[xorcist]

If anything, it should be safe to assume by now that capabilities don't scale linearly with model size.