Multimodal agents are making rapid progress on general computer-use tasks, yet existing benchmarks remain largely confined to browsersand basic desktop applications, falling short in professional software workflows that dominate real-world scientific and industrial practice. To close this gap, we introduce ProSoftArena, a benchmark and platform specifically for evaluating multimodal agents in professional software environments. We establish the first capability hierarchy tailored to agent use of professional software and construct a benchmark of 437 realistic work and research tasks spanning 6 disciplines and 13 core professional applications. To ensure reliable and reproducible assessment, we build an executable real-computer environment with an execution-based evaluation framework and uniquely incorporate a human-in-the-loop evaluation paradigm. Extensive experiments show that even the best-performing agent attains only a 24.4\% success rate on L2 tasks and completely fails on L3 multi-software workflow. In-depth analysis further provides valuable insights for addressing current agent limitations and more effective design principles, paving the way to build more capable agents in professional software settings.We will release all data and codes to forster research in this critical area.

Diffusion models have emerged as state-of-the-art generative models, capable of producing high-quality synthetic outputs. These generated contents serve two key purposes in practice: as end-user facing products (“Generated Content as Product”, GCAP) and for data augmentation in machine learning pipelines (“Generated Content as Data”, GCAD). The expedited deployment of diffusion models has heightened awareness of their potential vulnerabilities, among which backdoor attacks are identified as a prominent concern. However, existing backdoor attack methods primarily focus on the conventional GCAP scenario and are constrained by their reliance on perceptible, pixel-based triggers. The lack of exploration on imperceptible attacks and the neglect of GCAD scenario may exacerbate security risks and broaden the reach of adversarial impacts. To unveil hidden attack surfaces, we address the noted challenges and propose Evil Diffusion (ELF), a novel framework for launching stealthy backdoor attacks on diffusion models, including two algorithms tailored for the GCAP and GCAD scenarios, respectively. First, we introduce ELF-P, a two-stage training algorithm that autonomously learns a frequency-based stealthy trigger for the GCAP scenario. ELF-P employs perceptual and defense-resistant constraints in the frequency domain to craft a human-imperceptible trigger that is resilient to common steganography defenses, enabling highly covert backdooring of diffusion models. Second, we present ELF-D, an algorithm designed for the GCAD scenario that poisons a conditional diffusion model to compromise specific downstream tasks. ELF-D introduces perturbations into task-critical frequency components of generated images, allowing compromised diffusion models to subtly undermine the performance of downstream models trained on their outputs. Extensive experiments validate the effectiveness of the proposed ELF framework in both scenarios, unveiling the hidden vulnerabilities in diffusion models and emphasize the broader, potentially harmful implications of these attacks.

Multimodal large language models (MLLMs), which integrate language and visual cues for problem-solving, are crucial for advancing artificial general intelligence (AGI). However, current benchmarks for measuring the intelligence of MLLMs suffer from limited scale, narrow coverage, and unstructured knowledge, offering only static and undifferentiated evaluations. To bridge this gap, we introduce MDK12-Bench, a large-scale multidisciplinary benchmark built from real-world K–12 exams spanning six disciplines with 141K instances and 6,225 knowledge points organized in a six-layer taxonomy. Covering five question formats with difficulty and year annotations, it enables comprehensive evaluation to capture the extent to which MLLMs perform over four dimensions: 1) difficulty levels, 2) temporal (cross-year) shifts, 3) contextual shifts, and 4) knowledge-driven reasoning. We propose a novel dynamic evaluation framework that introduces unfamiliar visual, textual, and question form shifts to challenge model generalization while improving benchmark objectivity and longevity by mitigating data contamination. We further evaluate knowledge-point reference-augmented generation (KP-RAG) to examine the role of knowledge in reasoning. Key findings reveal limitations in current MLLMs in multiple aspects and provide guidance for enhancing model reasoning, robustness, and AI-assisted education.

Existing image forgery detection (IFD) methods either exploit low-level, semantics-agnostic artifacts or rely on multimodal large language models (MLLMs) with high-level semantic knowledge. Although naturally complementary, these two information streams are highly heterogeneous in both paradigm and reasoning, making it difficult for existing methods to unify them or effectively model their cross-level interactions. To address this gap, we propose ForenAgent, a multi-round interactive IFD framework that enables MLLMs to autonomously generate, execute, and iteratively refine Python-based low-level tools around the detection objective, thereby achieving more flexible and interpretable forgery analysis. ForenAgent follows a two-stage training pipeline combining Cold Start and Reinforcement Fine-Tuning to enhance its tool interaction capability and reasoning adaptability progressively. Inspired by human reasoning, we design a dynamic reasoning loop comprising global perception, local focusing, iterative probing, and holistic adjudication, and instantiate it as both a data-sampling strategy and a task-aligned process reward. For systematic training and evaluation, we construct FABench, a heterogeneous, high-quality agent-forensics dataset comprising 100k images and approximately 200k agent-interaction question-answer pairs. Experiments show that ForenAgent exhibits emergent tool-use competence and reflective reasoning on challenging IFD tasks when assisted by low-level tools, charting a promising route toward general-purpose IFD. The code will be released after the review process is completed.

Training open-ended agents via reinforcement learning (RL) is hindered by the lack of verifiable gold answers and scalable rubrics. Moreover, even near the model's capability boundary, long-horizon open-ended agentic tasks often yield brittle and unstable rewards, resulting in weak or noisy rollout contrast that obscures fine-grained optimization signals for group-based policy learning. To address these challenges, we propose ARISE-RL, a novel full-cycle self-evolution framework that couples a task/rubric Generator and a reasoning Solver through rubric-mediated co-evolution. The Generator grounds tool-related rubric criteria in real tool observations and is rewarded for producing valid, intermediate-difficulty tasks aligned with the Solver's evolving capability boundary. The Solver, in turn, learns from fine-grained rubric satisfaction signals through multi-step reasoning and tool use. We further introduce Reward-Gated Self-Evolution Distillation (RG-SED), which selectively distills a memory-augmented variant of the same policy back into itself only when the memory yields empirical reward improvement, thereby reducing distribution mismatch and avoiding blind imitation of noisy guidance. Finally, to support rigorous evaluation, we present ECR-Bench, an expert-calibrated rubric benchmark suite covering single-tool deep research and multi-tool travel planning. Extensive experiments demonstrate that ARISE-RL consistently achieves robust and stable overall state-of-the-art performance across all evaluated benchmarks.

Chest X-ray (CXR) imaging is one of the most widely used diagnostic modalities in clinical practice, encompassing a broad spectrum of diagnostic tasks. Recent advancements have seen the extensive application of reasoning-based multimodal large language models (MLLMs) in medical imaging to enhance diagnostic efficiency and interpretability. However, existing multimodal models predominantly rely on ”one-time” diagnostic approaches, lacking verifiable supervision of the reasoning process. This leads to challenges in multi-task CXR diagnosis, including lengthy reasoning, sparse rewards, and frequent hallucinations. To address these issues, we propose CX-Mind, the first generative model to achieve interleaved ”think-answer” reasoning for CXR tasks, driven by curriculum-based reinforcement learning and verifiable process rewards (CuRL-VPR). Specifically, we constructed an instruction-tuning dataset, CX-Set, comprising 708,473 images and 2,619,148 samples, and generated 42,828 high-quality interleaved reasoning data points supervised by clinical reports. Optimization was conducted in two stages under the Group Relative Policy Optimization framework: initially stabilizing basic reasoning with closed-domain tasks, followed by transfer to open-domain diagnostics, incorporating rule-based conditional process rewards to bypass the need for pretrained reward models. Extensive experimental results demonstrate that CX-Mind significantly outperforms existing medical and general-domain MLLMs in visual understanding, text generation, and spatiotemporal alignment, achieving an average performance improvement of 25.1% over comparable CXR-specific models. On a real-world clinical dataset (Rui-CXR), CX-Mind achieves a mean recall@1 across 14 diseases that substantially surpasses the second-best results, with multi-center expert evaluations further confirming its clinical utility across multiple dimensions. CX-Mind establishes a new paradigm for constructing interpretable, and high-performing medical MLLMs.

Recent advancements in Large Multimodal Models (LMMs) have significantly improved multimodal understanding and generation. However, these models still struggle to generate tightly interleaved image-text outputs, primarily due to the limited scale, quality and instructional richness of current training datasets. To address this, we introduce \textbf{InterSyn}, a dataset that features: (1) large scale, comprising 1.8M multimodal samples; (2) high quality, supported by our proposed \textbf{Self-Evaluation with Iterative Refinement (SEIR)} method for rigorous automated quality refinement; (3) rich instructional diversity, ensured through diverse well-designed question templates, based on human preferences and covering a 3500-topic hierarchy. These characteristics make InterSyn particularly well-suited for training LMMs in interactive image–text generation capabilities. To evaluate the capabilities, we propose \textbf{SynJudge}, a reliable automatic evaluator that aligns closely with human judge and outputs four interpretable scores: Text Content Completeness (TCC), Image Content Completeness (ICC), Image Quality (IQ), and Image–Text Synergy (ITS). These scores are complementary, covering both content and quality as well as cross-modal interaction, thereby forming a comprehensive evaluation framework. Experimental results on InterSyn subsets of up to 200K samples show that 25K–50K already yield substantial improvements, while scaling to 100K/200K brings further gains in TCC, ICC, and especially ITS, highlighting InterSyn’s: (1) scalability, as performance consistently improves with more data; (2) efficiency, as significant gains are achievable even with smaller subsets, making it accessible to researchers with varying computational resources.

Reinforcement learning has substantially improved the performance of LLM agents on tasks with verifiable outcomes, but it still struggles on open-ended agent tasks with vast solution spaces (e.g., complex travel planning). Due to the absence of objective ground-truth for these tasks, current RL algorithms largely rely on reward models that assign scalar scores to individual responses. We contend that such pointwise scoring suffers from an inherent discrimination collapse: the reward model struggles to distinguish subtle advantages among different trajectories, resulting in scores within a group being compressed into a narrow range. Consequently, the effective reward signal becomes dominated by noise from the reward model, leading to optimization stagnation. To address this, we propose ArenaRL, a reinforcement learning paradigm that shifts from pointwise scalar scoring to intra-group relative ranking. ArenaRL introduces a process-aware pairwise evaluation mechanism, employing multi-level rubrics to assign fine-grained relative scores to trajectories. Additionally, we construct an intra-group adversarial arena and devise a tournament-based ranking scheme to obtain stable advantage signals. Empirical results confirm that the built seeded single-elimination scheme achieves nearly equivalent advantage estimation accuracy to full pairwise comparisons with O(N^2) complexity, while operating with only O(N) complexity, striking an optimal balance between efficiency and precision. Furthermore, to address the lack of full-cycle benchmarks for open-ended agents, we build Open-Travel and Open-DeepResearch, two high-quality benchmarks featuring a comprehensive pipeline covering SFT, RL training, and multi-dimensional evaluation. Extensive experiments show that ArenaRL substantially outperforms standard RL baselines, enabling LLM agents to generate more robust solutions for complex real-world tasks.

Generative world models for autonomous driving (AD) have become a trending topic. Unlike the widely studied image modality, in this work we explore generative world models for LiDAR data. Existing generation methods for LiDAR data only support single frame generation, while existing prediction approaches require multiple frames of historical input and can only deterministically predict multiple frames at once, lacking interactivity. Both paradigms fail to support long-horizon interactive generation. To this end, we introduce LaGen, which to the best of our knowledge is the first framework capable of frame-by-frame autoregressive generation of long-horizon LiDAR scenes. LaGen is able to take a single-frame LiDAR input as a starting point and effectively utilize bounding box information as conditions to generate high-fidelity 4D scene point clouds. In addition, we introduce a scene decoupling estimation module to enhance the model's interactive generation capability for object-level content, as well as a noise modulation module to mitigate error accumulation during long-horizon generation. We construct a protocol based on nuScenes for evaluating long-horizon LiDAR scene generation. Experimental results comprehensively demonstrate LaGen outperforms state-of-the-art LiDAR generation and prediction models, especially on the later frames.

Effectively retrieving, reasoning, and understanding multimodal information remains a critical challenge for agentic systems. Traditional Retrieval-augmented Generation (RAG) methods rely on linear interaction histories, which struggle to handle long-context tasks, especially those involving information-sparse yet token-heavy visual data in iterative reasoning scenarios. To bridge this gap, we introduce VimRAG, a framework tailored for multimodal Retrieval-augmented Reasoning across text, images, and videos. Inspired by our systematic study, we model the reasoning process as a dynamic directed acyclic graph that structures the agent states and retrieved multimodal evidence. Building upon this structured memory, we introduce a Graph-Modulated Visual Memory Encoding mechanism, with which the significance of memory nodes is evaluated via their topological position, allowing the model to dynamically allocate high-resolution tokens to pivotal evidence while compressing or discarding trivial clues. To implement this paradigm, we propose a Graph-Guided Policy Optimization strategy. This strategy disentangles step-wise validity from trajectory-level rewards by pruning memory nodes associated with redundant actions, thereby facilitating fine-grained credit assignment. Extensive experiments demonstrate that VimRAG consistently achieves state-of-the-art performance on diverse multimodal RAG benchmarks.

Test-time adaptation using vision- language models (such as CLIP) to quickly adjust to distributional shifts of downstream tasks has shown great potential. Despite significant progress, existing methods are still limited to single- task test- time adaptation scenarios and have not effectively explored the issue of multi- task adaptation. To address this practical problem, we propose a novel Hierarchical Knowledge Prompt Tuning (HKPT) method, which achieves joint adaptation to multiple target domains by mining more comprehensive source domain discriminative knowledge and hierarchically modeling task- specific and task- shared knowledge. Specifically, HKPT constructs a CLIP prompt distillation framework that utilizes the broader source domain knowledge of large teacher CLIP to guide prompt tuning for lightweight student CLIP from multiple views during testing. Meanwhile, HKPT establishes task- specific dual dynamic knowledge graph to capture fine- grained contextual knowledge from continuous test data. To fully exploit the complementarity among multiple target tasks, HKPT employs an adaptive task grouping strategy for achieving intertask knowledge sharing. Furthermore, HKPT can seamlessly transfer to basic single- task test- time adaptation scenarios while maintaining robust performance. Extensive experimental results in both multi- task and single- task testtime adaptation settings demonstrate that our HKPT significantly outperforms state- of- the- art methods.

The dynamic nature of real-world information demands efficient knowledge editing in multimodal large language models (MLLMs) to ensure continuous knowledge updates. However, existing methods often struggle with precise matching in large-scale knowledge retrieval and lack multi-level guidance for coordinated editing, leading to less reliable outcomes. To tackle these challenges, we propose CARML, a novel retrieval-augmented editing framework that integrates conflict-aware dynamic retrieval with multi-level implicit and explicit guidance for reliable lifelong multimodal editing. Specifically, CARML introduces intra-modal uncertainty and inter-modal conflict quantification to dynamically integrate multi-channel retrieval results, so as to pinpoint the most relevant knowledge to the incoming edit samples. Afterwards, an edit scope classifier discerns whether the edit sample semantically aligns with the edit scope of the retrieved knowledge. If deemed in-scope, CARML refines the retrieved knowledge into information-rich continuous prompt prefixes, serving as the implicit knowledge guide. These prefixes not only include static knowledge prompt that capture key textual semantics but also incorporate token-level, context-aware dynamic prompt to explore fine-grained cross-modal associations between the edit sample and retrieved knowledge. To further enhance reliability, CARML incorporates a "hard correction" mechanism, leveraging explicit label knowledge to adjust the model’s output logits. Extensive experiments across multiple MLLMs and datasets indicate the superior performance of CARML in lifelong multimodal editing scenarios.

Pre-trained vision-language models, such as CLIP, are driving advancements in person re-identification by mining semantic information. Current approaches utilize globally learnable textual prompts to generate coarse-level, holistic yet ambiguous descriptions of individuals, which are then served as constraints for the fine-tuning of CLIP to learn visual representations of the pedestrians. However, relying exclusively on this type of prompt learning for the text encoder of CLIP overlooks the crucial fine-grained details of individuals and fails to model the necessary downstream adaptation capacity for the image encoder of CLIP. To address these limitations, we propose a novel Multi-Granularity and Multi-Modal Prompt Learning (MMPL) for person re-identification to fully unleash the substantial potential inherent in CLIP for acquiring discriminative representations. The MMPL encompasses a two-stage training procedure. In the first training stage, MMPL meticulously orchestrates the Hierarchical Prompt Learning (HPL) to refine crucial and distinctive information from hierarchical patch-level visual features. Aligning textual prompts with these subtle visual cues across diverse granularities, this process establishes patch-to-token level correspondences, ultimately yielding the creation of high-fidelity multi-granularity textual prompts. In the second training stage, MMPL integrates Collaborative Prompt Learning (CPL), generating supplementary visual prompts and fostering multi-modal interactive learning to aid CLIP’s image encoder in narrowing the semantic gap between modalities, leveraging CLIP’s extensive multi-modal knowledge to enhance feature representation. Comprehensive experimental evaluation across four widely recognized person re-identification benchmarks substantiates the effectiveness of our MMPL.

LLMs have shown strong performance on human-centric reasoning tasks. While previous evaluations have explored whether LLMs can infer intentions or detect deception, they often overlook the individualized reasoning styles that influence how people interpret and act in social contexts. Social deduction games (SDGs) provide a natural testbed for evaluating individualized reasoning styles, where different players may adopt diverse but contextually valid reasoning strategies under identical conditions. To address this, we introduce InMind, a cognitively grounded evaluation framework designed to assess whether LLMs can capture and apply personalized reasoning styles in SDGs. InMind enhances structured gameplay data with round-level strategy traces and post-game reflections, collected under both Observer and Participant modes. It supports four cognitively motivated tasks that jointly evaluate both static alignment and dynamic adaptation. As a case study, we apply InMind to the game Avalon, evaluating 11 state-of-the-art LLMs. General-purpose LLMs, even GPT-4o frequently rely on lexical cues, struggling to anchor reflections in temporal gameplay or adapt to evolving strategies. In contrast, reasoning-enhanced LLMs like DeepSeek-R1 exhibit early signs of style-sensitive reasoning. These findings reveal key limitations in current LLMs’ capacity for individualized, adaptive reasoning, and position InMind as a step toward cognitively aligned human–AI interaction.

Identifying key nodes in social networks plays a crucial role in timely blocking false information. Existing key node identification methods usually consider node influence only from the propagation structure perspective and have insufficient generalization ability to unknown scenarios. In this paper, we propose a novel Multi-perspective Memory Enhanced Network (MMEN) for identifying key nodes in social networks, which mines key nodes from multiple perspectives and utilizes memory networks to store historical information. Specifically, MMEN first constructs two propagation networks from the perspectives of user attributes and propagation structure and updates node feature representations using graph attention networks. Meanwhile, the memory network is employed to store information of similar subgraphs, enhancing the model's generalization performance in unknown scenarios. Finally, MMEN applies adaptive weights to combine the node influence of the two propagation networks to select the ultimate key nodes. Extensive experiments demonstrate that our method significantly outperforms previous methods.

Understanding information cascades in networks is a fundamental issue in numerous applications. Current researches often sample cascade information into several independent paths or subgraphs to learn a simple cascade representation. However, these approaches fail to exploit the hierarchical semantic associations between different modalities, limiting their predictive performance. In this work, we propose a novel Hierarchical Information Enhancement Network (HIENet) for cascade prediction. Our approach integrates fundamental cascade sequence, user social graphs, and sub-cascade graph into a unified framework. Specifically, HIENet utilizes DeepWalk to sample cascades information into a series of sequences. It then gathers path information between users to extract the social relationships of propagators. Additionally, we employ a time-stamped graph convolutional network to aggregate sub-cascade graph information effectively. Ultimately, we introduce a Multi-modal Cascade Transformer to powerfully fuse these clues, providing a comprehensive understanding of cascading process. Extensive experiments have demonstrated the effectiveness of the proposed method.

The proliferation of fake news with image and text in the internet has triggered widespread concern. Existing research has made important contributions in cross-modal information interaction and fusion, but fails to fundamentally address the modality gap among news image, text, and news-related external knowledge representations. In this paper, we propose a novel Natural Language-centered Inference Network (NLIN) for multi-modal fake news detection by aligning multi-modal news content with the natural language space and introducing an encoder-decoder architecture to fully comprehend the news in-context. Specifically, we first unify multi-modal news content into textual modality by converting news images and news-related external knowledge into plain textual content. Then, we design a multi-modal feature reasoning module, which consists of a multi-modal encoder, a unified-modal context encoder and an inference decoder with prompt phrase. This framework not only fully extracts the latent representation of cross-modal news content, but also utilizes the prompt phrase to stimulate the powerful in-context learning ability of the pre-trained large language model to reason about the truthfulness of the news content. In addition, to support the research in the field of multi-modal fake news detection, we produce a challenging large scale, multi-platform, multi-domain multi-modal Chinese Fake News Detection (CFND) dataset. Extensive experiments show that our CFND dataset is challenging and the proposed NLIN outperforms state-of-the-art methods.

Recently, misinformation incorporating both texts and images has been disseminated more effectively than those containing text alone on social media, raising significant concerns for multi-modal fact-checking. Existing research makes contributions to multi-modal feature extraction and interaction, but fails to fully enhance the valuable semantic representations or excavate the intricate entity information. Besides, existing multi-modal fact-checking datasets are primarily focused on English and merely concentrate on a single type of misinformation, thereby neglecting a comprehensive summary and coverage of various types of misinformation. Taking these factors into account, we construct the first large-scale Chinese Multi-modal Fact-Checking (CMFC) dataset which encompasses 46,000 claims. The CMFC covers all types of misinformation for fact-checking and is divided into two sub-datasets, Collected Chinese Multi-modal Fact-Checking (CCMF) and Synthetic Chinese Multi-modal Fact-Checking (SCMF). To establish baseline performance, we propose a novel Entity-enhanced and Stance Checking Network (ESCNet), which includes Multi-modal Feature Extraction Module, Stance Transformer, and Entity-enhanced Encoder. The ESCNet jointly models stance semantic reasoning features and knowledge-enhanced entity pair features, in order to simultaneously learn effective semantic-level and knowledge-level claim representations. Our work offers the first step and establishes a benchmark for evidence-based, multi-type, multi-modal fact-checking.

Restoring clear frames from rainy videos poses a significant challenge due to the swift motion of rain streaks. Traditional frame-based visual sensors, which record dense scene content synchronously, struggle to capture the fast-moving rain information. Conversely, the novel bio-inspired event camera, known for its high temporal resolution and low latency, effectively records the motion trajectories of rapidly falling rain through asynchronously generated sparse event sequences. In light of these attributes, we introduce a novel event-driven convolutional spiking network for video deraining. For video restoration, we employ a Convolutional Neural Network as the backbone, extracting dense features from video sequences to map rain-soaked frames to clear ones. To remove rain, we meticulously design a bio-inspired Spiking Neural Network that adapts to the sparse event sequences, capturing features of falling rain. We then establish a bimodal feature fusion module that combines dense convolutional features with sparse spiking features. This fusion aids the backbone in accurately pinpointing rain streaks across spatiotemporal dimensions. Thus, our diverse network extracts and collaborates information from both events and videos, enhancing deraining performance. Experiments conducted on synthetic and real-world datasets prove that our network markedly surpasses existing video deraining techniques.