API status

How each call has been answering, measured by the API gateway itself. The page refreshes every minute.

One region. The API runs on one server in Helsinki, Finland. There is no second region to fail over to, and no service-level agreement.

Reading the gateway's numbers…

How to read this

p95: 95 of 100 calls in the hour were answered faster than this. The gateway counts in steps (250 ms, 500 ms, 1 s, 2 s, 5 s…), so it says “under”.
Server errors are 5xx answers. Refusals you can fix (400, 422, 429) don't count.
Some operations have a latency budget; a day over it shows amber. Operations without one show only whether they answered.
Numbers are measured where the gateway answers, not at your end; add your own network time.
A weekly accuracy scorecard (answers checked against reference data) will appear here once it runs every week.