I would like to be able to use scored revisions with a test server. So for example I want to try how well En.Wikipedia model works on Simple English Wikipedia or Commons. It would not be a good idea to use the live system (ores) for this.
This depends on ORES providing the scores for a wiki different from the one a model was trained on. E.g. right now, the ptwiki model only scores revisions from ptwiki, as there is no model trained for any other Portuguese wiki:
returns "not found" errors.