Service has been fully restored. A cluster update caused a portion of model inference requests to fail, temporarily degrading check quality. The update issue has been resolved, and all systems are operating normally.

Service is fully restored and mitigations are in place while we roll out a permanent firmware fix