Affected
Partial outage from 8:05 PM to 8:14 PM
One GPU worker went into a bad state, it was restarted and returned to normal operation
Some of the requests to the nightly endpoint (experimental in app) are returning errors