On 13 November 2026, many people couldn’t sign in to Claude. Later that day a second incident caused errors across claude.ai, the API and Claude Code. Both are resolved now, but outages happen to every online service. This guide shows you how to tell in under a minute whether Claude is really down or something is wrong on your end, how to get alerts, and how to keep working while it’s fixed.
Incident details come from Anthropic’s status page as of 14 November 2026. Fixes come from Anthropic’s help centre and the Claude Code docs.
What happened on 13 November
Anthropic’s status page lists two separate incidents that day (times in UTC):
| Incident | Affected | Timeline |
|---|---|---|
| Elevated errors on claude.ai, Claude Code, and platform.claude.com | claude.ai, Claude Console, Claude Code, Claude Cowork | Investigating 09:58. A fix was deployed at 11:05 and the incident was marked resolved at 11:58. |
| Elevated errors on claude.ai, the Claude API, and Claude Code | claude.ai (including the mobile apps), Claude API, Claude Code, Claude Cowork | Investigating 13:09, recovery seen by 14:36, resolved 15:38. Adding new connectors was briefly unavailable. |
The first incident was about sign-in. Anthropic’s first update said it was “investigating an issue with users authenticating to Claude.ai and Claude Code.” Anthropic hasn’t published a root cause for either incident. A separate problem with the Microsoft 365 connector on 12–13 November was put down to an incident on Microsoft’s side.
What you need
- A browser, to open the status page.
- For the command-line checks: a terminal with
curl(built into macOS, Linux and Windows 10/11). - For the Claude Code tips: Claude Code installed and signed in. If you’re new to it, our guide to agent skills in Claude Code covers the basics.
Step 1: Check the official status page
Go to status.claude.com. The old address, status.anthropic.com, now redirects there. The page shows each part of Claude separately:
- claude.ai: the web app and the mobile apps.
- Claude Console (platform.claude.com): the developer dashboard.
- Claude API (api.anthropic.com): what apps and scripts call.
- Claude Code and Claude Cowork.
- Claude on Cloud Platforms: Claude on AWS, Microsoft Foundry and Google Cloud.
This split matters. The Claude API wasn’t on the affected list for the first 13 November incident, which was about signing in to claude.ai and Claude Code. If only one component shows a problem, you know exactly which parts to avoid.
Anthropic’s help centre is clear on one point: “Capacity constraints are not outages.” A busy-period message doesn’t mean the service is down.
Step 2: Check it from the terminal
The status page runs on Atlassian Statuspage, which has a public JSON API. This is handy in scripts or when the page loads slowly:
curl -s https://status.claude.com/api/v2/status.jsonYou’ll get something like this:
{"page":{"id":"...","name":"Claude","url":"https://status.claude.com","time_zone":"Etc/UTC",...},
"status":{"indicator":"minor","description":"Partially Degraded Service"}}The indicator is none when everything is fine, then minor, major or critical. For a per-component view, use the summary endpoint. If you have jq installed, this prints one line per component:
curl -s https://status.claude.com/api/v2/summary.json | jq -r '.components[] | "\(.name): \(.status)"'Each line reads operational, degraded_performance, partial_outage or major_outage. To list recent incidents:
curl -s https://status.claude.com/api/v2/incidents.json | jq -r '.incidents[:5][] | "\(.created_at) \(.status) \(.name)"'Step 3: Get alerted instead of refreshing
The status page has a Subscribe to updates button. It offers email, SMS, Slack, Microsoft Teams, webhook and RSS. If you’d rather use a feed reader, add one of these:
https://status.claude.com/history.rss
https://status.claude.com/history.atomOn an incident’s own page you can also subscribe to updates for just that incident. Teams that rely on Claude in production should send the webhook or Slack alert to a shared channel, so nobody has to guess.
Step 4: If the status page is green, fix your side
If the page shows everything as operational but Claude still won’t load or sign you in, the problem is probably local. Anthropic’s help centre suggests these steps:
- Turn off your VPN and reload.
- Disable browser extensions.
- Clear your browser’s cache and cookies.
If that doesn’t help, trying another browser or the mobile app is a quick way to narrow down where the problem is.
If the status page does show a system-wide incident, Anthropic’s advice is simply to wait: “it’s best to wait for service restoration before attempting further troubleshooting.” Signing out and in again repeatedly during an authentication incident is unlikely to help.
Step 5: Keep Claude Code working through blips
Claude Code already handles short failures. Its docs say it “retries transient failures up to 10 times with exponential backoff before showing you an error.” That covers server errors, overloaded responses and timeouts, plus temporary 429 throttles. You can tune this with environment variables:
# Change the number of retries (default 10)
export CLAUDE_CODE_MAX_RETRIES=15
# For unattended sessions: keep retrying 429 and 529 capacity errors indefinitely
export CLAUDE_CODE_RETRY_WATCHDOG=1If you see a 529 overloaded error, run /model in Claude Code and pick a different model. The docs explain that “capacity is tracked per model”, so another model may have room. A 529 also “doesn’t count against your quota”.
If your company already has Claude through a cloud provider, Claude Code can use it instead of Anthropic’s API. Anthropic doesn’t publish this as an outage workaround, but it’s a supported setup and it can help when only Anthropic’s own endpoints are affected. Check the “Claude on Cloud Platforms” rows on the status page first. You’ll need your own cloud account, model access and credentials set up in advance:
# Amazon Bedrock
export CLAUDE_CODE_USE_BEDROCK=1
# Google Cloud's Agent Platform (formerly Vertex AI)
export CLAUDE_CODE_USE_VERTEX=1Each one needs extra settings, such as region and project, so follow the Bedrock or Google Cloud setup page.
Step 6: Make your own API code outage-friendly
If you call the Claude API from your own app, handle these status codes on purpose:
| Code | Type | What it means and what to do |
|---|---|---|
| 429 | rate_limit_error | You hit a rate limit or a spend cap. If there’s a retry-after header, wait that long. A spend-cap 429 has no such header and keeps failing until access resumes. |
| 500 | api_error | An internal error at Anthropic. Retry with exponential backoff, and contact support with the request ID if it persists. |
| 529 | overloaded_error | The API is temporarily overloaded. Back off and retry, or fall back to another model. |
Add a fallback in your app, such as a cached answer, a queue or a friendly “try again soon” message, so users don’t see a raw error.
Common errors and fixes
- “Waiting for authentication…” never finishes in Claude Code. Check the status page for an authentication incident first. If it’s green, see Anthropic’s Claude Code installation and authentication troubleshooting article.
- “API Error: 500 Internal server error.” Claude Code has already retried. Wait a moment, then check status.claude.com, as the error message itself suggests.
- 529 overloaded in the middle of work. Switch models with
/modeland continue. It isn’t your usage limit. - A connector (Microsoft 365, Google Calendar) fails but chat works. Connector incidents are listed separately and are sometimes caused by the other company’s service. Check the status page before you reconnect anything.
- Only the iOS or Android app fails. The mobile apps share claude.ai’s status. Try the web version, then update the app.
Wrapping up
When Claude stops responding, check status.claude.com first. It tells you within seconds whether the problem is on Anthropic’s side, and which part is affected. Subscribe once so alerts come to you, keep the local fixes in mind for when the page is green, and set up Claude Code’s retries and model switching before you need them. To get more out of Claude when it’s running, see our explainer on Claude Fable 5.2 and our guide to Claude Whiteboards.
Sources
- Claude status page and its incident history for 13 November 2026
- Anthropic Help Center: Troubleshoot Claude error messages
- Anthropic Help Center: Log in to your Claude account
- Claude Code docs: Errors and Environment variables
- Claude API docs: Errors
Write a response