Platform status

Checking the platform…

Reading the latest figures from the API.

Task success rate: Loading…; Of the tasks that finished in the last 24 hours, the share that returned a token. Only those are charged.
Median solve time: Loading…; Half of the tasks that succeeded took this long or less, from createTask to the token.
API availability: Loading…; The share of the last 24 hours' minutes in which the API was answering requests.

How the figures are measured

Every figure covers the 24 hours before it was read, in UTC, and comes from our own database: the tasks that finished, and a mark each API server writes every 30 seconds while it is running.
A rate from a handful of tasks says little, so the success rate and the median wait until at least 20 tasks have finished, or succeeded, in the window, and say so meanwhile.
A minute in which no API server was running counts against availability. While the API has run for less than a day, availability covers the time since it started.
The figures are cached for a minute. If this page cannot reach the API at all, it says so rather than showing old numbers.