mlsanityData to 14 Oct 2026

PCRI: Measuring Context Robustness in Multimodal Models for Enterprise Applications

arXiv 2509.23879 ↗First posted 28 Sep 2025cs.CV cs.AI cs.CLcited by 1

On the map: Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning

Abstract (arXiv)

The reliability of Multimodal Large Language Models (MLLMs) in real-world settings is often undermined by sensitivity to irrelevant or distracting visual context, an aspect not captured by existing evaluation metrics. We introduce the Patch Context Robustness Index (PCRI), the first systematic and interpretable score for quantifying MLLM robustness to variations in visual context granularity, measuring performance changes between localized image patches and full-image input. Applying PCRI to 19 state-of-the-art MLLMs across 15 vision-language benchmarks, we find that most leading models remain brittle to background noise, with only a few, such as InternVL2-26B and Qwen2VL-72B, demonstrating consistent robustness across tasks. PCRI analysis also highlights how different model architectures handle and integrate visual context, offering actionable diagnostic insight for both researchers and practitioners. PCRI enables rigorous comparison of context robustness, supporting principled model selection and guiding the development of future architectures and training strategies for robust, real-world deployment.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 30 of its 49 references that are in the map. First posted on arXiv 28 Sep 2025.

Builds on

30 of 49 references.

Closest ancestors · 7

Same direction, newest first

  1. 2024Survey of Large Multimodal Model Datasets, Application Categories and Taxonomy23 Dec
  2. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling6 Dec
  3. 2024Mini-InternVL: A Flexible-Transfer Pocket Multimodal Model with 5% Parameters and 90% Performance21 Oct
  4. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution18 Sep
  5. 2024LLaVA-OneVision: Easy Visual Task Transfer6 Aug
  6. 2024How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suites25 Apr
  7. 2023InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks21 Dec

Other references · 23

  1. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answeringcited by 1,919
  2. 2024Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phonecited by 1,917
  3. 2019Towards VQA Models That Can Readcited by 1,718
  4. 2022ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoningcited by 1,479
  5. 2015Microsoft COCO Captions: Data Collection and Evaluation Servercited by 1,462
  6. 2018VizWiz Grand Challenge: Answering Visual Questions from Blind Peoplecited by 858
  7. 2025Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scalingcited by 812
  8. 2016A Diagram Is Worth A Dozen Imagescited by 804
  9. 2024VLMEvalKit: An Open-Source Toolkit for Evaluating Large Multi-Modality Modelscited by 455
  10. 2016Analyzing the Behavior of Visual Question Answering Modelscited by 119
  11. 2022Benchmarking Robustness of Multimodal Image-Text Models under Distribution Shiftcited by 24
  12. 2023TRIPS: Efficient Vision-and-Language Pre-training with Text-Relevant Image Patch Selectioncited by 18
11 more
  1. 2023Towards Perceiving Small Visual Details in Zero-shot Visual Question Answering with Multimodal LLMscited by 15
  2. 2024VCR: A Task for Pixel-Level Complex Reasoning in Vision Language Models via Restoring Occluded Textcited by 15
  3. 2025SweEval: Do LLMs Really Swear? A Safety Benchmark for Testing Limits for Enterprise Usecited by 15
  4. 2025Hard Negative Mining for Domain-Specific Retrieval in Enterprise Systemscited by 14
  5. 2024LLM for Barcodes: Generating Diverse Synthetic Data for Identity Documentscited by 12
  6. 2024VLTP: Vision-Language Guided Token Pruning for Task-Oriented Segmentationcited by 11
  7. 2025FS-DAG: Few Shot Domain Adapting Graph Networks for Visually Rich Document Understandingcited by 10
  8. 2025DAIQ: Auditing Demographic Attribute Inference from Question in LLMscited by 10
  9. 2025Who's Asking? Investigating Bias Through the Lens of Disability Framed Queries in LLMscited by 6
  10. 2025SpeechWeave: Diverse Multilingual Synthetic Text & Audio Data Generation Pipeline for Training Text to Speech Modelscited by 5
  11. 2024Efficient Vision-and-Language Pre-training with Text-Relevant Image Patch Selectioncited by 1

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Do Image-Text Metrics Respect Semantic Invariances?cited by 2

Similar, not cited

Close in wording, no citation either way

  1. 2025MVI-Bench: A Comprehensive Benchmark for Evaluating Robustness to Misleading Visual Inputs in LVLMs
  2. 2025RCI: A Score for Evaluating Global and Local Reasoning in Multimodal Benchmarks
  3. 2025MMLongCite: A Benchmark for Evaluating Fidelity of Long-Context Vision-Language Models
  4. 2025Evaluating Robustness of Vision-Language Models Under Noisy Conditions
  5. 2026Scaling Vision-Language Models Is Not Enough to Mitigate Bias
  6. 2025Demystifying the Visual Quality Paradox in Multimodal Large Language Models
  7. 2025The Perceptual Observatory Characterizing Robustness and Grounding in MLLMs
  8. 2025Res-Bench: Benchmarking the Robustness of Multimodal Large Language Models to Dynamic Resolution Input