Is ChatGPT, Claude, Gemini or DeepSeek down right now?
Live status of 38 AI APIs and apps. We combine each vendor’s official status page with our own probes from monitoring nodes around the world every 2 minutes — so you see problems even when the official page still says “operational”. Last check: 15s ago.
Recent incidents
Unable to start new Workspace Agent runs
Service Degradation for Kimi K3.1 Pro
Service Degradation for Kimi K3.1 Pro
Elevated errors affecting logins, sign-ups, and ChatGPT features
Database creation failures in APAC
Service Degradation for GLM 5.3
FAQ
How does ModelBank detect outages?
Every 2 minutes our monitoring network (1) reads each vendor’s official status page and (2) sends an unauthenticated request to the vendor’s public API endpoint. If the API responds at all (even with 401 “missing key”), the front door is up. Timeouts, connection errors and 5xx responses are counted as failures.
Why does ModelBank say “down” while the official page says “operational”?
Official status pages are updated by humans and often lag 10–30 minutes behind real problems. Our probe is automated. Conversely, a model-specific issue may appear on the official page before it affects the API front door.
Can I get this data programmatically?
Yes — use the free JSON API and embeddable status badges (see the Developers section in the footer).
Get outage alerts
An instant alert when a major AI API goes down — and another when it recovers. No spam.
Email · Browser push · Slack / Discord · Feishu / DingTalk / WeCom · RSS — free, no sign-up