arXiv cs.LG
· Papers
GUI-Perturbed: Domain Randomization Reveals Systematic Brittleness in GUI Grounding Models
arXiv:2604.14262v2 Announce Type: replace Abstract: GUI grounding models report over 85% accuracy on standard benchmarks, yet drop 27-56 percentage points when instructions require spatial reasoning rather than direct element naming. Current benchmarks miss this because they evaluate each screenshot once with a single