The problem has been identified and resolved, and all systems are operating normally.

A recent change to our API key handler introduced a caching issue, causing some requests to fail.