All infrastructure problems have been resolved and model predictions and training are available again for the relevant hardware types.

The black-forest-labs/flux-schnell model is fully operational now. Thank you for your patience!

Incident is resolved.