Skip to content

Open data

APIdown's incident and uptime history, free to download in CSV or JSON. Useful for research, vendor due diligence, blog posts, or plotting your own charts. No account, no key.

Incident history

Every outage and degradation we have detected across all tracked APIs, with severity, start and resolution times, and duration.

5,000 most recent incidents

Incident history dataset columns
ColumnMeaning
incident_idStable identifier; also the URL suffix at /incidents/{id}
api_slug, api_name, categoryWhich API the incident belongs to
severitycritical, major, or minor
statusinvestigating, identified, monitoring, or resolved
titleShort human description
started_at, resolved_atISO 8601 timestamps; resolved_at empty while open
duration_minutesEmpty while the incident is still open

Monthly uptime

One row per API per calendar month: measured uptime, incident count, total downtime, and longest single outage.

Last 24 months, all tracked APIs

Monthly uptime dataset columns
ColumnMeaning
api_slug, api_name, categoryWhich API the row describes
monthCalendar month as YYYY-MM (UTC)
uptime_pctMeasured uptime for that month, three decimals
incident_countIncidents overlapping the month
downtime_minutesTotal downtime, overlapping incidents merged
longest_outage_minutesLongest single outage in the month

How to use it

  • Attribution. These datasets are free to use, including commercially. We ask that you credit “APIdown.net” and link to https://apidown.net wherever you publish figures derived from them.
  • What the numbers are. Everything here is measured from anonymised client-side signals sent by applications running the APIdown SDK. They are our independent measurement, not the vendor's own reporting, and the two will differ. Client-side measurement includes the network path, so routing and DNS failures count against an API here.
  • What is not included. No user, account, subscriber, or API-key data appears in these files, and no per-request or per-customer data — only public APIs and the incidents recorded against them.
  • Freshness. Both datasets are generated on request and cached for an hour. For live values use the /v1 API instead.

Need this programmatically, filtered, or in bulk beyond the row limits? The public /v1 API covers the same data with query parameters.