Major outage·OpenAI
GPT-5 and o3 unavailable for 3+ hours
May 11, 14:23 UTC
3h 25m
All frontier reasoning endpoints returned 503s. Caused by a misconfigured deployment in the inference layer.
Live tracker · Updated 2026-05-15
Is ChatGPT, Claude, or Gemini down?
Live status across every major AI provider, plus recent incident history with severity and duration. One place to check when your API call just timed out.
Provider health
Last 30 days
OpenAI
frontier
1 incident3h 25m
View status page →
Anthropic
frontier
1 incident1h 27m
View status page →
Google AI / Gemini
frontier
1 incident5h 30m
View status page →
xAI / Grok
frontier
No incidents
View status page →
Mistral
frontier
No incidents
View status page →
DeepSeek
frontier
No incidents
View status page →
Perplexity
specialised
No incidents
View status page →
Cohere
specialised
No incidents
View status page →
Groq (inference)
specialised
No incidents
View status page →
Together AI
specialised
No incidents
View status page →
GitHub Copilot
specialised
No incidents
View status page →
Incidents in last 30 days
3
Across tracked providers
Total downtime
10h 22m
Aggregate, all providers
Most affected
OpenAI
By incident count
Recent incidents
Partial outage·Anthropic
Elevated latency on Claude API
Apr 28, 09:15 UTC
1h 27m
P99 latency on claude-4-7-opus exceeded 30s. Background indexing job created head-of-line blocking.
Major outage·Google AI / Gemini
Vertex AI Gemini endpoints down in us-central1
Apr 22, 20:00 UTC
5h 30m
Region-wide outage affecting Gemini 3 Pro requests routed through us-central1.
Incident data is curated from each provider's official status page. Click any provider tile above to view their live status feed. Severity, duration, and affected services are pulled from the original post-mortems where available.
Frequently asked questions
Is ChatGPT down right now?
Check the OpenAI tile in the provider grid above. Green dot means no incidents currently logged on their status page; red dot means an active incident is in progress. Click the tile to view OpenAI's official status feed.
Why does Claude / GPT / Gemini keep failing?
Frontier AI APIs run on massive GPU fleets that are saturated near capacity most of the time. When a single component fails (a routing layer, a model deployment, a region) the impact tends to be visible immediately. Major providers average 1 to 3 user-visible incidents per month, with most resolving in under two hours.
Where do you get the incident data?
We pull from each provider's official status page, which they publish primarily for transparency and SLA accountability. Each incident links back to the source. We don't include incidents tagged as 'maintenance' unless they materially affected users.
How can I be alerted to AI outages?
Subscribe to our weekly newsletter (link in the footer). For real-time alerts, follow each provider's status page directly — most support RSS, email, and webhook subscriptions. Some teams pipe these into Slack or PagerDuty.
What should I do during an AI outage?
If your workload is critical, run a multi-provider fallback: try Claude if GPT is down, try Gemini if Claude is down, etc. Most modern AI SDKs (LangChain, LlamaIndex, OpenRouter) support automatic provider fallback. For chat use, ChatGPT, Claude, Gemini, and Grok rarely all go down at once.
Related on Meridian48
The 48° Brief — Weekly
The week in AI, Pakistan tech, and global business.
One email, every Friday morning. Curated and written by Faizan Khan. No filler. No tracking pixels. Unsubscribe in one click.
Join the first readers of Meridian48.