ViMU is the first benchmark for evaluating video models on metaphorical and subtextual understanding using hint-free questions grounded in multimodal evidence.
super hub Mixed citations
Qwen3-VL Technical Report
Mixed citation behavior. Most common role is background (47%).
840
Pith papers citing it
Background
47% of classified citations
abstract
We introduce Qwen3-VL, the most capable vision-language model in the Qwen series to date, achieving superior performance across a broad range of multimodal benchmarks. It natively supports interleaved contexts of up to 256K tokens, seamlessly integrating text, images, and video. The model family includes both dense (2B/4B/8B/32B) and mixture-of-experts (30B-A3B/235B-A22B) variants to accommodate diverse latency-quality trade-offs. Qwen3-VL delivers three core pillars: (i) markedly stronger pure-text understanding, surpassing comparable text-only backbones in several cases; (ii) robust long-context comprehension with a native 256K-token window for both text and interleaved multimodal inputs, enabling faithful retention, retrieval, and cross-referencing across long documents and videos; and (iii) advanced multimodal reasoning across single-image, multi-image, and video tasks, demonstrating leading performance on comprehensive evaluations such as MMMU and visual-math benchmarks (e.g., MathVista and MathVision). Architecturally, we introduce three key upgrades: (i) an enhanced interleaved-MRoPE for stronger spatial-temporal modeling across images and video; (ii) DeepStack integration, which effectively leverages multi-level ViT features to tighten vision-language alignment; and (iii) text-based time alignment for video, evolving from T-RoPE to explicit textual timestamp alignment for more precise temporal grounding. Under comparable token budgets and latency constraints, Qwen3-VL achieves superior performance in both dense and Mixture-of-Experts (MoE) architectures. We envision Qwen3-VL serving as a foundational engine for image-grounded reasoning, agentic decision-making, and multimodal code intelligence in real-world workflows.
hub tools
citation-role summary
background 121
method 61
baseline 50
dataset 5
other 4
citation-polarity summary
claims ledger
- abstract We introduce Qwen3-VL, the most capable vision-language model in the Qwen series to date, achieving superior performance across a broad range of multimodal benchmarks. It natively supports interleaved contexts of up to 256K tokens, seamlessly integrating text, images, and video. The model family includes both dense (2B/4B/8B/32B) and mixture-of-experts (30B-A3B/235B-A22B) variants to accommodate diverse latency-quality trade-offs. Qwen3-VL delivers three core pillars: (i) markedly stronger pure-text understanding, surpassing comparable text-only backbones in several cases; (ii) robust long-con
authors
co-cited works
representative citing papers
CiteVQA requires models to cite specific document regions with bounding boxes alongside answers and finds that even the strongest MLLMs frequently cite the wrong region, with top SAA scores of only 76.0 for closed models and 22.5 for open-source ones.
SenseBench is the first physics-based benchmark with 10K+ instances and dual protocols to evaluate VLMs on remote sensing low-level perception and diagnostic description, revealing domain bias and specific failure modes.
EgoMemReason is a new benchmark showing that even the best multimodal models achieve only 39.6% accuracy on reasoning tasks that require integrating sparse evidence across days in egocentric video.
RuleSafe-VL creates 2,166 rule-conditioned cases from 93 atomic rules and 92 relations across three policy families to diagnose where VLMs fail at rule-based content moderation reasoning.
TraceAV-Bench is the first benchmark for multi-hop trajectory reasoning over long audio-visual videos, showing top models reach only 51-68% accuracy with substantial room for improvement.
PureDocBench shows document parsing is far from solved, with top models at ~74/100, small specialists competing with large VLMs, and ranking reversals under real degradation.
MedHorizon benchmark reveals current multimodal LLMs achieve only 41.1% accuracy on long medical videos due to failures in sparse evidence retrieval and procedural reasoning.
WindowsWorld benchmark shows leading GUI agents achieve under 21% success on multi-application professional tasks, with failures especially on conditional judgment across three or more apps and inefficient execution.
MM-JudgeBench shows substantial cross-lingual performance variance in 22 LVLM judges, with model size and architecture as poor predictors of multilingual robustness.
EVE enables verifiable self-evolution of MLLMs by using a Challenger-Solver architecture to generate dynamic executable visual transformations that produce VQA problems with absolute execution-verified ground truth.
VLMs hallucinate by prioritizing contradictory on-screen text over visual content, addressed via the VisualTextTrap benchmark with 6,057 human-validated samples and the VTHM-MoE dual-encoder framework using dimension-specific experts and adaptive routing.
RefereeBench shows that even the strongest video MLLMs reach only around 60% accuracy on multi-sport refereeing tasks and struggle with rule application and temporal grounding.
VLM-UnBench demonstrates that prompt-based training-free unlearning in VLMs leaves forget accuracy near the no-instruction baseline except under oracle conditions that reveal the target concept.
ScreenParse dataset and ScreenVLM model deliver dense screen parsing that outperforms larger VLMs on PageIoU and transfers to better UI grounding.
GUIGuard-Bench is a new benchmark with annotated GUI screenshots that measures privacy recognition, planning fidelity under protection, and utility impact for trajectory-based GUI agents.
Cultural commonsense in India is mostly regional, with only 39.4% agreement across five regions, and LLMs achieve just 13.4-20.9% accuracy while over-representing North and Central areas.
Molmo2 delivers state-of-the-art open-weight video VLMs with new grounding datasets and training methods that outperform prior open models and match or exceed some proprietary ones on pointing and tracking tasks.
S1-MMAlign is a new large-scale dataset of 15.5 million semantically enhanced scientific image-text pairs created via an AI recaptioning pipeline to improve multimodal understanding.
ToG-Bench is the first benchmark for task-oriented spatio-temporal video grounding in egocentric videos, with explicit-implicit dual grounding and one-to-many object scenarios across 100 ScanNet clips and 2704 instructions.
GaussDet enables open-vocabulary and referring segmentation in 3D Gaussians by learning instance features and aggregating votes from 2D detectors, improving referential grounding by 16.7% mIoU in zero-shot setting.
Goku supplies a 2M-scale dataset, synthesis pipeline, decoupled dual-branch model, and 1000-case benchmark for multi-task instruction-based video editing, reporting up to 8% gains in instruction following.
OmniCoT is a new panoramic reasoning benchmark with 6.7K eval, 1K real, and 14.3K training examples plus a two-stage SFT+GRPO training method to enforce global 360-degree consistency.
MuseBench shows state-of-the-art MLLMs achieve only 48.29% accuracy on intent-level audiovisual arts understanding versus 87.18% for human experts.
citing papers explorer
-
Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Models
Embodied-R1.5 is an 8B EFM achieving SOTA on 16 of 24 embodied VLM benchmarks, fine-tunable to outperform leading VLAs, with claimed zero-shot real-robot generalization.
-
Agent Skills Should Go Beyond Text: The Case for Visual Skills
The paper proposes that reusable agent skills should incorporate visual elements alongside text, introduces three forms of visual skills and an automatic conversion system, and reports better performance on GUI and visual-centric tasks.
-
Towards Understanding Modality Interaction in Multimodal Language Models via Partial Information Decomposition
PID applied to MLLMs identifies task-specific modality interaction profiles that generalize across models, extend to tri-modal cases, and yield initial performance gains via reweighting.
-
Linear Scaling Video VLMs for Long Video Understanding
StateKV is an inference-time technique that replaces quadratic self-attention prefill in video VLMs with a fixed-capacity importance-based recurrent state, keeping accuracy near full attention on long-video benchmarks without retraining.
-
Personalize Your Large Vision-language Models With In-context Prompt Tuning
ICPT adds an adaptive-length projection module and two geometric regularizations to enable efficient, high-accuracy personalization of LVLMs across complex multi-concept tasks.
-
VisionPulse: Dynamic Visual Sparsity for Efficient Multimodal Reasoning
VisionPulse is a step-wise visual token pruning method for LMMs that retains 5% of tokens per step, shortens reasoning traces by 11.2%, and maintains accuracy.
-
Cross-Modal Clinical Knowledge Integration for Mammography Report Generation
MammoRG integrates cross-modal prior clinical knowledge and BI-RADS terminology via two-stage training to generate mammography reports with higher clinical consistency than prior direct image-to-text methods.
-
Archon: A Unified Multimodal Model for Holistic Digital Human Generation
Archon unifies seven modalities via modality-specific tokenizers and an autoregressive backbone pretrained on 72 tasks, plus a 4x-efficient video reparameterization and stepwise 'Thinking in Modality' procedure, and reports superior or comparable results on digital-human tasks.
-
Grounded 3D-Aware Spatial Vision-Language Modeling
GR3D is a VLM that combines explicit 2D, implicit 2D, and monocular 3D grounding mechanisms to improve performance on spatial understanding benchmarks.
-
GenEraser: Generalizable Video Object Removal via Balanced Text-Mask Guidance and Decoupled Locator-Preserver
GenEraser proposes MC-MoE with bipartite text guidance, LD-CFG fusion, and a decoupled locator-preserver architecture for generalizable video object and effect removal, claiming 2.16 dB and 1.44 dB gains on ROSE and VOR-Eval benchmarks.
-
Masked Diffusion Vision-Language Models for Temporal Action Localization
Adapts MDVLMs to TAL via planned training objective and step-level IoU reward, reporting gains over autoregressive baselines on ActivityNet and THUMOS datasets.
-
OccamToken: Efficient VLM Inference with Training-Free and Budget-Adaptive Token Pruning
OccamToken replaces absolute token ranking with register-anchored relative evidence testing to enable adaptive, high-ratio visual token pruning in VLMs while preserving most accuracy.
-
STAMP: Training Explicit Memory for Mobile GUI Agents in Controllable and Scalable Virtual Environments
STAMP trains explicit memory for mobile GUI agents via virtual environments with controlled memory injection, achieving SOTA on the new Memory-World benchmark.
-
Mobile-Aptus: Confidence-Driven Proactive and Robust Interaction in MLLM-based Mobile-Using Agents
Mobile-Aptus uses supervised fine-tuning followed by semantic similarity retrieval and direct preference optimization to calibrate confidence scores in mobile agents, yielding over 17% average task success improvement on four benchmarks.
-
GEM: Generative Supervision Helps Embodied Intelligence
GEM adds generative depth supervision to VLM pre-training and reports improved results on embodied benchmarks plus real-world robot execution.
-
ABot-OCR Technical Report
ABot-OCR is a new end-to-end VLM for direct image-to-Markdown transcription using a custom data engine and structure-constrained RL optimization, reporting SOTA scores of 92.81/93.30 on OmniDocBench v1.5/v1.6.
-
When Think-with-Image Meets Safety: What Determines Multimodal Jailbreak Robustness?
Explicit image-tool interaction in VLMs cuts multimodal jailbreak ASR by ~30% on average; the effect is attributed to a safety-relevant shift in hidden representations rather than image semantics or text traces.
-
Personalized Generative Models for Contextual Debiasing
DecoupleGen personalizes diffusion models to create images with uncommon contexts for debiasing object recognition, yielding consistent gains on scene classification tasks.
-
RAPTOR+: A Visually Grounded Vision-Language Framework to Improve Clinical Trust and Auditability in Automated Cancer Referral Processing
RAPTOR+ shows fine-tuned VLMs achieve higher reading accuracy and substantially better evidence grounding than zero-shot models on 223 colorectal cancer referral forms.
-
Rethinking VLM Representation for VLA Initialization
Experiments indicate original VLM representations are crucial for VLA performance, LoRA outperforms full finetuning, and staged robot-data pretraining yields the strongest initialization.
-
Double Triangle Annotation: A Scalable Human-in-the-Loop Framework for High-Precision Historical Document Annotation
Double Triangle Annotation uses parallel MLLM consensus in two layers to reach WER 0.003 on 1887-1906 French medical directories while auto-accepting 85% of 13,595 fields via model agreement.
-
Does Seeing More Mean Knowing More? Mono-Anchored Advantage Normalization for Multi-Source Visual Reasoning
MARS introduces mono-anchored advantage normalization to quantify information gain from multi-source integration in RLVR, yielding 3.2% and 4.9% gains on GRPO and DAPO.
-
AOEPT: Breaking the Implicit Modality-Reduction Bottleneck in Modality-Missing Prompt Tuning
AOEPT proposes modal-contextualized prompts that distill global modality priors to restore reasoning scope in multimodal transformers under missing-modality conditions.
-
VaaWIT: Visual-Aware Adaptation of Large Language Models for Multilingual Web Image Translation
VaaWIT proposes DSAM and VAA modules to adapt LLMs for multilingual web image translation, claiming outperformance over open-source baselines on benchmarks.
-
Appearance-Invariant Detection of Suggestive Motion via Laban Movement Descriptors
Laban-based kinematic descriptors on SMPL poses achieve 68% accuracy in detecting suggestive motion, comparable to appearance-free video models.
-
Smart-Insertion-V: Photorealistic Video Insertion via a Closed-Loop Feedback Dual-Stream Framework
Smart-Insertion-V is a dual-stream closed-loop framework with Dual-World-View RoPE and a Decoupled Guidance Module that inserts reference objects into videos while achieving stylistic harmony despite domain gaps.
-
Leveraging Foundation Models for Causal Generative Modeling
FM-CGM is a framework that uses a large reasoning model and text-to-image diffusion model for zero-shot visual causal reasoning via concept extractor, manipulator, counterfactual generator, and Causal Semantic Guidance mechanism.
-
GlowGS: Generative Semantic Feature Learning for 3D Gaussian Splatting in Nighttime Glow Scenes
GlowGS improves 3D Gaussian Splatting in nighttime glow scenes via semantic feature generation from diffusion models and novel-view semantic learning with vision foundation models.
-
Swift Sampling: Selecting Temporal Surprises via Taylor Series
Swift Sampling is a training-free frame selection method that uses Taylor expansions on video latent trajectories to pick temporally surprising frames, outperforming uniform sampling on long-video QA tasks.
-
Bernini: Latent Semantic Planning for Video Diffusion
Bernini is a framework that uses an MLLM planner to output semantic representations for a DiT renderer to generate or edit videos, reporting SOTA benchmark performance.
-
MuKV: Multi-Grained KV Cache Compression for Long Streaming Video Question-Answering
MuKV adds multi-grained KV cache compression at patch-frame-segment levels plus semi-hierarchical retrieval to raise accuracy and cut memory in long video question-answering.
-
Enhancing Multimodal Large Language Models for Safety-Critical Driving Video Analysis
A data-fusion pipeline generates pseudo-labels from video, telematics, and CV models to fine-tune QwenVL-2.5 with DoRA adapters, yielding reported gains in detecting and explaining safety-critical driving events.
-
One Sentence, One Drama: Personalized Short-Form Drama Generation via Multi-Agent Systems
A hierarchical multi-agent framework converts a single sentence into a short drama using debate-based scripting, 3D-grounded first frames for spatial consistency, and multi-stage reviewer loops.
-
Foresee-to-Ground: From Predictive Temporal Perception to Evidence-Driven Reasoning for Video Temporal Grounding
F2G improves video temporal grounding accuracy by decoupling event identification from boundary measurement using predictive temporal perception to create citable evidence segments for LLM reasoning.
-
GenEvolve: Self-Evolving Image Generation Agents via Tool-Orchestrated Visual Experience Distillation
GenEvolve introduces a self-evolving agent framework for image generation using tool-orchestrated trajectories and Visual Experience Distillation to achieve claimed SOTA results on benchmarks.
-
Frequency-Domain Regularized Adversarial Alignment for Transferable Attacks against Closed-Source MLLMs
FRA-Attack uses high-pass DCT feature alignment and frequency-domain gradient regularization to boost adversarial transferability across 15 MLLMs from 7 vendors.
-
IndusAgent: Reinforcing Open-Vocabulary Industrial Anomaly Detection with Agentic Tools
IndusAgent achieves state-of-the-art zero-shot performance on industrial anomaly benchmarks by using a custom Indus-CoT dataset, dynamic tool orchestration, and gated RL to optimize anomaly classification, localization, and reasoning.
-
QwenSafe: Multimodal Content Rating Description Identification via Preference-Aligned VLMs
QwenSafe adapts Qwen3-VL-8B via SFT and DPO on a metadata2CRD synthesis pipeline to classify 12 Apple CRDs, reporting large gains in positive-class recall over Qwen3-VL, LLaVA-1.6, and Gemini-2.5-Flash.
-
PixVerve: Advancing Native UHR Image Generation to 100MP with a Large-Scale High-Quality Dataset
PixVerve introduces a 95K ultra-high-resolution image-text dataset and training strategies that enable native 100-megapixel text-to-image generation together with a new evaluation benchmark.
-
Stage-adaptive Token Selection for Efficient Omni-modal LLMs
SEATS adaptively selects and removes non-text tokens before and inside the LLM layers of omni-modal models, yielding 9.3x FLOPs reduction and 4.8x prefill speedup at 10% token retention while keeping 96.3% performance.
-
EgoCoT-Bench: Benchmarking Grounded and Verifiable Operation-Centric Chain of Thought Reasoning for MLLMs
EgoCoT-Bench provides 3,172 verifiable QA pairs across perception, anticipation, and reasoning tasks on egocentric videos, revealing that many MLLMs give answer-correct but evidence-inconsistent explanations.
-
SafeAlign-VLA: A Negative-Enhanced Safe Alignment Framework for Risk-Aware Autonomous Driving
SafeAlign-VLA uses counterfactual safety pairing and anchor-based group relative policy optimization to incorporate negative data for safer VLA-based autonomous driving.
-
Beyond Waypoints: Dual-Heatmap Grounding for Cross-Embodiment Semantic Navigation
A vision-language model outputs dual heatmaps for navigation affordance and facing to ground semantic instructions into executable free space, achieving higher affordance rates than waypoint regression across simulated robot embodiments.
-
Causal Evidence for Attention Head Imbalance in Modality Conflict Hallucination
Causal path-patching analysis across five MLLMs identifies distributed hallucination-driving attention heads and localized resisting heads whose imbalance biases generation toward erroneous text over visual evidence; a conditional intervention MACI suppresses the driving heads and cuts hallucination
-
FAGER: Factually Grounded Evaluation and Refinement of Text-to-Image Models
FAGER is a new agentic framework that creates structured factual rubrics to evaluate and refine text-to-image outputs for implicit factual correctness across science, history, products, and culture.
-
CrossView Suite: Harnessing Cross-view Spatial Intelligence of MLLMs with Dataset, Model and Benchmark
CrossView Suite supplies a 1.6M-sample dataset, scene-disjoint benchmark, and explicit-alignment framework to advance MLLMs from single-view perception to cross-view spatial intelligence.
-
What's Holding Back Latent Visual Reasoning?
Latent visual reasoning fails in current models because standard datasets make oracle latents uninformative and inference-time latents collapse away from useful representations.
-
Self-Evolving Spatial Reasoning in Vision Language Models via Geometric Logic Consistency
SAGE adds duality consistency as an auxiliary reward in GRPO training with a dynamic operation pool to improve spatial reasoning robustness and generalization in VLMs.
-
SafeLens: Deliberate and Efficient Video Guardrails with Fast-and-Slow Screening
SafeLens presents a fast-and-slow video guardrail framework that filters the SafeWatch dataset to 2.4% and adds Chain-of-Thought traces to achieve state-of-the-art moderation performance at reduced inference cost.
-
AffordVLA: Injecting Affordance Representations into Vision-Language-Action Models via Implicit Feature Alignment
AffordVLA improves VLA models for robotic manipulation by implicitly injecting affordance perception through feature alignment with a zero-shot teacher, claiming SOTA results in simulation and real-world tests.