Is Claude Down Again? Claude Status Today and Recent Outages

Is Claude down again right now?

Claude is up but partly degraded as of 03:40 UTC on 15 November 2026: the only open incident is a Sign in with Apple problem on the Claude iOS app, which mainly affects new accounts ^[1][6]. The Claude API, Claude Code, Claude Cowork and the Console (platform.claude.com) all show as operational ^[1].

Anthropic opened the incident at 07:11 UTC on 14 November and was still investigating it in its last update at 20:12 UTC that day ^[6]. Its advice: new users can sign up with Google or email instead ^[6]. Existing users who are already signed in are not described as affected ^[6].

So if Claude works for you but a friend cannot create an account on an iPhone, this is the reason. If chats fail for you on the web, in Claude Code or through the API, the cause is not this incident, and you should check the status page again, because a new incident may have opened after this check.

The Pouyam guide Is Claude Down? Claude AI Outage Status and Fixes explains the error codes 500, 529 and 429 and the basic checks. This page focuses on why Claude keeps going down and how often it happens.

Why does Claude keep going down this week?

There were five separate incidents between 12 and 14 November, each with a different cause or scope, so it looked like one long outage even though it was not ^{[2][3][4][5][6]}.

1. Microsoft 365 connector, 12 to 13 November: mail, calendar and search tools in the connector, and Claude for Outlook, returned errors from 16:03 UTC on 12 November to 16:40 UTC on 13 November ^[4]. Anthropic says the cause was a Microsoft 365 service incident ^[4].
2. Sign-in outage, 13 November, 09:58 to 11:58 UTC: users could not authenticate on claude.ai and Claude Code; the Console was also listed as affected ^[2]. A fix was deployed at 11:05 UTC ^[2].
3. Wider outage, 13 November, 13:09 to 15:38 UTC: errors on claude.ai, the API, Claude Code and connectors such as Docs and Sheets, plus Claude Managed Agents and Claude Security ^[3]. Anthropic said scheduled tasks could run more than two hours late during recovery ^[3].
4. Purchases, 14 November, 02:34 to 04:03 UTC: errors on purchases, plan upgrades and cancellations; chat itself was not affected ^[5].
5. Sign in with Apple on iOS, from 14 November: still open at the time of writing ^[6].

Some headlines linked the 13 November sign-in outage to Microsoft ^[10]. Anthropic's status page attributes only the Microsoft 365 connector incident to Microsoft; it has not published a cause for the sign-in outage or for the wider outage that afternoon ^[2][3][4].

How often does Claude go down?

Anthropic's status page lists 50 incidents between 14 August and 15 November 2026, or roughly one every two days ^[8]. Most were short and limited to particular models, connectors or products such as the developer Console.

By month, there were 13 incidents from 14 August, 13 in September, 17 in October and 7 in the first half of November ^[8]. Anthropic rated two of them as critical: a service disruption on 16 August and errors on several models on 23 October, which lasted 36 and 34 minutes ^[8].

The incidents that take down claude.ai, Claude Code and the API together are much rarer. Before 13 November, the last one was on 29 September, when errors began at 14:00 UTC and services recovered at 14:59 UTC; Anthropic said some messages sent in that hour may not have been saved ^[7].

A count of incidents is not the same as downtime. Many entries describe a few minutes of errors on a single model, such as four minutes of overload errors on Fable 5.2 and Mythos 5.2 on 10 November ^[8]. The table below lists the larger incidents since late September.

Is it Claude or just me?

Open status.claude.com and compare what fails for you with the component list: if your service shows green and only you have errors, the problem is more likely your account, network or browser ^[1].

1. Note what fails: signing in, sending a message, one model only, a connector, or payments. Each has appeared as its own incident this autumn ^[8].
2. Check the component on the status page: claude.ai, Claude Code, the Claude API, the Console and Claude Cowork are listed separately ^[1].
3. Read the wording. A message that you have reached your usage limit is a plan limit, not an outage, and it can appear when every service is green ^[9].
4. Try another surface. During the 13 November sign-in outage, the API and Claude on AWS, Google Cloud and Microsoft Foundry kept working ^[9].
5. Try another network or a private browser window to rule out a local problem.

Third-party outage trackers collect user reports, which helps show whether many people see errors, but only Anthropic's page says what the company has confirmed.

What should I do when Claude goes down again?

Stay signed in, wait for the incident update, and avoid paying twice or resetting your account while an incident is open ^[5][7][9].

• Do not sign out during sign-in problems. On 29 September Anthropic asked signed-in users not to sign out, because single sign-on and Sign in with Apple were unavailable ^[7].
• Do not pay again. After the 14 November purchase errors, Anthropic said delayed plan changes and payments would be applied and that there was no need to pay again ^[5].
• Save long prompts outside the chat. Messages sent during the 29 September outage may not have been saved ^[7].
• Use the API if you have a key. In the 13 November morning outage, the API kept working; Claude Code can run on an API key at per-token API prices ^[9].
• Expect delays after recovery. After the 13 November afternoon outage, Anthropic said scheduled tasks and routines could run more than two hours late ^[3].

For a work deadline, keep a second tool ready. The Pouyam guide ChatGPT vs Gemini vs Claude in 2026: Models, Prices and Key Differences compares the main alternatives.

How can I get alerts when Claude goes down?

Subscribe to updates on status.claude.com, which offers notifications and RSS and Atom feeds of incidents ^[1][8]. That way you see Anthropic's own wording when an incident opens, changes status and closes.

Incidents move through stages such as investigating, identified, monitoring and resolved ^[2][3][4]. "Monitoring" means a fix is in place but Anthropic is still watching; errors can return, as Anthropic warned during the 13 November afternoon outage ^[3].

The final update often gives the exact impact window in Pacific and UTC time, which is useful if you need to explain a failed job or a missed deadline ^[4][5].

Will Claude keep going down?

Anthropic's status entries for the two 13 November outages give no cause, and no post-incident report was found as of 15 November, so there is no official explanation of whether they share a cause ^[2][3]. Without that, any claim about the root cause is a guess.

The status record shows a pattern rather than a single fault: frequent short model errors, occasional connector failures (one of which Anthropic attributed to Microsoft), and rare platform-wide outages ^[4][8]. Claude Code and Cowork were also listed as affected in the 13 November sign-in incident ^[2].

For now, the practical view is simple: check the status page first, separate outages from usage limits, and keep important work saved outside the chat window. This page reflects the status page as of 03:40 UTC on 15 November 2026.