The Short Answer
Go to the vendor's own status page first. Claude is at status.claude.com, ChatGPT at status.openai.com, Gemini on the Google Workspace Status Dashboard, and Copilot at status.cloud.microsoft. If the page shows nothing and you still cannot work, the steps below help you tell an unannounced outage from a problem on your side.
Step 1: Open the Official Status Page
| Assistant | Status page | Alerts you can subscribe to |
|---|---|---|
| Claude | status.claude.com | Email, SMS, Slack, Microsoft Teams, webhook, RSS |
| ChatGPT and the OpenAI API | status.openai.com | Email, SMS, Slack, RSS |
| Gemini app and Gemini in Workspace | google.com/appsstatus/dashboard | Atom feed. No email option on the page |
| Gemini through Google Cloud | status.cloud.google.com | Atom feed |
| Microsoft Copilot, consumer | status.cloud.microsoft | None found |
| Microsoft 365 Copilot, work | Microsoft 365 admin centre, under Health, then Service health | Email, up to two addresses per admin |
| Perplexity | status.perplexity.com | Email, SMS, Slack, RSS |
| Grok | status.x.ai | RSS feed |
| DeepSeek | status.deepseek.com | Email, RSS |
Mistral also has a status page at status.mistral.ai, which we could not read automatically.
Step 2: Look at the Right Component
Status pages list products separately, and they fail separately. On March 11, 2026 the Claude app was down while the Claude API was unaffected. Check the line for what you actually use.
- Claude: claude.ai, the API, Claude Code, the console and Cowork are separate lines.
- OpenAI: ChatGPT, the APIs and Codex are separate groups, and inside ChatGPT, login and conversations are separate components.
- Microsoft: in the admin centre, Copilot issues carry the prefix CP or CW. See Microsoft 365 incident ID.
Step 3: Check Whether Login Is the Problem
Many 2026 outages were authentication failures with the model itself healthy. Claude's first major incident on November 13, 2026 was a login failure, and three of ChatGPT's larger outages this year traced to authentication systems. If you are already signed in on another device or through the API, try that before assuming everything is down.
Step 4: If the Status Page Is Green
Status pages lag. Anthropic's first post about its Microsoft 365 connector on November 12 came at 17:52 UTC, and Microsoft's own suite-wide advisory opened at 18:26 UTC. Before the vendor confirms:
- Check a crowd source: a sharp rise on Downdetector suggests it is not only you. Read it as a signal, not a measure. See what a report count means.
- Rule out your side: try another network, a private browser window, and another device.
- Check dependencies: if only one feature fails, such as a connector to another product, the fault may be in that other product.
- Wait ten minutes and reload the status page before changing anything in your own setup.
Step 5: Set Up Alerts Once
Subscribe to the status pages of the assistants your team depends on, and send the alerts to a shared channel instead of one person's inbox. Claude, OpenAI and Perplexity all offer Slack delivery directly. For Google, add the Atom feed to a feed reader or a channel integration. If you rely on dozens of vendors, an aggregator can do this in one place. See Downdetector vs StatusGator vs IsDown.
How Often Does This Happen?
Often enough to plan for. Anthropic's status page showed 99.54% uptime for claude.ai over the 90 days to November 13, 2026, and OpenAI's showed 99.95% for ChatGPT conversations. Google and Microsoft publish no uptime figure for Gemini or Copilot. Full history is in our AI assistant outage tracker, and what to do next is in how to keep a team working when an AI assistant is down.
Sources
Each status page was opened on November 13, 2026: Claude, OpenAI, Google Workspace, Microsoft, Perplexity and DeepSeek. Subscription options can change.