOG=1

You can also give 529s a longer backoff:

Terminal
export CLAUDE_CODE_OVERLOADED_RETRY_MAX_DELAY_MS=120000

Switch models when one is overloaded

When one model is under heavy load, another may be fine. The model configuration docs describe fallback model chains: when the primary model is overloaded or unavailable, Claude Code can switch to the next one for that turn.

Terminal
claude --fallback-model sonnet,haiku

The switch lasts for the current turn only, and rate-limit or billing errors never trigger it. You can also switch by hand with /model.

Is the Claude API down?

For your own apps, the HTTP status code tells you most of it. From the API errors docs:

  • 529 overloaded_error: the API is temporarily overloaded. The docs note this can happen during high traffic across all users.
  • 500 api_error: an unexpected internal error. Retry with exponential backoff, and contact support with the request ID if it persists.
  • 429 rate_limit_error: your organization hit a rate limit or spend cap. That's not an outage.
  • 401 authentication_error: a problem with your API key, such as an expired or revoked key.

The official SDKs already retry 429s and 5xx errors twice by default. Our guide to Claude API rate limits has retry code and explains 429 vs 529 in detail. Log the request_id from every failed response so support can trace it.

What to do while Claude is down

  • Check the right component on status.claude.com and subscribe to updates.
  • Wait out capacity messages. The Help Center says to wait a few minutes and try again.
  • In Claude Code, let retries run or set a fallback model.
  • Save your work. If a response cuts off mid-way, the Claude Code docs say it keeps the completed part. Ask Claude to continue once service is back.
  • For production apps, queue requests and retry with backoff rather than failing hard. Work that can wait can go through the Batch API.

Anthropic doesn't publish an outage schedule, and we don't track incident history here. The status page is the source for that.

Learn to build with Claude Code

The Stride AI Coding course teaches Claude Code and Codex for building client tools, written for non-programmers, with weekly live calls to review your builds. Many of the free guides cover Claude Code setups, and the Claude Code community is where members compare theirs. Start with the Claude Code guide if you're new.

FAQ

How do I know if Claude AI is down?

Check status.claude.com, Anthropic's official status page. It shows each service separately, including claude.ai, the Claude API and Claude Code. If every component is Operational, the problem is more likely a usage limit, a capacity message or your network.

Why does Claude say it's at capacity when the status page is green?

The Help Center says the "unexpected capacity constraints" message is normal load management, not an outage, so it doesn't appear on the status page. Wait a few minutes and try again.

Is Claude Code down or is it my setup?

Run /status and /doctor in Claude Code. A 529 overloaded error points to Anthropic, a 429 or session limit message points to your usage, and connection or certificate errors point to your network or proxy.

What is the difference between a Claude 429 and 529 error?

A 429 means your organization or plan hit a limit. A 529 means the API is temporarily overloaded for everyone. Both are worth retrying with backoff, but a 429 with no retry-after header on the API usually means a spend cap, which retrying won't fix.

Can I get notified when Claude goes down?

Yes. Subscribe on status.claude.com by email, SMS, Slack, Microsoft Teams, webhook or RSS and Atom feed. That way you hear about incidents without refreshing the page.

Build it with us in Stride AI Academy

Four courses, weekly live calls and a community of people building and selling AI. Or start with 320+ free guides.

Keep reading

Free AI Coding guides