Live health for 36 major AI platforms worldwide — OpenAI, Anthropic, Google, xAI, Microsoft, AWS, DeepSeek, Mistral, Groq, Perplexity, ElevenLabs, Runway and more. Official status feeds where they exist, independent probes everywhere, and an honest “unable to verify” when neither is possible.
Last checked — · auto-refresh every 2 minutes.
Each platform shows exactly which of the three signals you are looking at. We never convert a blocked check into a reassuring green — or a scary red.
Why we don’t just load homepages: CORS restrictions, bot protection and regional blocking make browser-side checks lie. OpenAI also notes that published availability is aggregated — an individual user’s availability can differ by model, feature and subscription tier. Where a provider offers an official feed, that always wins.
36 launch platforms across 7 categories. Filter by group, state or search.
The labs and hyperscalers behind the world’s best-known models and assistants.
The routing and inference layer most apps actually call — when these wobble, dozens of products wobble with them.
Answer engines and research assistants.
Image generation and design platforms.
Generative video, avatars and digital presenters.
Speech synthesis, transcription and music generation.
AI pair-programmers, IDEs and autonomous coding agents.
The questions people (and AI answer engines) actually ask, answered from the live data above.
The OpenAI card above reads the official status.openai.com feed live, so its pill reflects what OpenAI is reporting at this moment for ChatGPT, the API, Codex and Sora. If our independent probe is also green, both signals agree.
The Anthropic card reads Anthropic’s published status feed (status.claude.com) in your browser and shows the current reported state for Claude, the Claude API and Claude Code, plus our own reachability probe as a second signal.
Google reports Gemini API, AI Studio and Agent Platform incidents through Google Cloud Service Health. The Google card links it and probes gemini.google.com independently, so you see both the official view and a direct check.
Three ways, clearly labeled per card. OFFICIAL = read live from the provider’s published status feed. PROBE = our own reachability and latency check of a public endpoint. UNABLE TO VERIFY = the check was blocked — which is not the same as down.
Providers such as Meta, Baidu, Tencent, Midjourney’s app and several gateways block browser-based checks or publish no machine-readable feed. A blocked check only means we could not see it from your browser — never that the service is down. Each card links the official status page so you can confirm in one click.
OpenAI, Anthropic, Google Cloud, Microsoft Azure, AWS, xAI, DeepSeek, Mistral, Cohere, NVIDIA, Databricks, Snowflake, Cerebras, OpenRouter, Groq, Together AI, Fireworks, Replicate, Hugging Face, fal.ai, Modal, RunPod, Cloudflare, Vercel, Lambda, Perplexity, Adobe, Midjourney, Stability AI, Runway, Synthesia, ElevenLabs, GitHub, Cursor, Windsurf, Lovable and many others publish status information — linked on every card.
Official feeds are re-fetched and endpoints re-probed every 2 minutes, and you can force a fresh pass anytime with "Refresh now". All timestamps shown are UTC.
Yes. OpenAI notes that its published availability is aggregated — an individual user’s availability can differ by model, feature, plan and region. Official feeds are a strong signal, but treat them as platform-level, not a guarantee for every account.
Providers (OpenAI, Anthropic, Google…) train and serve their own models. Gateways (OpenRouter, GroqCloud, Together AI, Fireworks, Cloudflare Workers AI…) resell or host many models behind one API — so a gateway incident can affect dozens of apps built on it.
No. Market Me Global is independent. Official statuses are read from each provider’s public feed with attribution and links; probes are our own measurements from your browser.
Market Me Global builds automation on top of these exact platforms — with fallbacks, health checks and graceful degradation designed in from day one.
