# Benchmarking MLLM-based Web Understanding: Reasoning, Robustness and Safety

arXiv: https://arxiv.org/abs/2509.21782 · HTML: https://mlsanity.com/paper/2509.21782/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 10 of its 44 references that are in the map.
- First posted on arXiv 26 Sep 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- Multimodal LLMs / vision-language models / [Chart understanding and anomaly detection](https://mlsanity.com/mllm/chart-understanding-and-anomaly-detection/)
- LLM agents and tool use / [Multimodal and embodied agents](https://mlsanity.com/agents/multimodal-and-embodied-agents/)

## Builds on (most cited first, 10 of 10)

- [MMBench: Is Your Multi-modal Model an All-around Player?](https://mlsanity.com/paper/2307.06281/) (2023)
- [Large Language Model-Brained GUI Agents: A Survey](https://mlsanity.com/paper/2411.18279/) (2024)
- [Interaction2Code: Benchmarking MLLM-based Interactive Webpage Code Generation from Interactive Prototyping](https://mlsanity.com/paper/2411.03292/) (2024)
- [DesignBench: A Comprehensive Benchmark for MLLM-based Front-end Code Generation](https://mlsanity.com/paper/2506.06251/) (2025)
- [MRWeb: An Exploration of Generating Multi-Page Resource-Aware Web Code from UI Designs](https://mlsanity.com/paper/2412.15310/) (2024)
- [SlideCoder: Layout-aware RAG-enhanced Hierarchical Slide Generation from Design](https://mlsanity.com/paper/2506.07964/) (2025)
- [A Survey on the Safety and Security Threats of Computer-Using Agents: JARVIS or Ultron?](https://mlsanity.com/paper/2505.10924/) (2025)
- [QueryAttack: Jailbreaking Aligned Large Language Models Using Structured Non-natural Query Language](https://mlsanity.com/paper/2502.09723/) (2025)
- [PairBench: Are Vision-Language Models Reliable at Comparing What They See?](https://mlsanity.com/paper/2502.15210/) (2025)
- [EfficientUICoder: A Bidirectional Token Compression Framework for Efficient MLLM-Based UI Code Generation](https://mlsanity.com/paper/2509.12159/) (2025)

## Built on by (most cited first, 6 of 6)

- [EfficientPosterGen: Semantic-aware Efficient Poster Generation via Token Compression and Accurate Violation Detection](https://mlsanity.com/paper/2603.00155/) (2026)
- [WebVR: Benchmarking Multimodal LLMs for WebPage Recreation from Videos via Human-Aligned Visual Rubrics](https://mlsanity.com/paper/2603.13391/) (2026)
- [TREAT: A Code LLMs Trustworthiness / Reliability Evaluation and Testing Framework](https://mlsanity.com/paper/2510.17163/) (2025)
- [VulTriage: Triple-Path Context Augmentation for LLM-Based Vulnerability Detection](https://mlsanity.com/paper/2605.09461/) (2026)
- [DCVD: Dual-Channel Cross-Modal Fusion for Joint Vulnerability Detection and Localization](https://mlsanity.com/paper/2605.11015/) (2026)
- [Scenes as an Attack Surface: Uncovering Scene-Grounded Safety Failures in MLLMs](https://mlsanity.com/paper/2610.14094/) (2026)

## Abstract (arXiv)

Multimodal large language models (MLLMs) are increasingly deployed as the core reasoning engine for web-facing systems, powering GUI agents and front-end automation that must interpret page structure, select actionable widgets, and execute multi-step interactions reliably. However, existing benchmarks largely emphasize visual perception or UI code generation, showing insufficient evaluation on the reasoning, robustness and safety capability required for end-to-end web applications. To bridge the gap, we introduce a comprehensive web understanding benchmark, named WebRRSBench, that jointly evaluates Reasoning, Robustness, and Safety across eight tasks, such as position relationship reasoning, color robustness, and safety critical detection, etc. The benchmark is constructed from 729 websites and contains 3799 QA pairs that probe multi-step inference over page structure, text, widgets, and safety-critical interactions. To ensure reliable measurement, we adopt standardized prompts, a protocolized and deterministic evaluation pipeline, and multi-stage quality control combining automatic checks with targeted human verification. We evaluate 11 MLLMs on WebRRSBench. The results reveal significant gaps: models still struggle with compositional and cross-element reasoning over realistic layouts, show limited robustness when facing perturbations in user interfaces and content such as layout rearrangements or visual style shifts, and are rather conservative in recognizing and avoiding safety critical or irreversible actions. Our code and appendix are available at https: //github.com/annoy-worker/WebRSSBench.