Is Claude down? Claude status

No. Claude is up and running normally.

Operational Last checked (Jul 28, 2026 17:23 UTC) 29 ms Open Claude ↗

This verdict covers the Claude.ai consumer web app (claude.ai); Claude Code, Claude Cowork, the API platform and Claude for Government are tracked as separate groups below.

  • Official signal Operational
  • Independent web check (North America: Virginia, US) Operational
  • User reports 0 in the last hour
  • North America: Virginia, US Operational 29 ms · checked Jul 28, 2026 17:23 UTC

Independent reachability is currently checked from Virginia, US.

Independent monitor. Not affiliated with or endorsed by Claude.

6 / 6 Components up
0 Open incidents
29 ms Web endpoint latency
0 Reports (1h)
98.35% Uptime (30d)

We check Claude continuously using a dual-source method. Read more about our methodology.

System components

6 of 6 operational
  • Claude.ai Operational
  • Claude Code Operational
  • Claude Cowork Operational
  • Claude Console (platform.claude.com) Operational
  • Claude API (api.anthropic.com) Operational
  • Claude for Government Operational

Component status comes from Claude’s official status page. Official source: status.claude.com. Last synced .

Outage reports

0 in the last hour

If you landed here, Claude probably froze mid-response, refused to load, or threw an error just when you needed it. Take a breath: the vast majority of these problems are quick to identify and quick to fix. Sometimes the trouble is on Claude’s side, and sometimes it is something small and local that you can sort out in a minute or two. Below is a calm, practical walkthrough to help you figure out what is happening and get back to work.

Illustration representing Claude status

Common reasons Claude stops working

Anthropic runs Claude as several separate surfaces, each with its own status: the Claude.ai app, Claude Code, Claude Cowork, the Console and the API. A problem on one does not mean the others are down.

  • Claude.ai versus Code, Cowork and the API. The chat app at claude.ai is distinct from Claude Code (the terminal and IDE tool), Claude Cowork, the Console at platform.claude.com and the API at api.anthropic.com. The component list on this page shows each one, so an API or Claude Code incident is easy to separate from a claude.ai problem.
  • Capacity and “overloaded” errors. When demand is high Claude can return an overloaded or capacity error. It is temporary and usually clears if you retry a moment later, and it can appear on one surface while the others are fine.
  • Two different sign-ins. Logging in to claude.ai is separate from the API keys used by the Console and by developer tools. An API-key or Console problem can leave the claude.ai app working normally.
  • Your side of the connection. A dropped network, a stale session or an extension interfering with the page can imitate an outage when Claude is actually up.

How to fix Claude when it is not working

Work through these steps from the top. Most people are back up and running before they reach the end.

  1. Hard refresh the page. Reload while bypassing the cache (Ctrl+Shift+R on Windows or Linux, Cmd+Shift+R on a Mac). This clears a surprising number of loading and display glitches.
  2. Open a private or incognito window. A fresh window with no extensions and no stored data quickly tells you whether the problem is your normal browser setup.
  3. Sign out and back in. This refreshes your session and clears up many login loops and permission errors.
  4. Clear your cache and cookies. If a private window worked, clearing stored site data in your everyday browser usually fixes it for good.
  5. Switch networks. Move from Wi-Fi to mobile data, or the other way around. If it works on one network but not the other, the issue is local to your connection.
  6. Disable extensions and turn off your VPN. Switch off ad blockers, privacy add-ons, and any VPN or proxy, then reload. Re-enable them one at a time to find the culprit.
  7. Update or restart the mobile app. Close it fully, reopen it, and check your app store for an update if you are on a phone or tablet.
  8. Wait out rate limits. If you are told you have reached a limit, pause for a while. These caps reset over time, and waiting is the only real fix.

Is Claude down, or is it just you?

The fastest way to tell the difference is to separate a wide outage from a local problem. Start with this page: it shows a live verdict alongside uptime history and recent user reports, so you can see at a glance whether others are affected right now. If the signals here look healthy but it is still broken for you, the trouble is most likely on your end.

To confirm, try two quick tests. First, open Claude in a private window to rule out your cache and extensions. Second, switch to a different network, such as your phone’s mobile data. If it works in either of those, the service itself is fine and you have narrowed the cause to your setup. For a step by step approach that applies to any AI tool, see our guide on what to do when an AI tool is down.

How we check Claude

We determine Claude’s status from two independent signals: Claude’s official status page, which reports Claude.ai, Claude Code, Claude Cowork, the API platform and Claude for Government as separate components, and our own reachability check that connects to the service directly. Only the Claude.ai consumer component drives the main verdict; an API, Code or Console incident is shown in its own group without marking the consumer app down. Using both means we can tell the difference between an officially acknowledged incident and a quieter problem that has not been posted yet, which gives you a more reliable picture than either source alone. You can read more about how this works on our methodology page.

Alternatives to Claude during an outage

If Claude is genuinely down and you have something that cannot wait, another assistant can usually cover the same ground while you wait for service to return. A few solid options:

  • ChatGPT for general writing, coding help, and everyday questions.
  • Gemini for research, summaries, and tasks tied into the wider Google ecosystem.
  • DeepSeek as a capable, lower cost alternative for reasoning and code.

It is worth keeping one of these handy as a backup so a single outage never brings your day to a stop. When Claude is back, you can pick up right where you left off. Most outages are short, and most local issues are even quicker to fix, so it usually will not be long before everything is working again.

What Breaks When Claude Has Trouble

Claude is available as a web app and mobile app from Anthropic, with a separate API and Console for developers. The signal most people notice during peak hours is an “Overloaded” or capacity error, which means Anthropic’s systems are temporarily saturated rather than offline. Other friction points include hitting the message and usage limits on the Free and Pro plans, attachments or long documents that fail to process, and the occasional sign-in loop where your session does not persist.

Because Claude handles long context and large file attachments, some failures cluster around big inputs: a very long conversation or a heavy upload may error while short prompts still go through. Projects, which group chats and reference files together, can also load slowly when the service is degraded.

  • “Overloaded” or capacity messages during high-traffic periods
  • Reaching the per-window message limit on Free or Pro
  • File attachments or long-context requests timing out
  • Sign-in not sticking or repeated re-authentication

Uptime, last 30 days

98.35%
98.92% 7-day
98.35% 30-day
Building 90-day

Web endpoint latency

Recent incidents

Embed this status badge

Show Claude's live status on your own site, docs or README. It updates automatically and links back here.

Claude status, via is-down.ai
HTML
Markdown

On WordPress? Our free AI Status Widget plugin shows Claude's live status with cards, badges and full uptime panels.

Frequently asked questions

Is Claude down right now or am I just rate limited?

Start with the live verdict at the top of this page, which pairs Anthropic's status with our own reachability test. A true outage affects many users and usually shows up in our reports feed, while a rate limit is specific to your account and clears after a wait. If Claude loads but refuses new messages, you are more likely capped than down.

What does the "Overloaded" error in Claude mean?

It means Anthropic's capacity is temporarily maxed out, typically during busy periods, so your request was turned away rather than the whole service being offline. Retrying after a short pause often succeeds once load eases. If you see it constantly across many attempts, check the verdict above to confirm whether it is a wider incident.

Why did Claude stop letting me send messages today?

Free and Pro plans include usage limits that reset on a rolling window, and long conversations consume more of that allowance. When you reach the cap, Claude pauses new messages until the window resets rather than showing a server error. Starting a fresh, shorter chat can also reduce how quickly you hit the limit.

Is the Claude app outage the same as the Anthropic API being down?

Not necessarily. The Claude web and mobile apps and the Anthropic API and Console are distinct surfaces, so the consumer app can work while the API is degraded or vice versa. Developers should consult the API and Console status for their integrations. This page tracks the Claude app experience for end users.

Why does Claude fail on long documents or large attachments?

Large files and very long contexts are heavier to process, so they can time out or error even when short prompts work normally. If an upload fails, try a smaller file, split the content, or trim the conversation length. Persistent failures across small inputs too would point to a broader service problem you can confirm above.

Why can't I sign in to Claude or stay logged in?

Sign-in problems often come from an expired or blocked session rather than a full outage, especially with strict cookie or privacy settings. Try a private window, clear cookies for the Anthropic domain, and request a fresh login link if you use email sign-in. If sign-in fails for many people at once, it usually reflects an authentication incident shown in the verdict above.