# GUI-Perturbed: Domain Randomization Reveals Systematic Brittleness in GUI Grounding Models

arXiv: https://arxiv.org/abs/2604.14262 · HTML: https://mlsanity.com/paper/2604.14262/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 24 of its 30 references that are in the map.
- First posted on arXiv 15 Apr 2026.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 24 of 24)

- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World](https://mlsanity.com/paper/1703.06907/) (2017)
- [Mind2Web: Towards a Generalist Agent for the Web](https://mlsanity.com/paper/2306.06070/) (2023)
- [OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments](https://mlsanity.com/paper/2404.07972/) (2024)
- [UI-TARS: Pioneering Automated GUI Interaction with Native Agents](https://mlsanity.com/paper/2501.12326/) (2025)
- [SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agents](https://mlsanity.com/paper/2401.10935/) (2024)
- [OS-ATLAS: A Foundation Action Model for Generalist GUI Agents](https://mlsanity.com/paper/2410.23218/) (2024)
- [ScreenSpot-Pro: GUI Grounding for Professional High-Resolution Computer Use](https://mlsanity.com/paper/2504.07981/) (2025)
- [An Illusion of Progress? Assessing the Current State of Web Agents](https://mlsanity.com/paper/2504.01382/) (2025)
- [OpenCUA: Open Foundations for Computer-Use Agents](https://mlsanity.com/paper/2508.09123/) (2025)
- [OmniACT: A Dataset and Benchmark for Enabling Multimodal Generalist Autonomous Agents for Desktop and Web](https://mlsanity.com/paper/2402.17553/) (2024)
- [Aria-UI: Visual Grounding for GUI Instructions](https://mlsanity.com/paper/2412.16256/) (2024)
- [Scaling Computer-Use Grounding via User Interface Decomposition and Synthesis](https://mlsanity.com/paper/2505.13227/) (2025)
- [GTA1: GUI Test-time Scaling Agent](https://mlsanity.com/paper/2507.05791/) (2025)
- [Widget Captioning: Generating Natural Language Description for Mobile User Interface Elements](https://mlsanity.com/paper/2010.04295/) (2020)
- [SpatialLadder: Progressive Training for Spatial Reasoning in Vision-Language Models](https://mlsanity.com/paper/2510.08531/) (2025)
- [UI-Vision: A Desktop-centric GUI Benchmark for Visual Perception and Interaction](https://mlsanity.com/paper/2503.15661/) (2025)
- [Mind2Web 2: Evaluating Agentic Search with Agent-as-a-Judge](https://mlsanity.com/paper/2506.21506/) (2025)
- [EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience](https://mlsanity.com/paper/2601.15876/) (2026)
- [WorldGUI: An Interactive Benchmark for Desktop GUI Automation from Any Starting Point](https://mlsanity.com/paper/2502.08047/) (2025)
- [GUI-Robust: A Comprehensive Dataset for Testing GUI Agent Robustness in Real-World Anomalies](https://mlsanity.com/paper/2506.14477/) (2025)
- [GuirlVG: Incentivize GUI Visual Grounding via Empirical Exploration on Reinforcement Learning](https://mlsanity.com/paper/2508.04389/) (2025)
- [How do Visual Attributes Influence Web Agents? A Comprehensive Evaluation of User Interface Design Factors](https://mlsanity.com/paper/2601.21961/) (2026)
- [GLAD: Generalizable Tuning for Vision-Language Models](https://mlsanity.com/paper/2507.13089/) (2025)

## Built on by (most cited first, 3 of 3)

- [GUI-C2: Coarse-to-Fine GUI Grounding via Difficulty-Aware Reinforcement Learning](https://mlsanity.com/paper/2605.30884/) (2026)
- [ScreenHaystack: Finding Blind Zones in GUI Grounding](https://mlsanity.com/paper/2609.32036/) (2026)
- [Can Agents Work for Everyone? Cross-User Reliability for Mobile GUI Agents in Personalized User Interfaces](https://mlsanity.com/paper/2610.07972/) (2026)

## Abstract (arXiv)

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 fixed instruction. We introduce GUI-Perturbed, a controlled perturbation framework that independently varies visual scenes and instructions to measure grounding robustness. Evaluating three 7B models from the same architecture lineage, we find that relational instructions cause systematic accuracy collapse across all models, a 70% browser zoom produces statistically significant degradation, and rank-8 LoRA fine-tuning with augmented data degrades performance rather than improving it. By perturbing along independent axes, GUI-Perturbed isolates which specific capability axes are affected-spatial reasoning, visual robustness, reasoning calibration-providing diagnostic signal that aggregate benchmarks cannot. We release the dataset, augmentation pipeline, and a fine-tuned model.