Is Claude Down? Claude AI Outage Status and Fixes

Is Claude down right now?

No, Claude is not fully down as of 20:43 UTC (23:43 Istanbul) on 13 November 2026, but Anthropic's status page shows "Partially Degraded Service" ^[1]. At that check, claude.ai and Claude Cowork were listed as degraded, while the Claude API, Claude Code, the Claude Console and the cloud-platform versions were listed as operational ^[1].

The two big incidents of the day are both marked resolved ^[2][3]. The one item still open is a problem with the Microsoft 365 connector, which Anthropic says was caused by a Microsoft 365 service incident and which was in the monitoring stage at its last update, at 15:40 UTC on 13 November ^[4].

Status pages change by the minute during an incident. The fastest way to get the current answer is to open Anthropic's official status page, now at status.claude.com (the old status.anthropic.com address redirects there) ^[1]. If the page shows all green and Claude still fails for you, the problem is more likely on your side, and the steps further down can help.

What happened in the Claude outage on 13 November 2026?

Claude had two separate major incidents on 13 November 2026: a sign-in failure in the morning (UTC) and a wider error spike in the afternoon that also reached the API ^[2][3]. Anthropic's status page describes them as elevated errors; it does not call either a global outage ^[2][3].

Incident 1. At 09:58 UTC Anthropic began investigating an issue with users authenticating to claude.ai and Claude Code; the incident title also names platform.claude.com ^[2]. A fix was deployed at 11:05 UTC and the incident was marked resolved at 11:58 UTC ^[2]. The Economic Times reported that Anthropic's first post came at 4:58 a.m. Eastern, which is the same moment ^[6].

Incident 2. At 13:09 UTC a new incident opened for elevated errors on claude.ai, the Claude API and Claude Code ^[3]. Anthropic added connectors such as Docs and Sheets to the list at 13:52, then Claude Security and Managed Agents at 14:03 ^[3]. Recovery on claude.ai and the API was reported at 14:36 (with a warning that errors could return), monitoring began at 15:33 and the incident was resolved at 15:38 UTC, with its impact rated major ^[3].

User reports rose sharply during the day. The Economic Times reported that Downdetector recorded 2,895 reports around 3:45 PM, against a baseline of two; the report does not state the time zone ^[6]. Android Authority's live article was updated at 1:14 PM ET to say the status site showed the outage resolved for claude.ai, the API, Claude Code and connectors, while noting the Microsoft 365 connector was still having problems ^[7].

Which Claude services were affected?

The status page lists claude.ai, Claude Code and Claude Cowork as affected in both incidents, the Console in the first, and the Claude API in the second ^[2][3]. In practice, that covered the web and app chat, the coding tool in the terminal and editor, the agent workspace, and developer apps calling the API during the afternoon incident.

Some features had knock-on effects after recovery. In its 15:33 UTC update, Anthropic said adding a new connector was temporarily unavailable, scheduled tasks and routines could run more than two hours late with some runs skipped, and Claude Security scans were delayed ^[3].

The Microsoft 365 connector problem is separate. It opened on 12 November at 17:52 UTC and Anthropic attributes it to a Microsoft 365 service incident, so requests for mail, calendar, SharePoint and Teams chat through the connector, and Claude for Outlook, were affected even when the rest of Claude worked ^[4].

How do I check if Claude is down for everyone?

Check Anthropic's official status page first, then compare with your own symptoms ^[1].

1. Open status.claude.com. Each component, such as claude.ai, the API or Claude Code, has its own colour; an open incident appears at the top with time-stamped updates ^[1].
2. Note the time of the latest update. Anthropic posts times in UTC on incident pages; add three hours for Istanbul ^[2][3].
3. Try a second surface. If the website fails but the API or Claude Code works, the problem is limited to one service; the status snapshot at 20:43 UTC on 13 November showed exactly that, with claude.ai degraded and the API and Claude Code operational ^[1].
4. Use outage trackers such as Downdetector only as a rough signal: they count user complaints, not confirmed failures ^[6].
5. If you depend on Claude for work, look for a subscribe option on the status page so you hear about the next incident without checking.

What can I do while Claude is down?

Wait for the official fix, but try a few quick checks first, since some problems are local ^[1].

• Login loops: during the morning incident the failure was in authentication ^[2]. Repeatedly signing in rarely helps then; wait for the status page to show recovery, then sign out and in once.
• Refresh and clear: reload the page, try a private browser window or the mobile app, and turn off extensions that alter pages.
• Save your work: copy long prompts to a text file before sending, so a failed request does not lose them.
• Check scheduled jobs: after the 13 November incident some scheduled tasks and routines ran late or were skipped ^[3]. Re-run anything important by hand.
• Use another tool for urgent work: the Pouyam guide ChatGPT vs Gemini vs Claude in 2026: Models, Prices and Key Differences compares alternatives.

Do not download "Claude fix" apps or browser extensions advertised during an outage, and do not enter your password on a page that is not claude.ai. Fake support pages can appear whenever a well-known service is in the news.

What do Claude API errors 529, 500 and 429 mean?

Anthropic's API documentation says 529 means the API is temporarily overloaded, 500 is an unexpected internal error, 429 means you hit a rate limit and 504 is a timeout ^[5].

The documentation advises retrying 500 errors with exponential backoff, which means waiting a little longer after each failed attempt ^[5]. Anthropic's official SDKs retry some errors automatically, twice by default ^[5]. The documentation describes a 429 as your organization hitting a rate limit or spend limit ^[5], which points to your own usage, not an outage; slow down or check your limits in the Console.

Every API response carries a request ID, and the documentation asks you to include it when contacting support, so log it ^[5]. Other good habits for apps built on the API: set a timeout, show users a clear message instead of a blank screen, and keep a fallback path for critical features.

Why does Claude go down, and will it happen again?

Anthropic has not published a cause for the two 13 November incidents as of 20:43 UTC; the status pages describe symptoms and fixes, not root causes ^[2][3]. Large AI services depend on sign-in systems, data centres, model servers and outside services; a fault in any one can show up as elevated errors.

The Microsoft 365 connector problem shows how outside providers matter: Anthropic attributes that incident to a Microsoft 365 service incident ^[4]. Outages happen at every large online service. The useful response is to follow the official status page, keep local copies of important work and avoid single points of failure in anything you build.

Is Claude down only in my country?

Anthropic's status page does not give a country breakdown, so it cannot tell you which countries were affected by the 13 November incidents ^[2][3]. Press coverage on the same day, such as Android Authority's, described the outage as widespread ^[7].

If the status page is green and Claude still fails only for you, look closer to home: your internet provider, a company firewall, a VPN that routes you through a blocked location, or a browser extension. Try another network, such as mobile data, before assuming an outage.