✦ AI vendor status
When an API suddenly starts failing, first check whether the vendor has a problem. This page gathers what each vendor's official status page says right now: one light per vendor, open it for components and incidents in progress. Every figure comes from the vendor; I don't measure them.
Outside China
OpenAI
Status page ↗Components
Anthropic (Claude)
Status page ↗Components
Google Gemini (Vertex AI)
Status page ↗From the Vertex Gemini API entry on Google Cloud's status page; for the Gemini API with an AI Studio key, see the next row
Components
Its status page only renders in a browser and has no public data feed; check it for the Gemini API (AI Studio keys)
Mistral AI
Status page ↗Components
Cohere
Status page ↗Components
xAI (Grok)
Status page ↗From the status page's RSS, keeping api.x.ai items only (the page also posts Cursor and Grok app incidents)
Components
Groq
Status page ↗Components
Fireworks AI
Status page ↗Components
Cerebras
Status page ↗Components
Perplexity
Status page ↗Components
Amazon Bedrock
Status page ↗From the AWS Health Dashboard Bedrock RSS for 6 regions: N. Virginia, Oregon, Frankfurt, Ireland, Tokyo, Singapore
Components
Azure OpenAI
Status page ↗From Azure's status RSS, keeping items that mention Azure OpenAI / AI Foundry; Azure only posts while something is wrong
Components
In China
DeepSeek
Status page ↗Components
Kimi (Moonshot AI)
Status page ↗Components
MiniMax
Status page ↗Components
No public status page for its models; outages are usually posted as console notices
No public status page for its models; outages are usually posted as console notices
No public status page for its models; outages are usually posted as console notices