BEAR-Bench is a bilingual benchmark for multimodal large language models, featuring 1,000 human‑annotated questions derived from text‑rich business and scientific documents in English and Russian. It evaluates 16 MLLMs, including Gemini 3.1 Pro and Qwen3.5‑397B, revealing significant performance gaps even for the strongest systems. The benchmark also serves to compare hallucination‑detection methods by analyzing model failures on these complex documents.
By Liubov Chubarova, Alexandra Kuleshova, Daniil Volkov, Kirill Sultanov, Alexey Zaytsev
The study investigates how software developers, architects, and AI practitioners select and integrate Large Language Models (LLMs) into modern software systems. Interviews with 22 professionals reveal that functional criteria—such as performance, accuracy, cost, and specific features—dominate model choice, while security concerns are rarely considered. The research highlights a pervasive neglect of established software supply‑chain security lessons, leading to vulnerabilities like malicious components, data leakage, and unintended behavior, and offers actionable recommendations for a proactive, security‑by‑design approach.
By Mahzabin Tamanna, Elizabeth Lin, Sparsha Gowda, Laurie Williams, Dominik Wermke
The paper introduces Fair-ASR, a protocol for evaluating black‑box jailbreak attacks using a shared target‑call budget, addressing the bias of prior studies that rely solely on attack success rate. Re‑evaluating 11 attacks under this protocol shows significant shifts in rankings and highlights that many methods are not efficient in both target and attacker calls. The authors then present ReCode, a budget‑efficient attack that combines desensitization rewriting with low‑cost primitives, achieving 85% ASR on GPT‑5 with only 7.19 attacker calls per request under a 20‑target‑call budget.
By Zhida He, Xiaoyu Wen, Han Qi, Ziyuan Zhou, Peng Yu, Jiajia Li, Chaochao Lu, Qiaosheng Zhang
LoopVLA introduces a recurrent Vision‑Language‑Action architecture that learns to refine multimodal representations, predict actions, and estimate when further refinement is unnecessary. By iteratively applying a shared Transformer block and producing a sufficiency score at each step, it decouples refinement from fixed layer indices and aligns confidence scores with action quality through a self‑supervised objective. Experiments on LIBERO, LIBERO‑Plus, and VLA‑Arena demonstrate that LoopVLA reduces model parameters by 45% and boosts inference throughput up to 1.7× while matching or surpassing strong baselines in task success.
By Boyang Shen, Kaixiang Yang, Hao Wang, Qiuyu Yu, Qiang Xie, Qiang Li, Zhiwei Wang
The paper introduces OVRSISBench, a unified benchmark for open‑vocabulary remote sensing image segmentation, and evaluates existing OVS/OVRSIS models, uncovering their shortcomings in remote sensing contexts. Leveraging insights from this evaluation, the authors propose RSKT‑Seg, a new framework featuring a Multi‑Directional Cost Map Aggregation module, an Efficient Cost Map Fusion transformer, and a Remote Sensing Knowledge Transfer module. Experiments on the benchmark demonstrate that RSKT‑Seg outperforms strong baselines by +3.8 mIoU and +5.9 mACC while achieving twice the inference speed.
By Bingyu Li, Haocheng Dong, Da Zhang, Zhiyuan Zhao, Junyu Gao, Xuelong Li
DiSCO is a zero‑shot, black‑box defense for text‑to‑image models that operates solely at the prompt level. It expands prompts with a distribution‑guided suffix using beam search and contrastive scoring against safe and unsafe image pools generated by the target model, iteratively refining until safe content is produced. The method improves safety on the I2P benchmark under various red‑teaming attacks, reducing attack success rates by 37.7% and 25.13% while preserving semantic fidelity and image coherence.
By Tong Zhang, Motasem Alfarra, Carlos Hinojosa, Christos Louizos, Bernard Ghanem
The paper introduces ADAPT, an adversarial disentangled prompt tuning framework designed to improve the robustness of vision‑language models. ADAPT employs a dual‑prompt strategy: a target prompt learns robust features while a set of decoy prompts capture pseudo‑robust, non‑generalizable shortcuts. By enforcing orthogonality between target and decoy prompts, the method mitigates robust generalization overfitting and provides a theoretical error bound for unseen classes, leading to significant empirical robustness gains.
By Yang Chen, Zhan Zhuang, Yanbin Wei, Zebin Chen, Hua Liu, Yu Zhang
Prism‑GRPO enhances the GRPO reinforcement‑learning algorithm for vision‑language‑action policies by adding a weighted trajectory‑level execution‑quality score to binary success rewards. This approach splits groups with identical outcomes into a quality spectrum, preserving training signal while ensuring successes always outrank failures. Experiments on four RoboTwin tasks show Prism‑GRPO achieves higher success and quality at matched rollout budgets, reaching target success rates with up to 56% fewer rollouts and mitigating reward‑hacking behaviors that transfer to real‑robot deployment.
By Zeyun Deng, Yuzhe Lu, Yawei Wang, Linbo Liu, Qing Ping, Han Ding, Guande Wu, Panpan Xu, Jun Huan
The 10th AI City Challenge, held alongside ECCV 2026, celebrates a decade of benchmarking for intelligent transportation, smart cities, and physical AI. Since its 2017 inception focused on vehicle detection, classification, and tracking, the challenge has expanded into a comprehensive benchmark suite covering multi‑camera perception, multimodal reasoning, synthetic‑to‑real learning, generative forecasting, and privacy‑preserving evaluation. The 2026 edition saw 325 registered teams from 26 countries, with six main tracks—spanning multi‑camera 3D perception, transportation safety captioning and VQA, traffic anomaly reasoning, text‑based person anomaly search, generative traffic video forecasting, and cross‑city object detection—plus two out‑of‑domain leaderboards for fisheye traffic‑violation understanding and pedestrian situated‑intent VQA.
By Zheng Tang, Shuo Wang, David C. Anastasiu, Ming-Ching Chang, Anuj Sharma, Quan Kong, Munkhjargal Gochoo, Jun-Wei Hsieh, Tomasz Kornuta, Zhedong Zheng, Renran Tian, Judah Goldfeder, Fulgencio Navarro, Yuxing Wang, Yizhou Wang, Sameer Satish Pusegaonkar, Anqi Li, Nalin Dadhich, Ridham Kachhadiya, Dhanishtha Patil, Haoquan Liang, Jiajun Li, Han Zhang, Yilin Zhao, Zaid Pervaiz Bhat, Shuyu Yang, Ashutosh Kumar, Rong Wang, Rafael Martin Nieto, Peter Christiansen, Ahmed Abduljawad, Mohanrasu Shanmugam, Nadeem Shaik, Sujit Biswas, Xunlei Wu, Vidya Murali, Rama Chellappa
DeAR (Decentralized Agentic Reasoning) is a new framework that replaces centralized protocols with autonomous peer‑to‑peer collaboration for agentic reasoning. It introduces three key mechanisms: decentralized capability grounding for agent specialization, thought map navigation for targeted peer interactions, and topology updates for adaptive error correction. Across nine multimodal reasoning and text‑based QA benchmarks, DeAR consistently outperforms recent baseline methods, demonstrating that decentralized and adaptive collaboration improves accuracy in knowledge‑intensive reasoning tasks.
By Xing Wei, Changmeng Zheng, XiaoYong Wei, Xiufen Ye, Qing Li
VLCP (Vision Language Control Policy) is a training‑free robot manipulation approach that keeps a vision‑language model (VLM) frozen and uses it to generate short Python control functions. Unlike traditional methods that retry a fixed policy, VLCP rewrites the control code every K steps based on multi‑view RGB, proprioceptive state, and state delta, allowing failures to be corrected within the same episode. In a 57‑task MuJoCo/RoboVerse benchmark, VLCP achieves 35.1% pooled success versus 3.5% for a single‑query baseline, with a 27.3% within‑episode recovery rate on failed grasps and efficient token usage.
By Dhia Naouali, Minghan Wu, Claudia Wong, Abhinav Puthran, Omar G. Younis
The paper introduces MemJack, a memory‑augmented multi‑agent framework that automatically generates jailbreak attacks on Vision‑Language Models (VLMs) using benign natural images as visual anchors. MemJack discovers visual anchors, camouflages them semantically, evaluates responses, repairs via reflection, and replans dynamically, forming a closed‑loop attack pipeline. The authors also create MemJack‑Bench, a dataset of over 113,000 interactive multimodal jailbreak trajectories, and show that MemJack achieves a 71.48% attack success rate against Qwen3‑VL‑Plus, reaching 90% under extended budgets, outperforming other baselines on natural‑image evaluation.
By Jianhao Chen, Haoyang Chen, Hanjie Zhao, Haozhe Liang, Zheng Wang, Tieyun Qian
Co‑RL is a multi‑agent reinforcement learning framework that trains several decoupled models without shared parameters, using rewards generated by their peers. By increasing cohort diversity—through heterogeneous model families, varying sizes, and rephrased training samples—Co‑RL reduces self‑reinforcing feedback loops, preserves behavioral diversity, and prevents training collapse. Across both text‑only and multimodal benchmarks, Co‑RL outperforms base models and prior label‑free methods, achieving gains of 3.0‑8.6% on seven text benchmarks and 2.3‑7.2% on four multimodal benchmarks, while matching or surpassing supervised approaches without any ground‑truth labels.
By Yunhao Yang, Yuexin Bian, Yunjie Tian, Di Fu, Tianjin Huang, Yuanyuan Shi, Ziang Xiao, Nuno Vasconcelos, Yijiang Li
The paper introduces a novel framework that converts structural framing plans from PDF drawings into editable finite‑element model drafts without requiring task‑specific detector training. It combines a deterministic geometry extraction stage—estimating scale, recognizing five entity classes, and assembling a drafting grammar—with an agentic vision‑language refinement stage that proposes corrections, performs admission tests, and ensures fail‑closed transactions. Evaluation on a 100‑plan benchmark shows high accuracy, with scale within 0.1% and recall/precision values above 0.86 for all component types.
By Mohammad Talebi-Kalaleh, Qipei Mei
The paper introduces a capability‑centric data infrastructure for generalist image generation, integrating task‑specific supervision with a curriculum that aligns with the dependencies among generative capabilities. It employs three interoperable data engines—text‑image grounding, inter‑image transformation, and image‑knowledge association—alongside caption experts to harmonize text‑to‑image and editing supervision. The system curates massive corpora (440M T2I images, 120M editing pairs, 27M image‑entity pairs) and trains multimodal diffusion models (3B and 6B parameters) from scratch, achieving broad visual coverage and versatile rendering as shown by CPI‑Bench and qualitative tests.
By Xingjian Wang, Zhao Wang, Taihang Hu, Jun Zheng, Qing Jin, Qinye Zhou, Zhengtao Wu, Yongchao Du, Zuan Gao, Chao Lin, Yefeng Shen, Xiaoli Xu, Zhengze Xu, Hao Yan, Yuhang Yu, Mingzhou Zhang, Mengting Chen
M3TR is a temporal retrieval‑enhanced multi‑modal framework for predicting micro‑video popularity. It introduces a Mamba‑Hawkes Process module to model user feedback as self‑exciting events, capturing long‑range temporal dependencies. A temporal‑aware retrieval engine then identifies historically relevant videos by combining multi‑modal content similarity with popularity trajectory similarity, augmenting the target video’s features for improved prediction accuracy.
By Jiacheng Lu, Weijian Wang, Mingyuan Xiao, Yang Hua, Tao Song, Bo Peng, Cheng Hua, Haibing Guan
The paper introduces TiMi, a framework that enhances time series transformers with a Multimodal Mixture-of-Experts (MMoE) module to incorporate multimodal data, especially textual information, into forecasting. TiMi leverages large language models to generate future inferences that guide predictions, eliminating the need for explicit representation alignment. Experiments show TiMi achieves state‑of‑the‑art performance on sixteen real‑world multimodal forecasting benchmarks, outperforming advanced baselines while maintaining adaptability and interpretability.
By Jiafeng Lin, Yuxuan Wang, Huakun Luo, Jianmin Wang, Zhongyi Pei
The review discusses how two‑phase heat transfer—critical for boiling, condensation, and thermal management—poses challenges for data reuse due to its complex interfacial physics. It surveys open datasets, machine‑learning techniques, and reusable software, organizing them with a spatial‑plus‑temporal dimensionality taxonomy (S+TD) that links data types to AI tasks such as regression, sequence learning, and image/video analysis. The paper proposes a roadmap for physics‑aware open data, including metadata standards, maturity labels, benchmark splits, and community databanks, emphasizing that progress in two‑phase AI relies as much on robust data infrastructure as on model design.
By Christy Dunlap, Ridwan Olabiyi, Firas Al-Hindawi, Hari Pandey, Stephen Pierson, Daniel Curl, Braden Stevens, Mohammad Ishraq Hossain, Annapurna Parjuli, Chinmaya Joshi, Ashif Iquebal, Han Hu
The paper introduces Uni-SafeBench, a safety benchmark designed to evaluate the holistic safety of Unified Multimodal Large Models (UMLMs) across six safety categories and seven task types. It also presents Uni-Judger, a framework that separates contextual safety from intrinsic safety to enable rigorous assessment. Evaluations reveal that current unified models do not consistently maintain the safety alignment of their underlying language models, and open‑source UMLMs perform significantly worse on safety than specialized multimodal models, especially in generation tasks.
By Zixiang Peng, Yongxiu Xu, Qin-Yi Zhang, Jiexun Shen, Yi-Fan Zhang, Hongbo Xu, Yubin Wang, Gaopeng Gou
MultiSigBERT is a unified framework that performs multimodal sequential survival modeling in oncology by integrating narrative clinical reports, numerical measurements, and structured variables. The method converts free-text reports into sentence embeddings, compresses them with modality-specific PCA, and concatenates them with structured covariates to create joint temporal trajectories. These trajectories are encoded using the Signature transform from Rough Paths theory, and the resulting high-dimensional features are fed into a LASSO-regularized Cox model, achieving a concordance index of 0.743 on an independent test set of over 2,500 patients.
By Paul Minchella, St\'ephane Chr\'etien, Guillaume Metzler, Lo\"ic Verlingue, R\'emi Vaucher