Efficient Multimodal Inference

Follow this topic Ask about this topic

Momentum

28 papers in the last four weeks, up 211% on the four weeks before. 0.3% of all new papers.

Jul 13Week of Sep 28

Latest papers 157

Jun 23, 2026cs.CV

Towards Fast and Effective Long Video Understanding of Multimodal Large Language Models via Adaptive Quasi-Gaussian Sampling

Long video understanding remains a daunting challenge for Multimodal Large Language Models (MLLMs) due to the excessive computation and memory footprint. Thus, keyframe selection is often adopted to mitigate this shortcoming, which however still suffers from low flexibility and high noise due to its hard sampling principle. In this paper, we define video frame selection as a problem of Quasi-Gaussian Sampling, and propose an adaptive and training-free approach termed AdaQ. Inspired by the 3-

σ

rule of Gaussian distribution, the objective of AdaQ is to achieve the optimal 3-

σ

interval for different examples, i.e., a smaller 3-

σ

interval for the local query and a larger one for the global query, thereby facilitating robust and adaptive frame sampling. To validate AdaQ, we apply it to four MLLMs with three embedding models. The extensive experimental results not only show its obvious performance gains over the default MLLMs and the SOTA keyframe selection methods, e.g., helping Qwen3-VL-8B outperform GPT4o by 15.8% on average by using only 64 frames, but also confirm its superior robustness and high efficiency for long-video understanding, e.g., only 1 hyper-parameter needs to be set.

Kun Zhang, Chenxin Fang, Tao Chen +4

Key Laboratory of Multimedia Trusted Perception and Efficient Computing, Ministry of Education of China, Xiamen University, 361005, P.R. China

Efficient Multimodal Inference Multimodal Large Language Models Long-Video Understanding Video Frame Selection

Jun 23, 2026cs.CV

Spectral Evolution-Guided Token Pruning in Multimodal Large Language Models

Reducing visual token redundancy is critical for accelerating Multimodal Large Language Models (MLLMs) without degrading cross-modal reasoning performance. Existing token pruning methods typically rely on single-layer signals, such as attention scores or token similarities, which overlook the cross-layer transformation of visual representations and may exhibit positional bias in multimodal token sequences. To address this limitation, we propose a training-free token pruning framework based on Cross-Layer Spectral Evolution (CLSE). Instead of measuring token importance from single-layer feature magnitudes, CLSE quantifies how token representations evolve across Transformer layers in the frequency domain. This evolution reflects the transition from high-frequency structural details to low-frequency semantic abstractions. We observe that tokens with stronger spectral redistribution across layers are more likely to be semantically active and should therefore be preserved. By modeling cross-layer token dynamics, CLSE provides a stable importance criterion that mitigates positional bias. Extensive experiments on both image and video benchmarks demonstrate that CLSE achieves a superior trade-off between efficiency and accuracy under aggressive token reduction. Across multiple MLLMs, CLSE reduces FLOPs, KV cache memory, and latency while maintaining competitive or improved performance.

Bin Chen, Yuxiang Cai, Yadan Luo +3

School of Software Technology, Zhejiang University, Ningbo, China · Zhejiang Key Laboratory of Digital-Intelligence Service Technology, China · The University of Queensland, St Lucia, QLD, Australia +2

Efficient Multimodal Inference Multimodal Large Language Models Token Pruning Multimodal Token Compression Visual Token Pruning

Jun 22, 2026cs.DC

Kamera: Unified Position-Invariant Multimodal KV Cache for Training-Free Reuse

Multimodal agents repeatedly re-examine the same video frames, UI screenshots, and rendered artifacts as their context window slides and reasoning iterates, yet every look-back re-encodes from scratch, because prefix caches serve reuse only at a fixed leading position. We show this recompute is avoidable, and identify exactly what naive KV reuse loses: the cross-chunk conditioning a chunk absorbs from its neighbours. This loss is asymmetric. The direct readout of a cached chunk is recovered exactly and for free by the standard state-merge. What remains is a diffuse, low-rank residue concentrated in deep layers, invisible to single-hop retrieval but precisely what multi-hop reasoning binds on. Blind reuse therefore leaves single-hop recall intact while halving multi-hop accuracy; this is the failure mode prior position-independent caches, designed for single-context or single-image reuse, do not address. We repair it with a small, training-free low-rank conditioning patch stored alongside each position-free chunk. Reuse reduces to one operator across MLA, GQA, and MHA: exact RoPE re-rotation to any target position, plus the patch that restores cross-chunk binding. This makes three window operations cheap: reorder (one patch serves every ordering of a cached set), sliding-window survival (surviving chunks relocate via rotation only, zero re-encode), and recall (an evicted chunk is rehydrated by its patch, never re-encoded). A rank-m patch recovers full task accuracy on cross-chunk-binding benchmarks, MM-NIAH across two attention families and two-page doc-QA, at a fraction of the KV footprint, and reconstructs re-prefill KV to within bf16 rounding in a production SGLang kernel across six backbones. The conditioning signal is strongest in redundant vision and video streams, making our solution most impactful where multimodal agents spend their recompute budget.

Bole Ma, Jan Eitzinger, Harald Koestler +1

Efficient Multimodal Inference KV Caching Multimodal Reasoning KV-Cache Reuse Multimodal Agents

Jun 21, 2026cs.AI

Efficient Multimodal Clinical Question Answering for Pulmonary Embolism Risk Assessment

Pulmonary embolism (PE) is a high risk cardiopulmonary condition whose management requires both timely diagnosis and reliable assessment of future clinical risk. Because PE care routinely combines computed tomography pulmonary angiography (CTPA), radiology interpretation, and longitudinal electronic health record (EHR) evidence, it provides a clinically meaningful setting for evaluating compact multimodal language models. In this work, we build a benchmark using efficient multimodal large language models (MLLMs) on INSPECT, a multimodal PE dataset containing 23,248 CTPA studies from 19,402 patients. We formulate eight diagnostic and prognostic tasks as structured clinical question answering problems and evaluate on typical efficient MLLMs under CTPA-Only, EHR-Only, and CTPA+EHR settings with zero-shot and few-shot prompting. Results show that Gemma4 E4B and Gemma4 E2B perform more strongly when EHR evidence is available, especially under CTPA+EHR input. Task level analysis further shows that PE diagnosis achieves higher performance than prognostic tasks, particularly readmission prediction. These observations suggest that compact multimodal models have the great potential in early stage PE risk detection and explanation.

Xiangyuan Xue, Yang Yu, Yan Gao +5

University of Auckland, Auckland, New Zealand · University of Cambridge, Cambridge, United Kingdom · University of Adelaide, Adelaide, Australia +1

Efficient Multimodal Inference Multimodal QA Medical VLMs Multimodal Model Evaluation Clinical Risk Prediction

Jun 17, 2026cs.CV

PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models

Multimodal large language models (MLLMs) have achieved remarkable progress in visual understanding tasks. However, most existing MLLMs rely on autoregressive generation, which limits their efficiency for perception tasks that require captioning multiple regions. In this work, we propose PerceptionDLM, a multimodal diffusion language model optimized for efficient parallel region perception. Built upon PerceptionDLM-Base, a strong foundational baseline that achieves state-of-the-art performance among open-source diffusion MLLMs, our architecture fully leverages the parallel decoding nature of DLMs. Specifically, we introduce efficient prompting and structured attention masking to enable simultaneous perception of multiple masked regions, allowing the model to generate region descriptions in parallel at both the sequence and token levels. This design significantly improves inference efficiency compared with existing approaches that process regions sequentially. To systematically evaluate the parallelism property of visual perception capability for DLMs, we construct a new Parallel Detailed Localized Captioning Benchmark (ParaDLC-Bench) by scaling the DLC-Bench to include multiple region masks per image, enabling joint evaluation of both caption quality and inference efficiency. Experiments demonstrate that PerceptionDLM maintains competitive performance in region captioning while achieving substantial speed improvements for multi-region perception tasks. Our results highlight the potential of multimodal diffusion language models for efficient, parallel visual perception. To the best of our knowledge, we are the first to achieve parallel region caption and perception by leveraging the advantages of diffusion language models. Code, models, and datasets are released.

Yueyi Sun, Yuhao Wang, Jason Li +8

Peking University MSALab, ByteDance

Efficient Multimodal Inference Image Captioning Parallel Decoding Multimodal Diffusion Models

Jun 17, 2026cs.CV

Visual-OPSD: Cross-Modal On-Policy Self-Distillation for Efficient Unified Multimodal Reasoning

Unified multimodal models (UMMs) interleave generated ''visual thoughts'' (VTs) with text reasoning to improve spatial tasks. This incurs roughly an order-of-magnitude inference cost from multi-step diffusion. We find this cost yields limited direct benefit. On ThinkMorph, removing or noising VTs barely changes accuracy across nine benchmarks. Once rendered, attention concentrates on the VT regardless of content. Yet a KL diagnostic shows that conditioning on a privileged VT trace shifts the model's completion distribution. This suggests the generation pathway encodes useful reasoning beyond the rendered pixels. Motivated by this gap, we propose Visual On-Policy Self-Distillation(Visual-OPSD). Teacher and student share identical weights but differ in context: the teacher sees privileged VTs while the student sees only the question. Token-level JSD distillation on on-policy student trajectories transfers the teacher's reasoning to a text-only student. Across nine benchmarks, Visual-OPSD improves over its generative teacher by

+3.40

pp with

14.3\times

speedup (10.0s vs. 142.8s per sample) and outperforms same-scale VLMs by

+63.83

pp on VSP. A Gaussian-noise control (

+0.40

pp vs.

+10.28

pp for real VTs) and

58.4\%

closure of the KL gap confirm that gains come from the semantic content of the generation pathway.

Pengyu Li, Zhitao Gao, Lingling Zhang +5

Xi’an Jiaotong University · Shaanxi Province Key Laboratory of Big Data Knowledge Engineering · MOE KLINNS Lab, Xi’an Jiaotong University +1

Cross-Modal Knowledge Distillation Efficient Multimodal Inference Unified Multimodal Models On-Policy Self-Distillation Multimodal Reasoning Multimodal Foundation Models+1

Jun 14, 2026cs.CV

Stepwise Token Selection for Efficient Multimodal Large Language Models

In multimodal large language models (MLLMs), inference cost is largely dominated by the visual token prefix rather than the language backbone, making token reduction a key factor for improving efficiency. Existing approaches typically assign independent importance scores to visual tokens and retain a fixed number of top-ranked tokens, implicitly assuming token independence and a uniform compression ratio across inputs. In this work, we reformulate visual token pruning as a sequential decision-making process. Specifically, we introduce a pointer-style selection mechanism that iteratively chooses informative tokens, conditioning each decision on previously selected ones, and dynamically determines when to stop via a learned termination action. This enables joint optimization of both the selected subset and its size. To enable end-to-end training under standard language modeling objectives, we design a differentiable relaxation based on a variance-preserving noise interpolation scheme, allowing gradients to propagate through the discrete selection process. Extensive experiments on LLaVA-v1.5-7B and Qwen2.5-VL-7B demonstrate that our approach consistently outperforms fixed-ratio baselines across different compression levels. Under aggressive pruning that removes 88.9% of visual tokens, our method preserves 94.6% of the original accuracy while achieving a 1.88x speed-up in prefill latency.

Landi He, Shawn Young, Lijian Xu

Shenzhen University of Advanced Technology

Efficient Multimodal Inference Multimodal Large Language Models Multimodal Token Compression Visual Token Pruning

Jun 13, 2026cs.CV

DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning

Reasoning capabilities of multimodal large language models (MLLMs) have improved considerably in recent years. Existing approaches typically rely on explicit chain-of-thought or continuous latent-space trajectories to enhance multi-step reasoning. However, these methods generally assume that an input admits a single latent interpretation and unfold reasoning along a fixed path or under a uniform computation budget. In real-world multimodal settings, visual observations are often subject to occlusion, blur, viewpoint variation, or semantic ambiguity, giving rise to multiple plausible interpretations. A uniform reasoning strategy not only limits the model's ability to explore multiple hypotheses but also incurs high memory usage and rollout cost. We present DLWM (Diverse Latent World Models), a multimodal reasoning framework that combines latent-space reasoning with reinforcement learning. First, we construct a set of diverse latent world hypotheses in continuous latent space, each capturing a different plausible interpretation of the visual input, and unfold latent reasoning independently on each hypothesis. An orthogonality-based diversity regularizer explicitly prevents hypothesis collapse. Second, we formulate the latent reasoning process as a resource-constrained sequential decision problem and introduce a resource-aware reinforcement learning policy that adaptively allocates computation across hypotheses, dynamically deciding whether to expand, terminate, or merge reasoning paths, thereby substantially reducing memory footprint and improving rollout efficiency. Experiments on multiple multimodal reasoning benchmarks demonstrate that DLWM outperforms existing methods by 2-5 points in accuracy while reducing memory usage by 24%.

David Huang, Lianlei Shan

University of Toronto · Tsinghua University

Efficient Multimodal Inference Multimodal Reasoning Latent Visual Reasoning Latent World Models RL for VLMs Multimodal Understanding

Jun 11, 2026cs.LG

ProMUSE: Progressive Multi-modal Uncertainty-guided Staged Evidential Alzheimer Disease Classification

Alzheimer's disease (AD) is a fatal disorder that destroys memory and cognitive skills in the elderly population. Most treatments for AD are effective in the early stage, leading to an increasing demand for early AD diagnosis. AD diagnosis increasingly relies on multimodal data such as clinical assessments, structural Magnetic Resonance Imaging (MRI), and Positron Emission Tomography (PET) imaging. However, MRI and PET acquisition remain costly and not universally accessible, making full-modality inference impractical in real-world clinical workflows. We propose ProMUSE, a Progressive Multi-modal Uncertainty Guided Staged Evidential Network that adaptively determines when additional modalities are necessary, helping reduce the overall cost of data acquisition while maintaining accuracy. ProMUSE first performs evidential classification using low-cost clinical data and quantifies uncertainty via a Dirichlet-based subjective logic model. When uncertainty exceeds a learned threshold, ProMUSE progressively incorporates MRI or PET features, fusing modality-wise belief and uncertainty through Dempster-Shafer theory to obtain a calibrated multimodal prediction. This staged acquisition strategy enables accurate diagnosis while minimizing reliance on expensive imaging. Experiments on ADNI, AIBL, and OASIS across CN-AD, CN-MCI, and MCI-AD tasks demonstrate that ProMUSE achieves competitive or superior accuracy compared to full-modality baselines while reducing MRI/PET usage by 50-90%, yielding substantial cost savings. These results highlight ProMUSE as a practical, uncertainty-aware, and resource-efficient solution for real-world AD screening.

Long Doan, Branden Chen, Ethan Litton +6

Department of Computer Science, Kennesaw State University · Department of Data Science and Analytics, Kennesaw State University · Department of Information Technology, Kennesaw State University +2

Efficient Multimodal Inference Evidential Deep Learning Medical Diagnosis Multimodal Classification Alzheimer's Disease Classification Multimodal Fusion

Jun 11, 2026cs.CV

LaME: Learning to Think in Latent Space for Multimodal Embedding via Information Bottleneck

Reasoning-driven universal multimodal embedding has advanced rapidly by introducing Chain-of-Thought (CoT) reasoning into the embedding pipeline. Despite the strong performance across both general and complex tasks, this paradigm suffers from two core limitations: (i) autoregressive CoT reasoning incurs high computational cost, making it impractical for low-latency retrieval; and (ii) embedding performance is heavily coupled with CoT annotation quality, making large-scale training unreliable. These raise fundamental questions: Is textual CoT the optimal form of reasoning for embedding, and can effective embedding reasoning be accomplished in latent space? To this end, we propose LaME (Latent Reasoning Multimodal Embedding), which formulates embedding-oriented latent reasoning as a weakly supervised information bottleneck. LaME employs K learnable reason tokens as a fixed-capacity bottleneck, completing all reasoning within a single forward pass. The two weak supervision signals structurally decouple contrastive from autoregressive objectives and eliminate dependence on CoT annotations, while a two-stage training pipeline ensures stable convergence. Experiments on MMEB-v2 and MRMR show that LaME achieves competitive performance, surpassing some explicit CoT-based models, while delivering 60x faster inference than explicit CoT methods and 2x faster than latent baselines with throughput comparable to discriminative embedding models. Code is available at https://github.com/PeppaWu/LaME.

Peixi Wu, Biao Yang, Feipeng Ma +7

University of Science and Technology of China · 2Kuaishou Technology · 3Zhejiang University +1

Efficient Multimodal Inference Multimodal Embedding Multimodal Contrastive Learning Latent Reasoning

Jun 10, 2026cs.LG

M*: A Modular, Extensible, Serving System for Multimodal Models

We are entering a new era of composite model architectures that integrate diverse components such as vision encoders, language backbones, diffusion and flow heads, audio codecs, action generators, and world-model predictors. Such architectures underpin a broad class of multimodal models, including unified multimodal models, omni models, speech-language models, vision-language-action policies, and world models. However, existing model serving frameworks were built on narrow assumptions about model structure, making them ill-suited to accommodate this new architectural diversity. Here we present M*, a universal serving system for efficient serving of composite AI models. M* represents models as dataflow graphs, processing requests spanning diverse modalities and tasks as traversals over these graphs. The core insight is a modular abstraction that supports arbitrary composition of model components, flexible placement onto a physical cluster, and model-agnostic optimizations within a distributed runtime. We call this abstraction the Walk Graph and show how it can concisely capture composite models from a broad range of families. We instantiate M* on representative models and find that it achieves, on average, 20% lower end-to-end latency than vLLM-Omni for text-to-image workloads on BAGEL, while delivering up to 2.9x lower real-time factor and 2.7x higher throughput for text-to-speech workloads on Qwen3-Omni. M* also outperforms the V-JEPA 2-AC rollout baseline for robotic planning by up to 12.5x. Thus, our work paves the road towards more efficient serving of complex models with minimal developer effort.

Atindra Jha, Naomi Sagan, Keisuke Kamahori +9

Stanford University · University of Washington · Carnegie Mellon University

Efficient Multimodal Inference Distributed Inference

Jun 10, 2026cs.CV

InternVideo3: Agentify Foundation Models with Multimodal Contextual Reasoning

Recent progress in foundation models has shifted toward agentic behavior involving multi-step reasoning and tool use. However, open-source efforts largely focus on text-dominant settings, leaving long-horizon multimodal tasks underexplored. This gap is evident in video tasks requiring sustained temporal understanding and iterative interaction. We present InternVideo3, a framework enhancing these capabilities via Multimodal Contextual Reasoning (MCR). MCR treats understanding as a closed-loop process over a shared, evolving context containing observations, instructions, reasoning, tool actions, and memory. This frames long-video understanding as evidence accumulation and verification. To ensure efficiency, we introduce Multimodal Multi-head Latent Attention (M^2LA), a token-preserving reparameterization compressing KV-cache states while retaining the full token stream. Our staged training includes continued pretraining, short-to-long supervised fine-tuning, rule-based reinforcement learning, and on-policy distillation. Experiments show InternVideo3 achieves strong performance on benchmarks like Video-MME, MLVU, and EgoSchema. We further instantiate the model as a video agent with retrieval tools, demonstrating robust evidence-grounded behavior. Our results suggest that efficient context handling and closed-loop reasoning are vital for adapting open multimodal models toward long-horizon visually grounded agency.

Ziang Yan, Sheng Xia, Jiashuo Yu +10

Shanghai AI Laboratory · Shanghai Innovation Institute · Nanjing University

Efficient Multimodal Inference Multimodal Reasoning Long-Video Understanding Multimodal Foundation Models Multimodal Agents Temporal Video Understanding

Jun 8, 2026cs.IR

ReasonRec: A Reasoning-Augmented Multimodal Agent for Unified Recommendation

Recent advances in multimodal recommenders excel at feature fusion but remain opaque and inefficient decision-makers, lacking explicit reasoning and self-awareness of uncertainty. We introduce ReasonRec, a reasoning-augmented multimodal agent structured around a three-stage explicit reasoning pipeline. Specifically, we propose a reasoning-aware visual instruction tuning strategy that systematically transforms diverse recommendation tasks into unified CoT prompts, enabling the VLM to explicitly articulate intermediate decision steps. Additionally, our evidence-horizon curriculum progressively enhances the reasoning complexity to better handle cold-start and long-tail user scenarios, significantly boosting model generalization. Furthermore, the uncertainty-guided delegation mechanism empowers the agent to assess its own confidence, strategically allocating computational resources to optimize both recommendation accuracy and inference efficiency. Comprehensive experiments on four standard recommendation tasks across five real-world datasets demonstrate that ReasonRec achieves over 30% relative improvement in key ranking metrics compared to state-of-the-art multimodal recommenders. Crucially, ReasonRec substantially reduces inference latency by dynamically delegating up to 35% of queries to efficient sub-models without compromising accuracy. Extensive ablation studies further confirm that each proposed reasoning and planning mechanism individually contributes substantially to ReasonRec's overall effectiveness. Collectively, our results illustrate a clear pathway towards interpretable, adaptive, and efficient multimodal recommendation through explicit reasoning and agentic design.

Yihua Zhang, Mingfu Liang, Jiyan Yang +11

Meta AI · Michigan State University · The University of North Carolina at Chapel Hill +1

Efficient Multimodal Inference Multimodal CoT Reasoning Recommender Systems Large Language Model-Based Recommendation Multimodal Recommendation Multimodal Agents+1

Jun 8, 2026cs.AI

From Senses to Decisions: The Information Flow of Auditory and Visual Perception in Multimodal LLMs

Multimodal Large Language Models (MLLMs) can listen and see, but how do audio and visual signals actually travel through the network to shape an answer? Despite their growing role in research and real-world applications, the internal pathways through which audio and visual tokens influence the final prediction remain poorly understood. In this study, we examine audio-visual information flow inside Audio-Visual Large Language Models (AVLLMs), tracing how AVLLMs route, utilize, and integrate audio and visual information across two input configurations, audio-visual video and multiple interleaved audio-visual items. We find that for audio-visual video, AVLLMs follow the sequential information flow pathway established for VLMs and VideoLLMs, with audio and visual contribution flowing along this pathway in proportion to the task's reliance on each modality. In settings with multiple interleaved audio-visual items, this routing shifts to different parallel streams. Furthermore, we demonstrate that audio-visual and other token types can be discarded once their information is transferred to LLM, with minimal impact on the model's prediction or even slight improvement, generalizing across multiple tasks and datasets, enabling more efficient inference. These findings hold across multiple models and scales, Qwen2.5-Omni and Video-SALMONN2 Plus at 3B and 7B scales, leading to hypotheses on why these flow structures emerge. Together, these results deliver the first coherent picture of how AVLLMs orchestrate sound and sight inside the network and lay the groundwork for the next wave of interpretability, design, and efficiency advances in audio-visual and broader MLLMs.

Wish Suharitdamrong, Muhammad Awais, Xiatian Zhu +1

Surrey Institute for People-Centred AI (PAI), University of Surrey, UK · Centre for Vision, Speech and Signal Processing (CVSSP), University of Surrey, UK

Efficient Multimodal Inference Multimodal Large Language Models Neural Network Interpretability Multimodal Model Interpretability Multimodal Understanding

Jun 8, 2026cs.AI

Late-Layer Fusion is Enough: Dual-Path Vision Token Routing for Multimodal Large Language Models under Visual Saturation

Multimodal large language models (MLLMs) commonly inherit the deep, symmetric Transformer backbone designed for unimodal text modeling, and apply the same computation uniformly to image and language tokens. This design overlooks a key modality asymmetry: image and text tokens differ substantially in information density, redundancy, and required reasoning depth. Through a layer-wise analysis of LLaVA-1.5, we observe that vision tokens tend to saturate in the middle layers. Specifically, text-to-image attention decreases from 0.68 at layer 0 to 0.07 by layer 4, and stabilizes near 0.04 after layer 18, whereas text tokens continue to benefit from deep semantic processing. These findings suggest a mismatch between architectural symmetry and depth-asynchronous modality evolution, resulting in redundant visual computation and possible drift in perceptual representations during deep task-specific adaptation. Motivated by this, we propose Dual-Path Vision Token Routing (DPVR), a modality-asymmetric routing framework for efficient MLLMs. Its core instantiation, DPVR-LF (Late-Layer Fusion), routes vision tokens at the saturation point into a one-layer trainable side branch, runs a thirteen-layer text-only forward that skips image positions in the deep stack, and re-fuses the visual and textual streams only at the final layer. With approximately 3% trainable parameters, DPVR-LF preserves competitive multimodal performance on standard benchmarks while reducing visual computation in the deep Transformer stack. The results challenge the conventional assumption that vision tokens must traverse all deep language-model layers, and indicate that a single late fusion layer can be sufficient for maintaining strong perceptual competence in LLaVA-style MLLMs.

Siyuan Liu, Jinyang Wu

aSchool of Mechanics and Engineering Science, Peking University, China · bDepartment of Automation, Tsinghua University, China

Efficient Multimodal Inference Multimodal Large Language Models Multimodal Fusion

Jun 7, 2026cs.CV

Look Less, Reason More: Block-wise Attention Skipping for Efficient Multimodal LLMs

Multimodal Large Language Models (MLLMs) face a significant inference bottleneck due to the quadratic computational cost of self-attention over long visual token sequences. However, we identify a critical inefficiency in current architectures: Visual Attention Saturation. Our analysis reveals that visual tokens rapidly establish their spatial structure and intra-modal relationships in early layers, rendering visual-to-visual self-attention in deeper layers computationally redundant. Conversely, Feed-Forward Networks (FFNs) in these layers remain essential for projecting visual features into the evolving textual semantic space. Leveraging this insight, we present Visual-Skip (V-Skip), a training-free inference paradigm that decouples spatial interaction from semantic evolution. Rather than discarding tokens, V-Skip imposes block-wise structured sparsity by selectively bypassing saturated visual self-attention modules. Furthermore, recognizing that varying downstream tasks demand distinct reasoning depths, V-Skip employs a lightweight, few-shot calibration to dynamically route the task-optimal sparsity path. Extensive experiments demonstrate that V-Skip effectively bypasses redundant vision attention to achieve block-wise sparsity, maintaining a 94.16% to 100.31% performance retention across diverse MLLMs. Ultimately, we prove that to reason more effectively, models do not need to discard what they see -- they simply need to "look less" at the right depth.

Jie Ma, Zhike Qiu, Jiayi Ji +2

Xiamen University

Efficient Multimodal Inference Multimodal Large Language Models Efficient Attention Sparse Attention

Jun 5, 2026cs.CL

M $^3$ Exam: Benchmarking Multimodal Memory for Realistic User-Agent Interactions

Language agents are increasingly deployed over accumulating multimodal information, yet existing benchmarks assume a human-human form with sparse visuals and straightforward content, evaluating neither reasoning over authentic multimodal file interaction nor the interpretation of concealed user information. We therefore introduce M

^3

Exam, a query-centric multimodal conversational memory benchmark built on realistic user-agent interaction, with multi-dimensional evaluation spanning cross-modal grounding and implicit information inference. Benchmarking MLLMs and memory systems reveals persistent gaps in cross-modal grounding, cross session reasoning, and the efficiency cost of accumulating multimodal context. We further propose M

^3

Proctor, a multimodal memory method that detects query modality bias and consumes raw visual sources only on demand, improving accuracy by 13% while cutting index-construction time and retrieved tokens by over 70%.

Zhengjun Huang, Wenxuan Liu, Zhoujin Tian +6

The Hong Kong University of Science and Technology · Peng Cheng Laboratory · Beijing University of Chemical Technology +4

Multimodal Grounding Efficient Multimodal Inference Multimodal Memory AI Agent Benchmarks Memory Agent Benchmarks LLM Reasoning+1

Jun 5, 2026cs.RO

Dreaming when Necessary: Advancing World Action Models with Adaptive Multi-Modal Reasoning

World Action Models (WAMs) offer a promising approach to embodied intelligence, yet existing methods rely heavily on video prediction as action priors and lack adaptive multimodal reasoning, limiting their effectiveness on long-horizon, complex tasks. We observe that WAMs require different multimodal reasoning modes under different execution contexts: textual reasoning is essential during task transitions to guide high-level action prediction, while visual reasoning is critical during fine-grained manipulation for precise control. Motivated by this observation, we propose \textbf{AdaWAM}, a world action model with adaptive multimodal reasoning abilities. AdaWAM integrates a lightweight dynamic router that autonomously triggers textual or visual reasoning as needed during task execution. Experiments on both simulated and real-world embodied tasks show that AdaWAM substantially improves inference efficiency while outperforming state-of-the-art embodied policies. Codes and demos are available at: https://adawam.github.io/.

Yinzhou Tang, Jingbo Xu, Yu Shang +4

Tsinghua University · Manifold AI

Efficient Multimodal Inference World Models for Robotics Multimodal Reasoning World Action Models

Jun 4, 2026cs.CV

DisasterBench: A Multimodal Benchmark for UAV-Based Disaster Response in Complex Environments

When a disaster unfolds, responders must answer not only what is happening, but also why it is happening, what will happen next, and what to do now, often from noisy low-altitude UAV views and under tight on-site compute constraints. However, most existing multimodal benchmarks emphasize perception (e.g., recognition/description), cover limited disaster types, and provide insufficient support for the multi-stage reasoning required in practical emergency response. We introduce DisasterBench, a multi-stage multimodal reasoning benchmark for UAV-Based disaster response in complex environments. DisasterBench spans 14 disaster-related scene types and 9 response-critical tasks across pre-, during-, and post-disaster stages, with fine-grained disaster-task mappings that explicitly test causal attribution, propagation prediction, damage analysis, and decision-oriented reasoning. To enable reasoning on the edge, we further propose DisasterVL, a lightweight multimodal model optimized with a three-stage pipeline combining domain instruction tuning, chain-of-thought-guided multimodal alignment, and reinforcement learning-based policy optimization. Experiments across 21 popular MLLMs show that our 2B-parameter DisasterVL outperforms all evaluated open-source models and substantially narrows the gap to state-of-the-art closed-source models, achieving GPT-4o-comparable reasoning accuracy with superior efficiency. The project page is available at https://github.com/TanmouTT/DisasterBench.

Tan Zhang, Quanyou Li, Lu Zhang +3

School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, 611731, China · School of Information and Communication Engineering, Dalian University of Technology, Dalian, 116024, China · School of Computing and Communications, Lancaster University, Lancaster, LA1 4YW, England +1

Efficient Multimodal Inference Multimodal Reasoning Disaster Response Multimodal Model Evaluation

Jun 4, 2026cs.MM

Beyond Generative Decoding: Discriminative Hidden-State Readout from a Native Omni-Modal LLM for Multimodal Sentiment Analysis

Multimodal sentiment analysis (MSA) infers human affect from language, acoustic, and visual signals. Recent methods increasingly adapt large multimodal models (LMMs) via generative readout: prompting the model to emit a sentiment score as a text string. While convenient, this ties continuous regression to discrete autoregressive decoding, incurring unmeasured costs. We revisit this readout mechanism and propose a discriminative formulation built on the Thinker module of a native omni-modal LLM (Qwen2.5-Omni-7B). Instead of text decoding, we map the final-layer hidden state of the last non-padding token to a continuous score via a lightweight regression head in a single forward pass. Using 4-bit quantization and low-rank adaptation (QLoRA), the entire 7B pipeline -- including video and audio processing -- trains on a single consumer GPU (RTX 5090, 32 GB) with 10-21 GB peak memory and 1.14% trainable parameters. Through a controlled comparison fixing the backbone, data, and LoRA configuration, we isolate the impact of the readout. On CMU-MOSI and CMU-MOSEI, our discriminative readout reaches state-of-the-art accuracy without task-specific feature engineering (MOSI: MAE 0.551, Corr 0.888; MOSEI: MAE 0.506, Corr 0.790) and exhibits strong multi-seed stability. In contrast, the generative readout -- even after equivalent supervised training -- more than doubles the mean absolute error, yields unparsable or out-of-range outputs (2.8% zero-shot), and suffers from higher latency. Modality ablations reveal a text-dominant regime on CMU-MOSI. Our findings indicate that how an LMM is read out is as consequential as how it is trained, demonstrating that a discriminative readout offers a more accurate, efficient, and reliable alternative for continuous MSA.

Bin Wen, Tien-Ping Tan

School of Computer Sciences, Universiti Sains Malaysia, Penang, Malaysia.

Efficient Multimodal Inference Sentiment Analysis Multimodal Large Language Models Multimodal Sentiment Analysis Multimodal Foundation Models Multimodal Understanding

Jun 3, 2026cs.CV

Hyper-ICL: Attention Calibration with Hyperbolic Anchor Distillation for Multimodal In-Context Learning

Multimodal In-Context Learning (ICL) has emerged as a practical inference paradigm for Multimodal Large Language Models, where a small set of interleaved image-text In-Context Demonstrations (ICDs) conditions the model to solve new tasks. Despite its flexibility, multimodal ICL incurs high inference latency and suffers from instability due to sensitivity to demonstration formatting, ordering, and content. To address these limitations, we propose Hyper-ICL, a lightweight, training-based framework for demonstration-free multimodal ICL that reconstructs demonstration effects directly without requiring ICDs at inference time. Hyper-ICL learns a parameter-efficient low-rank logit-level adapter that calibrates attention distributions to better match demonstration-induced attention redistribution. To capture how demonstration influence varies across queries, we introduce a query-adaptive modulation mechanism that adaptively controls intervention strength at token level across layers and heads based on the current query. Finally, we propose a layer-wise hyperbolic anchor distillation loss that aligns intermediate student features to a demonstration-conditioned teacher via Lorentz geodesic distance. This loss encourages the student to reconstruct the demonstration-query relationships induced by ICDs. Extensive experiments across six different multimodal benchmarks (including VQAv2, OK-VQA, and COCO Caption) demonstrate that Hyper-ICL consistently improves accuracy and stability over vanilla ICL and existing state-of-the-art methods.

Niloufar Alipour Talemi, Hossein Kashiani, Fatemeh Afghah

Holcombe Department of Electrical and Computer Engineering, Clemson University, Clemson, United States.

Efficient Multimodal Inference Multimodal ICL Multimodal Large Language Models In-Context Learning Attention Mechanisms Multimodal Foundation Models+2

Jun 3, 2026cs.LG

Differentiable Efficient Operator Search

Efficient multimodal foundation models often rely on manually designed token-reduction operators, such as pruning, merging, pooling, and adaptive reweighting. Although these operators appear different, we show that they can be interpreted as distinct regimes of a shared operator space. Based on this view, we introduce Efficient Operator Search, a differentiable framework that jointly searches where to reduce tokens, how many tokens to retain, and how reduced token information should be processed. The proposed search space parameterizes layer activation, retention budget, and operator behavior, while the search policy optimizes task performance under one-sided budget and cost constraints. This formulation recovers representative hand-designed baselines as special cases and further discovers hybrid operators beyond isolated manual designs. Experiments on multimodal benchmarks show that the searched operators achieve competitive accuracy-efficiency trade-offs, especially under aggressive visual-token reduction. These results suggest that efficient multimodal inference can be reframed from manual operator design to differentiable operator search.

Xiaohuan Pei, Jiyuan Zhang, Yuanfan Guo +4

1The University of Sydney · 2ByteDance · 3Shanghai Jiao Tong University +1

Efficient Multimodal Inference Multimodal Token Compression Multimodal Foundation Models Multimodal Learning

Jun 2, 2026cs.AI

Perceive Before Reasoning: A Pre-Reasoning Perception Framework for Efficient and Reliable Proactive Mobile Agents

Multimodal large language models (MLLMs) have substantially advanced mobile agents, yet proactive mobile assistance remains challenging because agents must decide when to intervene before determining how to assist. Existing systems often implement these two decisions within a unified MLLM-based pipeline, leading to goal misalignment between conservative intervention filtering and comprehensive assistance generation, as well as redundant inference when the agent should remain silent. To address these limitations, we propose the Pre-Reasoning Perception Framework (PRPF), a two-stage framework built on perceiving before reasoning. PRPF introduces a lightweight Multimodal Proactive Perceptor (MPP) for intervention gating and context compression, and activates the Proactive Agent Reasoner (PAR) only when intervention is warranted. Experiments on the ProactiveMobile benchmark show that PRPF substantially reduces false trigger rates (FTR) while improving success rates (SR) and inference efficiency over the ProactiveMobile baseline.

Zhijie Ding, Weinan Hong, Zicheng Zhu +6

HyperAI Team, Xiaomi Corporation · Zhongnan University of Economics and Law · 1HyperAI Team, Xiaomi Corporation +5

Efficient Multimodal Inference Proactive Assistance Multimodal Agents LLM Agent Reliability

Jun 1, 2026cs.LG

Spectral-Progressive Thought Flow for Lightweight Multimodal Reasoning

Multimodal spatial reasoning often relies on long chains of intermediate textual and visual thoughts, where accumulating visual tokens and dense cross-modal attention incur substantial computation and memory overhead. To address this challenge, we propose Spectral-Progressive Thought Flow (SpecFlow), a novel lightweight multimodal spatial reasoning framework that represents intermediate visual thoughts in a fixed-size discrete cosine space. By exploiting strong energy compaction, SpecFlow preserves global layout and relational structure while introducing high-frequency details only when increased spatial precision is required. To align visual state evolution with linguistic intent, classifier-free guidance enables autoregressive textual thoughts to steer flow-based updates of the visual workspace/state without expanding the context. As a result, SpecFlow maintains a bounded visual workspace whose updates depend only on the current visual state and accumulated textual trace, enabling long-horizon inference with stable latency and memory usage independent of reasoning depth. Empirical results show that SpecFlow achieves competitive or superior reasoning performance while reducing computation and KV cache costs by up to 2.1 times.

Yixian Shen, Zhiheng Yang, Qi Bi +6

Informatics Institute, University of Amsterdam, Amsterdam, The Netherlands · University of Thessaly, Volos, Greece · Department of Computer Science, University College London +3

Efficient Multimodal Inference Visual Spatial Reasoning Multimodal Reasoning Latent Visual Reasoning Multimodal CoT Reasoning Visually Grounded Reasoning+1

Jun 1, 2026cs.CV

Improving Visual Token Reduction via Rectifying Distortions for Efficient Multimodal LLM Inference

Recent advancements in Multimodal Large Language Models (MLLMs) have achieved remarkable success in vision-language tasks, yet the quadratic computational complexity arising from the vast number of visual tokens incurs significant memory and latency bottlenecks. While visual token reduction (VTR) strategies have been explored to mitigate this burden, existing methods overlook the positional and attentional consistency between the full and reduced sequences, resulting in a distorted representation. To this end, we propose RESTORE, a novel VTR framework that rectifies the positional and attentional distortions while maintaining efficiency. Specifically, we present a simple yet effective calibration method that restores lost visual attention by augmenting attention weights based on relative distances. We also introduce a distinctive anchor selection for token merging to mitigate information loss during feature averaging. Experimental results on multiple benchmarks demonstrate that our method consistently improves the accuracy of various reduction methods, achieving state-of-the-art performance while maintaining computational efficiency. Project page is available at https://cvlab.yonsei.ac.kr/projects/RESTORE

Hyeonwoo Cho, Donghyeon Baek, Yewon Kim +1

Yonsei University · Korea Institute of Science and Technology (KIST)

Efficient Multimodal Inference Multimodal Large Language Models Visual Token Merging Visual Token Compression

May 29, 2026cs.CV

VisionPulse: Dynamic Visual Sparsity for Efficient Multimodal Reasoning

With the rapid advancement of large multimodal models (LMMs), inference-time overhead has become a key bottleneck for real-world deployment. Existing methods typically prune visual tokens at prefill, assuming the required visual evidence remains static during reasoning. However, we empirically show that visual evidence is strongly step-dependent: only a sparse subset of visual tokens is critical at each decoding step, and the critical set evolves across reasoning. Furthermore, we identify a coupled bottleneck where redundant visual context can steer the model toward query-irrelevant regions, lengthening the reasoning trace. Guided by these insights, we propose VisionPulse, a step-wise visual token pruning framework during reasoning. VisionPulse computes a lightweight visual attention mass to estimate the step-wise retention budget by exploiting its strong positive correlation with LMMs' effective visual token usage and retain only the most critical tokens under this budget. By enforcing visual sparsity during reasoning, VisionPulse filters redundant visual context while preserving relevant visual evidence, shortening reasoning traces naturally. Extensive experiments show that VisionPulse only retains 5% of visual tokens per step with reasoning traces shortened by 11.2%, while keeping accuracy almost unchanged.

Hengbo Xu, Shengjie Jin, Yanbiao Ma +1

Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China.

Efficient Multimodal Inference Visual Attention Multimodal Reasoning VLM Reasoning Multimodal Foundation Models Visual Token Pruning+1

May 28, 2026cs.RO

MARS Policy: Multimodality Only When It Matters

Imitation learning has become a cornerstone for solving complex robotic manipulation tasks. In particular, multimodality, which enables robots to capture diverse yet valid behavioral patterns, has driven the rapid emergence of generative policies as a dominant paradigm in robot learning. However, achieving such multimodality typically relies on stochastic noise initialization and iterative denoising procedures, resulting in substantial training complexity and low inference efficiency. Meanwhile, not all phases of a robotic task inherently require behavioral diversity. Motivated by this insight, we propose the Modality-Adaptive Robot Sampling (MARS) policy, which adaptively invokes tailored stochasticity only when it is truly beneficial, while reverting to an efficient deterministic learning during single-modal phases. In other words, the proper amount of noise is injected only at the proper time. By selectively activating multimodal generation, MARS policy bridges the gap between the multimodal capability of generative policies and the superior training and inference efficiency of deterministic models. Empirical studies across 8 simulated and 4 real-world tasks demonstrate that MARS exhibits robust multimodal expressivity and high efficiency, with a 16.67% success rate improvement and an 83.20% inference latency reduction in real-world tests. Counterintuitively, MARS also outpaces deterministic policies in training efficiency on near-deterministic tasks by more effectively modeling nuanced action diversity.

Jindou Jia, Tuo An, Yuxuan Hu +7

1MARS Lab, Nanyang Technological University

Efficient Multimodal Inference Robot Policy Learning Robot Imitation Learning

May 27, 2026cs.AI

DREAM-R: Multimodal Speculative Reasoning with RL-Based Refined Drafting, Precise Verification, and Fully Parallel Execution

Speculative reasoning has recently been proposed as a means to accelerate reasoning-intensive generation in large multimodal models, but its effectiveness is often constrained by misalignment between speculative drafts and target-verified reasoning. In this work, we introduce DREAM-R, a framework that substantially improves the performance of speculative reasoning. At its core, DREAM-R employs Speculative Alignment Policy Optimization (SAPO), a reinforcement-learning objective that trains draft models to generate reasoning steps that are both faithful to target trajectories and concise. We further propose a Threshold-based Verification Mechanism (TBVM) that uses a ratio-based criterion to provide stable and interpretable acceptance of speculative steps only when positive evidence clearly dominates, thereby preventing error propagation. Building on these components, we develop a Fully Parallel Speculative Reasoning (FPSR) framework that parallelizes draft generation, target-side reasoning, and verification across multi-step reasoning, enabling early stopping and clean fallback. Experiments on reasoning-heavy benchmarks demonstrate up to speedup while preserving target-model accuracy, yielding substantial efficiency gains without compromising reasoning quality.

Yunhai Hu, Zining Liu, Xiangyang Yin +5

New York University · University of Pennsylvania · Cerebras Systems

Efficient Multimodal Inference RL for Language Model Reasoning Multimodal Reasoning

May 27, 2026cs.RO

ProgVLA: Progress-Aware Robot Manipulation Skill Learning

We present ProgVLA, a compact vision-language-action (VLA) model designed for reliable robot manipulation under tight compute and memory budgets. The model specifically focuses on efficiently processing long multi-modal sequences by maintaining an explicit representation of task progress over extended horizons. To this end, ProgVLA integrates two key components. First, a multi-modal encoder with a two-stage Perceiver resampling scheme compresses variable-length visual, language, and proprioceptive streams into a fixed set of control-ready context tokens, substantially reducing sequence length while preserving cross-modal grounding. Second, an auxiliary set of progress heads is trained with offline reinforcement learning (RL) objectives to jointly learn critics over normalized remaining-horizon targets. This provides the policy with an internal estimate of task progress and enables advantage- and success-weighted flow-matching imitation learning. On two well-established multi-task robot manipulation benchmarks, a 0.1B-parameter ProgVLA model reaches success rates that are competitive with, and on long-horizon and harder task tiers exceed, substantially larger pretrained baselines. Ablations indicate that the learned context resampler and task-adaptive visual fine-tuning are the largest single contributors, while progress-aware training provides a consistent additional gain that is concentrated on long-horizon and multi-object tasks. We further validate the approach in real-world toy-kitchen environments.

Seungsu Kim, Jinyoung Choi, Seungmin Baek +1

NAVER LABS Europe · NAVER LABS

Cross-Modal Learning Efficient Multimodal Inference Long-Horizon Robotic Manipulation Robot Skill Learning Robotic Manipulation Vision-Language-Action Models+1

May 27, 2026cs.CV

ROVER: Routing Object-Centric Visual Evidence for Grounded Multi-Image Reasoning

Multimodal Large Language Models (MLLMs) have increasingly localized and interleaved visual evidence for deliberative reasoning. Grounding-based approaches typically focus on regions of interest (RoIs) by injecting cropped image patches or RoI-specific features into the reasoning context. However, such designs can weaken holistic scene understanding and inter-object relations, while incurring decoding costs that scale with the number and size of RoIs. Alternatively, adaptive visual feature selection often requires fine-grained supervision or complex heuristics. To address these limitations, we propose ROVER (Routing Object-centric Visual Evidence for grounded multi-image Reasoning), a lightweight, learnable plugin for efficient global visual evidence routing. Upon each object grounding prediction, ROVER injects a step-specific token triplet to synergistically: (i) aggregate the ongoing reasoning context, (ii) distill intra-image cues into a visual working space via object-centric differential attention, and (iii) route and integrate history-aware evidence across objects and images within this space for subsequent reasoning. We integrate ROVER into Qwen2.5-VL-7B and develop an interleaved SFT-to-GRPO training pipeline. Strictly adhering to the original datasets and evaluation protocols, our method achieves the best performance on MM-GCoT (+4.8% answer accuracy, +14.6% grounding accuracy) and VideoEspresso (+8.6% answer accuracy). The VideoEspresso-trained model demonstrates strong transferability, outperforming the base model by +4.7% on average across diverse benchmarks.

Guannan Lv, Ren Nie, Hongjian Dou +1

Kuaishou Technology

Multimodal Grounding Efficient Multimodal Inference Visual Reasoning VLM Reasoning Visually Grounded Reasoning