LLM Status Hub

We make a real API call to each model — the same call your app would make — from 2 independent AWS regions (US-East, EU-West). The region column shows where that call originated, so latency differences reflect real geographic routing.

Latency is measured from request sent to full response received. We flag a model as degraded when median latency exceeds 1,000 ms or the error rate rises above 1%, and down when it stops responding entirely.

Regions monitored: US-East · EU-West

Unlike DownDetector or AI vendor status pages, ModelVantage runs active heartbeats — the same API calls your app makes — giving you specific, high-resolution data to optimize by model and region, not just a binary up/down flag.