Currently, due to heavy usage (more than 20 sequences/day on average), expect delays for model building.
Please read the documentation as well as some caveats based on the analysis of the server modules' performance before submitting a sequence. In particular, please do realise that the de novo approach is best only for smaller sequences (100 amino acids or less). Also, users submitting a large number of sequences simultaneously will be moved to a separate queue which would be (naturally) slower).
Check current prediction status of your job.