The source of the problem appears to have been that our API was unable to connect to one of its underlying data stores, most likely due to a networking interruption. This has recovered as of 00:02 UTC and traffic is being served normally once again. We will continue to monitor.

We're aware that our API is currently serving slow responses and 5XX-series responses to some traffic and are investigating.