# Order Matters: Exploring Order Sensitivity in Multimodal Large Language Models

arXiv: https://arxiv.org/abs/2410.16983 · HTML: https://mlsanity.com/paper/2410.16983/

## Key facts

- Cited by 20 papers 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 32 of its 51 references that are in the map.
- First posted on arXiv 22 Oct 2024.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM training and visual instruction tuning](https://mlsanity.com/mllm/mllm-training-and-visual-instruction-tuning/)

## Builds on (most cited first, 30 of 32)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks](https://mlsanity.com/paper/2312.14238/) (2023)
- [Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understanding](https://mlsanity.com/paper/2306.02858/) (2023)
- [Video-LLaVA: Learning United Visual Representation by Alignment Before Projection](https://mlsanity.com/paper/2311.10122/) (2023)
- [Fantastically Ordered Prompts and Where to Find Them: Overcoming Few-Shot Prompt Order Sensitivity](https://mlsanity.com/paper/2104.08786/) (2021)
- [MVBench: A Comprehensive Multi-modal Video Understanding Benchmark](https://mlsanity.com/paper/2311.17005/) (2023)
- [Large Language Models are not Fair Evaluators](https://mlsanity.com/paper/2305.17926/) (2023)
- [How Can We Know What Language Models Know?](https://mlsanity.com/paper/1911.12543/) (2019)
- [LLaMA-Adapter: Efficient Fine-tuning of Language Models with Zero-init Attention](https://mlsanity.com/paper/2303.16199/) (2023)
- [Multimodal Chain-of-Thought Reasoning in Language Models](https://mlsanity.com/paper/2302.00923/) (2023)
- [Generative Multimodal Models are In-Context Learners](https://mlsanity.com/paper/2312.13286/) (2023)
- [Learning How to Ask: Querying LMs with Mixtures of Soft Prompts](https://mlsanity.com/paper/2104.06599/) (2021)
- [MLLM-as-a-Judge: Assessing Multimodal LLM-as-a-Judge with Vision-Language Benchmark](https://mlsanity.com/paper/2402.04788/) (2024)
- [Hard Prompts Made Easy: Gradient-Based Discrete Optimization for Prompt Tuning and Discovery](https://mlsanity.com/paper/2302.03668/) (2023)
- [InternLM-XComposer2: Mastering Free-form Text-Image Composition and Comprehension in Vision-Language Large Model](https://mlsanity.com/paper/2401.16420/) (2024)
- [Prompting Visual-Language Models for Efficient Video Understanding](https://mlsanity.com/paper/2112.04478/) (2021)
- [MuRAG: Multimodal Retrieval-Augmented Generator for Open Question Answering over Images and Text](https://mlsanity.com/paper/2210.02928/) (2022)
- [GPT3Mix: Leveraging Large-scale Language Models for Text Augmentation](https://mlsanity.com/paper/2104.08826/) (2021)
- [TokenPacker: Efficient Visual Projector for Multimodal LLM](https://mlsanity.com/paper/2407.02392/) (2024)
- [Compositional Chain-of-Thought Prompting for Large Multimodal Models](https://mlsanity.com/paper/2311.17076/) (2023)
- [Self-Adaptive In-Context Learning: An Information Compression Perspective for In-Context Example Selection and Ordering](https://mlsanity.com/paper/2212.10375/) (2022)
- [On the Adversarial Robustness of Multi-Modal Foundation Models](https://mlsanity.com/paper/2308.10741/) (2023)
- [Wiki-LLaVA: Hierarchical Retrieval-Augmented Generation for Multimodal LLMs](https://mlsanity.com/paper/2404.15406/) (2024)
- [Split and Merge: Aligning Position Biases in LLM-based Evaluators](https://mlsanity.com/paper/2310.01432/) (2023)
- [High-fidelity 3D GAN Inversion by Pseudo-multi-view Optimization](https://mlsanity.com/paper/2211.15662/) (2022)
- [Exploring Diverse In-Context Configurations for Image Captioning](https://mlsanity.com/paper/2305.14800/) (2023)
- [Reordering Examples Helps during Priming-based Few-Shot Learning](https://mlsanity.com/paper/2106.01751/) (2021)
- [Hierarchical Context Merging: Better Long Context Understanding for Pre-trained LLMs](https://mlsanity.com/paper/2404.10308/) (2024)

## Built on by (most cited first, 20 of 20)

- [A Survey of Context Engineering for Large Language Models](https://mlsanity.com/paper/2507.13334/) (2025)
- [Who is in the Spotlight: The Hidden Bias Undermining Multimodal Retrieval-Augmented Generation](https://mlsanity.com/paper/2506.11063/) (2025)
- [Residual Decoding: Mitigating Hallucinations in Large Vision-Language Models via History-Aware Residual Guidance](https://mlsanity.com/paper/2602.01047/) (2026)
- [GraphSOS: Graph Sampling and Order Selection to Help LLMs Understand Graphs Better](https://mlsanity.com/paper/2501.14427/) (2025)
- [ViKey: Enhancing Temporal Understanding in Videos via Visual Prompting](https://mlsanity.com/paper/2603.23186/) (2026)
- [OMHBench: Benchmarking Balanced and Grounded Omni-Modal Multi-Hop Reasoning](https://mlsanity.com/paper/2508.16198/) (2025)
- [MLLM-based Speech Recognition: When and How is Multimodality Beneficial?](https://mlsanity.com/paper/2507.19037/) (2025)
- [Unexplored flaws in multiple-choice VQA make benchmarking unreliable](https://mlsanity.com/paper/2511.22341/) (2025)
- [Chain of Modality: From Static Fusion to Dynamic Orchestration in Omni-MLLMs](https://mlsanity.com/paper/2604.14520/) (2026)
- [Logit-Attention Divergence: Mitigating Position Bias in Multi-Image Retrieval via Attention-Guided Calibration](https://mlsanity.com/paper/2605.11591/) (2026)
- [MetaRA: Metamorphic Robustness Assessment for Multimodal Large Language Model-based Visual Question Answering Systems](https://mlsanity.com/paper/2605.19307/) (2026)
- [Detecting Differences Is Not Understanding Structure: Large Language Models Fail at Graph Isomorphism](https://mlsanity.com/paper/2606.09484/) (2026)
- [Lost at the End: Primacy Bias in Multimodal Retrieval-Augmented Question Answering](https://mlsanity.com/paper/2606.16494/) (2026)
- [Same Evidence, Different Answer: Auditing Order Sensitivity in Multimodal Large Language Models](https://mlsanity.com/paper/2606.26079/) (2026)
- [Mitigating Hallucinations on Object Attributes using Multiview Images and Negative Instructions](https://mlsanity.com/paper/2501.10011/) (2025)
- [Where is the multimodal goal post? On the Ability of Foundation Models to Recognize Highlights in Videos](https://mlsanity.com/paper/2601.16333/) (2026)
- [Where Do Vision-Language Models Fail? World Scale Analysis for Image Geolocalization](https://mlsanity.com/paper/2604.16248/) (2026)
- [Ask Twice, Look Twice: Prompt Echoing Resolves the Question-First Paradox in Vision-Language Models](https://mlsanity.com/paper/2607.15565/) (2026)
- [Same evidence, different judgments: Evidence noncommutative in vision/speech-text conflicts](https://mlsanity.com/paper/2609.26986/) (2026)
- [ActICL: Post-Training for Agentic Multimodal In-Context Learning](https://mlsanity.com/paper/2610.19475/) (2026)

## Abstract (arXiv)

Multimodal Large Language Models (MLLMs) utilize multimodal contexts consisting of text, images, or videos to solve various multimodal tasks. However, we find that changing the order of multimodal input can cause the model's performance to fluctuate between advanced performance and random guessing. This phenomenon exists in both single-modality (text-only or image-only) and mixed-modality (image-text-pair) contexts. Furthermore, we demonstrate that popular MLLMs pay special attention to certain multimodal context positions, particularly the beginning and end. Leveraging this special attention, we place key video frames and important image/text content in special positions within the context and submit them to the MLLM for inference. This method results in average performance gains of 14.7% for video-caption matching and 17.8% for visual question answering tasks. Additionally, we propose a new metric, Position-Invariant Accuracy (PIA), to address order bias in MLLM evaluation. Our research findings contribute to a better understanding of Multi-Modal In-Context Learning (MMICL) and provide practical strategies for enhancing MLLM performance without increasing computational costs.