Is Claude AI down?
Check Anthropic’s official status page at status.claude.com and read the timestamp, component and incident details. Claude is not one endpoint: claude.ai, the Console, API, Claude Code, Cowork and Claude for Government can have different states. If the relevant component is operational, test your account, browser, network or API request before assuming a global outage.
Why — the first-principles explanation
“Is Claude down?” compresses several different failures into one question. A provider incident can affect a single product surface, a feature, or particular model requests; an account, browser, VPN, firewall or quota problem can affect only you. Anthropic’s status page is the first-party record for service components and incident updates, not a guarantee that every individual request will succeed. Start by naming the surface: claude.ai chat, Claude Console, the API, Claude Code, Cowork or Claude for Government. Then read the incident’s impact and affected components instead of relying on a green badge or a social-media post. A status page can lag a newly emerging issue, and an operational provider can coexist with a local login or network failure. For an API error, separate transport and firewall problems from an HTTP error returned by Anthropic; Anthropic’s troubleshooting guidance specifically calls out firewall, network and VPN checks. For the web app, reproduce with a harmless minimal prompt, a private window or another network while keeping the same account and note the exact time and error. Do not paste private prompts or keys into an outage checker. If the service remains unavailable, follow the incident updates and use Anthropic’s Help Center support messenger; support options depend on account type and plan. Treat uptime percentages, third-party down-detector votes and a single successful test as context, not proof that the service is or is not down right now.
An example that makes it click
Suppose claude.ai shows a timeout but the API returns normal responses. That is not evidence that “Claude is down”; it points to a surface-specific issue, account session or browser path. Conversely, if the status page names claude.ai and the API as degraded, switching browsers will not repair the provider incident. Record the component, error, timestamp and whether a second network reproduces it, then choose either the incident updates or local troubleshooting path.
How to do it
- Open Anthropic’s official status page at https://status.claude.com/ and read its current timestamp, overall indicator and active incident banner.
- Match your problem to the component: claude.ai, Claude Console (platform.claude.com), Claude API (api.anthropic.com), Claude Code, Claude Cowork or Claude for Government.
- Open the incident details and check the affected components, impact wording and latest update. “Degraded performance” is not the same as a total outage, and a resolved incident is historical evidence rather than a current diagnosis.
- If a model or feature is named, retry a harmless request with an available alternative only when your account already offers one. Do not infer that a model switch fixes a provider-wide outage.
- If the relevant component is operational, reproduce once with a new conversation or minimal prompt. For the web app, try a hard refresh, private window, another supported browser and another network; disable extensions or VPNs only as a controlled test.
- For the API, save the HTTP status, response body, request ID and UTC time without exposing your API key. Check firewall, proxy and VPN rules; Anthropic documents these as possible causes of connection errors.
- Check whether the failure is limited to one account, organization, plan or quota. A green public status page cannot confirm private billing, authentication, permission or rate-limit state.
- Subscribe to status updates or wait for the next incident update instead of repeatedly refreshing. If the page stays green and the issue persists, use the Claude Help Center support messenger and include the component, timestamp and safe reproduction details.
- Recheck after the incident is marked resolved. Keep the status page, support article and troubleshooting steps as the source of truth; do not treat a third-party outage score, browser extension or cached screenshot as a permanent diagnosis.
Key facts
- Anthropic’s official status page is status.claude.com; its current indicator and incident history are time-sensitive.
- The status page separates claude.ai, Claude Console, Claude API, Claude Code, Claude Cowork and Claude for Government.
- A component can be degraded while another Claude surface remains operational, so identify the product before troubleshooting.
- Incident updates can describe a feature or model-specific problem rather than a total Claude outage.
- “Operational” means no provider incident is currently reported; it does not rule out login, account, browser, network, VPN, firewall, quota or permission problems.
- Anthropic’s API troubleshooting guidance calls out firewall, network and VPN configuration as possible causes of connection errors.
- A status page can be briefly behind a newly emerging incident, so record the time and exact error when reproducing.
- Third-party down detectors and social posts can help discover a pattern but are not a substitute for Anthropic’s component-level status and incident updates.
- Support access differs by Free, Pro, Max, Team, Enterprise and Console account roles; use the Help Center’s documented support path.
- Never share API keys, private prompts, customer data or authentication cookies while asking whether Claude is down.
Try Claude's $0 Free plan first, then compare current usage limits and billing before upgrading.
Diagnose Claude by surface
Check provider status first, then route to the right Claude product, plan, support or alternative-tool comparison without claiming a global outage from one failed request.
▶ The 60-second explainer (script)
Is Claude AI down right now? Start with Anthropic’s official status page, status dot claude dot com. Read the timestamp and match the problem to the right surface: claude dot ai, the Console, the API, Claude Code, Cowork, or Claude for Government. Claude is not one switch, so a partial incident can affect one surface while another works. If the page reports an incident, follow its updates and affected-component list. If your component is operational, test a harmless minimal request, then check your account, browser, extensions, VPN, network, firewall, quota and permissions. For API errors, save the HTTP status and request ID, never the API key. A green badge does not rule out a local problem, and a third-party down detector is not an official diagnosis. Still stuck? Use the Claude Help Center support messenger with the component, UTC time and safe reproduction details.
What authoritative sources say
People also ask
Is Claude AI down right now?
Check status.claude.com at the time of the failure. Read the current timestamp, overall indicator and affected component; this page changes as Anthropic investigates and resolves incidents.
What is the official Claude status page?
Anthropic’s official status page is https://status.claude.com/. It separates claude.ai, the Console, API, Claude Code, Cowork and Claude for Government and publishes incident updates.
Claude is slow but not completely down. Is that an outage?
It may be a degraded-performance incident, a model or feature issue, or a local network problem. Check the incident impact wording and whether the affected component matches your surface.
Why does claude.ai work while the Claude API fails?
They are separate service surfaces with different authentication and network paths. Check the API component, HTTP response and request ID, then review firewall, proxy and VPN rules.
The status page is green but Claude will not load. What should I do?
Reproduce with a harmless minimal prompt, then try a hard refresh, private window, another browser or another network. Check account, permissions, quota, extensions and VPN before contacting support.
Should I switch Claude models during an outage?
Only if an official incident names a model or feature and your account already offers another model. A model switch is not a guaranteed workaround for a provider-wide or account problem.
How do I contact Claude support if the site is down?
Use the support messenger in Anthropic’s Help Center when you can sign in. Support routes depend on plan and account role; include the component, UTC time, safe error text and reproduction steps, never an API key.
Can a third-party down detector prove Claude is down?
No. It can suggest a pattern, but confirm with Anthropic’s component-level status and incident updates. A local outage, account issue or cached test can produce a different result.
The same question, asked other ways
- Is Claude AI down right now?