Claude Down? How to Check Status Page and Error Messages

To check whether Claude is down, open status.claude.com. If an incident is posted on Anthropic's official status page, the problem is on the service side. Support points to this page for confirming service outages. When you cannot connect, there are three main causes. A full service outage, the usage limit on your own plan, or a capacity shortage, where demand is too high to answer but nothing is technically broken. The screen message differs in each case. Reading it tells you whether to wait or whether you can fix it yourself.

Claude down or not working: tell the cause from the screen message

The usage limit guide in Anthropic's help center sorts error messages by cause. Read one message exactly and your next step is clear.

Claude screen messages, causes and what to do (based on the help center guide)

Message on screen	Cause	What to do
"5-hour limit reached - resets [time]"	You used up the 5-hour session limit	Wait until the reset time
"You've reached your weekly limit. It resets [day] at [time]."	You used up the weekly limit	Wait for the reset, or if your plan supports it, turn on usage credits on the usage screen in settings
"5-hour limit resets [time] - continuing with usage credits"	You passed the limit but work continues on usage credits	Nothing to do. Just check your credit balance and monthly spend limit
"Due to unexpected capacity constraints, Claude is unable to respond to your message. Please try again soon."	Heavy demand from all users on the infrastructure	Resend in a few minutes. The help center says this is not an outage, and it does not appear on the status page
"There was an error logging you in"	Login failed (unspecified cause)	Turn off your VPN and browser extensions, then clear cache and cookies. If it continues, check the status page
"Your message will exceed the length limit for this chat."	The message you sent is too long	Split it, summarize it, or open a new chat

Limit messages always include a reset time. If you see a time, it is a limit on your plan, not an outage. To see how large each limit is, the Claude plan price and usage limit comparison covers the differences between Pro, Max, and Ultra.

Check whether Claude is down on status.claude.com

If the message has no limit wording, open the Claude status page. Active incidents appear at the top as an orange bar. Inside the bar, updates stack in time order: investigation started, cause identified, recovery. Times are in UTC, so add 9 hours to get Korea Standard Time.

Here is how to read that incident. The first post (Investigating) is 13:09 UTC on November 13, which is 22:09 in Korea. The impact then spread from claude.ai, the API, and Claude Code to the Docs and Sheets connectors, managed agents, and Claude Security. The 14:36 UTC update (23:36 in Korea) says recovery is visible but errors may return. Earlier the same day, at 11:58 UTC (20:58 in Korea), a separate incident with errors on claude.ai, Claude Code, and platform.claude.com was resolved. The incident history on the status page lists 5 incidents from November 8 to 13.

Service components to check on the status page and what they cover

Component	What it covers
claude.ai	Web chat. The mobile app is said to be affected in the same way
Claude Console (platform.claude.com)	Developer screen for managing API keys, usage, and billing
Claude API (api.anthropic.com)	Developer calls that build Claude into a service
Claude Code	Coding tool for the terminal and desktop
Claude Cowork and others	Cowork, Claude for government, and Claude on cloud platforms are listed separately

Check whether your product line shows Operational, Partial Outage, or Degraded Performance. As the help center notes, a capacity shortage is not an outage, so it does not appear on the status page. If there is no incident bar and every line is normal but Claude still fails, it is a capacity shortage or a problem in your own environment. Teams that care about outages can use the Subscribe button on the status page to get email, text, Slack, Teams, and webhook alerts.

Claude not working but no outage: what to check

If the status page is normal, first check whether the message is the capacity one. For that message, resend after a few minutes. If you keep getting stuck in the same chat, you can open a new chat. For a login error, follow the order the help center gives.

1. Turn off your VPN if it is on.
2. Turn off browser extensions, or open Claude in an incognito window.
3. Clear the browser cache and cookies, then log in again.
4. If it still fails, check the status page again.

There is a simple way to find the cause yourself. Open the same address in an incognito window too. If it works there but not in the normal window, an extension or cache is the cause, so finish the steps above. If it fails in both, the problem is on the service side, not yours. The success test is getting a reply when you send a short message in a new chat. If only long messages fail, suspect the length limit or a capacity shortage. This check is the author's suggestion, not from the help center.

Claude Code error messages and fixes

Claude Code error messages are more specific. The Claude Code error docs explain them as follows.

Claude Code error messages, causes and fixes (based on the official error docs)

Error message	Meaning	Fix
API Error: 500 Internal server error	Internal Anthropic error. Not caused by your prompt or settings	Check the status page, wait, then resend. If there is no incident notice, report it with /feedback
API Error: Repeated 529 Overloaded errors	The API is at capacity. Claude Code already retried and it does not count against your limit	Wait a few minutes or switch to another model with /model. Capacity is tracked per model
Opus is experiencing high load	Load is concentrated on one model	Pick another model such as Sonnet in /model
Request timed out	No response within the default 10 minutes	Try again. On slow networks, raise API_TIMEOUT_MS
Not logged in · Please run /login	No login session	Run /login
Server is temporarily limiting requests, Request rejected (429)	Request limiting	Try again shortly. A temporary 429 is retried automatically

The per-model capacity point is the most useful in practice. If 529 keeps repeating on the top model, switch one tier down with /model, keep working, and come back once it recovers. If you are new to Claude Code, start with the Claude Code first task and permission mode guide, and check plan limits in the Claude Code install, pricing, and usage limit summary. If you are blocked by a full limit, the guide to switching between multiple Claude Code accounts explains how to split usage across accounts.

API error codes developers should tell apart: 429 and 529

If you build Claude API into a service, you need to tell the codes in the API error docs apart to retry correctly. A 429 is a problem with your organization. A 529 is a problem with overall traffic.

Claude API HTTP error codes (based on the official error docs)

Code	Type	Meaning and fix
400	invalid_request_error	Problem with the request format. Hitting a spend limit set on an organization or workspace also returns 400
401	authentication_error	The API key is wrong or revoked
402	billing_error	Problem with billing information. Check in the Console
429	rate_limit_error	Your organization hit a rate limit or the monthly spend cap. A spend-cap 429 has no retry-after header and keeps failing until access is restored
500	api_error	Internal Anthropic error. Retry with exponential backoff, and contact support with the request ID if it continues
529	overloaded_error	Temporary overload from overall user traffic

Every response carries a request-id header. Send this value when you contact support so they can find the request quickly. The official SDKs automatically retry connection errors, rate limits, and 5xx errors twice by default, and follow the retry-after header when present. If you get a 429 and think your limit is too low, follow the help center guide: check your tier and limits on the Rate limits screen in the Console, and you can request an increase after using more than half of your current limit.

Steps that save time when Claude is down

1. Read the screen message for "limit" and a reset time. If present, follow the help center guide to wait or turn on usage credits.
2. If not, look at your product line on status.claude.com.
3. If there is an incident bar, convert the update times to your local time and wait for the recovery notice.
4. If there is no bar and the message says "capacity constraints", resend after a few minutes.
5. For a login error, check your VPN, extensions, and cache in that order.

If urgent work depends on Claude alone, you can also put the same question into another chatbot as a backup. This is the author's suggestion, not official guidance. The official guidance is to watch the status page updates and wait during an outage.

Frequently asked questions

Does the Claude mobile app have the same outage?

Yes. An update on the status page for the November 13, 2026 incident stated that the mobile app is affected in the same way as claude.ai. When the web is down, the app is often down too.

Where can I see past outage records?

You can browse them by date on the History screen of the status page. The same place offers Atom and RSS feeds (status.claude.com/history.atom).

Do I have to pay extra if I hit a limit?

Not necessarily. You can simply wait until the reset time. If your plan supports it, you can turn on usage credits on the usage screen in settings and keep working past the limit, paying for what you use.