| Qwen3.5-Omni Technical Report | | 79.0 | Qwen3.5-Omni-Plus | 2026-04-21 |
| InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models | ✓ Link | 78.7 | InternVL3-78B | 2025-04-14 |
| GLM-5.3-Flash: Frontier Intelligence, Flash Cost | ✓ Link | 77.8 | GLM-5.3-Flash | 2026-08-26 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 77.6 | Qwen3.5-397B-A17B | 2026-02-16 |
| PerceptionLM: Open-Access Data and Models for Detailed Visual Understanding | ✓ Link | 77.1 | PLM-8B | 2025-04-17 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 76.6 | Qwen3.5-122B-A10B | 2026-02-16 |
| Qwen3-VL Technical Report | ✓ Link | 76.5 | Qwen3-VL-235B-A22B-Instruct | 2025-11-26 |
| Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding | ✓ Link | 75.9 | Molmo2-8B | 2026-01-15 |
| Qwen3.6 | ✓ Link | 75.5 | Qwen3.6-27B | 2026-04-21 |
| LongCat-Flash-Omni Technical Report | ✓ Link | 75.2 | LongCat-Flash-Omni | 2025-10-31 |
| InternVideo3: Agentify Foundation Models with Multimodal Contextual Reasoning | ✓ Link | 75.0 | InternVideo3-8B | 2026-06-10 |
| GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning | ✓ Link | 74.9 | GLM-4.6V | 2025-07-01 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 74.8 | Qwen3.5-35B-A3B | 2026-02-16 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 74.6 | Qwen3.5-27B | 2026-02-16 |
| Qwen3.6 | ✓ Link | 74.6 | Qwen3.6-35B-A3B | 2026-04-21 |
| Cosmos 3: Omnimodal World Models for Physical AI | ✓ Link | 74.5 | Cosmos3-Super | 2026-06-01 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 74.4 | Qwen3.5-9B | 2026-02-16 |
| EgoVITA: Learning to Plan and Verify for Egocentric Video Reasoning | ✓ Link | 73.8 | EgoVITA (InternVL-3.5) | 2025-11-23 |
| Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution | ✓ Link | 73.6 | Qwen2-VL-72B | 2024-09-18 |
| HopChain: Multi-Hop Data Synthesis for Generalizable Vision-Language Reasoning | | 73.31 | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | 2026-03-17 |
| Unified Spatio-Temporal Token Scoring for Efficient Video VLMs | ✓ Link | 72.7 | Molmo2 (SigLIP2 So400M + Qwen3-4B) with STTS (30%) | 2026-03-18 |
| Annotations as Rollouts: Efficient and Scalable Reinforcement Learning for Video MLLMs | ✓ Link | 72.5 | Video-ORA-9B | 2026-08-20 |
| TLive-Omni: An Omni-Modal Understanding Model for E-Commerce Live Streaming | ✓ Link | 72.5 | TLive-Omni (9B) | 2026-08-21 |
| Streaming Video Instruction Tuning | ✓ Link | 72.3 | Streamo | 2025-12-24 |
| VideoAuto-R1: Video Auto Reasoning via Thinking Once, Answering Twice | ✓ Link | 72.0 | VideoAuto-R1 (Qwen3-VL-8B) | 2026-01-08 |
| Before It Fades: Reinforcing Temporal Representations at Inference Time in VideoLLMs | ✓ Link | 71.6 | InternVL2.5-8B + TAI | 2026-10-01 |
| AdaTooler-V: Adaptive Tool-Use for Images and Videos | ✓ Link | 71.5 | AdaTooler-V-7B | 2025-12-18 |
| MoE-ViE: Mixture of Experts Vision Encoder for Efficient Image and Video Understanding | ✓ Link | 71.5 | MoE-ViE-H/14 + Llama 3.1 8B (42M alignment) | 2026-08-18 |
| Capek 0.5: An Execution-Centric Vision-Language Model for Embodied Intelligence | | 71.32 | Capek0.5-35B-A3B | 2026-08-07 |
| From Pixels to Words -- Towards Native One-Vision Models at Scale | ✓ Link | 70.7 | NEO-ov 9B | 2026-05-27 |
| PixelUMM: Encoder-Free Unified Image and Video Understanding and Generation | ✓ Link | 70.53 | PixelUMM 8B MoT | 2026-09-29 |
| Qwen2.5-VL Technical Report | ✓ Link | 70.4 | Qwen2.5-VL-72B | 2025-02-19 |
| What Should a Streaming Video Model Remember? | | 70.4 | SelectStream-Qwen2.5-VL-7B | 2026-06-15 |
| VisReflect: Latent Visual Reflection for Fine-Grained Perception in Long Visual Context | ✓ Link | 70.4 | VisReflect-7B-Video | 2026-06-29 |
| Qwen2.5-Omni Technical Report | ✓ Link | 70.3 | Qwen2.5-Omni-7B | 2025-03-26 |
| CogniRoute: Learning to Route Social Evidence in Omni-Modal Models | | 70.2 | CogniRoute | 2026-06-18 |
| Clue Matters: Empower Video Reasoning with Brain-Inspired Latent Clue Learning | | 69.8 | ClueNet | 2026-09-08 |
| VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding | ✓ Link | 69.7 | VideoLLaMA3 7B | 2025-01-22 |
| Ovis2.5 Technical Report | ✓ Link | 69.7 | Ovis2.5-9B | 2025-08-15 |
| Attend Before Attention: Efficient and Scalable Video Understanding via Autoregressive Gazing | ✓ Link | 69.7 | NVILA-8B-HD-Video | 2026-03-12 |
| Qwen2.5-VL Technical Report | ✓ Link | 69.6 | Qwen2.5-VL-7B | 2025-02-19 |
| Perception Encoder: The best visual embeddings are not at the output of the network | ✓ Link | 69.4 | PE-Lang G | 2025-04-17 |
| Demo-ICL: In-Context Learning for Procedural Video Knowledge Acquisition | ✓ Link | 69.4 | Demo-ICL (SFT, 7B) | 2026-02-09 |
| LinVT: Empower Your Image-level Large Language Model to Understand Videos | ✓ Link | 69.3 | LinVT-Qwen2-VL
(7B) | 2024-12-06 |
| TLive-Omni: An Omni-Modal Understanding Model for E-Commerce Live Streaming | ✓ Link | 69.0 | TLive-Omni (4B) | 2026-08-21 |
| VideoKR: Towards Knowledge- and Reasoning-Intensive Video Understanding | ✓ Link | 68.9 | VideoKR (SFT + RL) on Qwen2.5-VL-7B-Instruct (Frames=128) | 2026-06-03 |
| Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoders | ✓ Link | 68.7 | Penguin-VL 8B | 2026-03-06 |
| AURA: Always-On Understanding and Real-Time Assistance via Video Streams | ✓ Link | 68.1 | AURA | 2026-04-05 |
| Tarsier: Recipes for Training and Evaluating Large Video Description Models | ✓ Link | 67.6 | Tarsier (34B) | 2024-06-30 |
| InternVideo2: Scaling Foundation Models for Multimodal Video Understanding | ✓ Link | 67.2 | InternVideo2 | 2024-03-22 |
| Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution | ✓ Link | 67.0 | Qwen2-VL-7B | 2024-09-18 |
| Qwen2.5-VL Technical Report | ✓ Link | 67.0 | Qwen2.5-VL-3B | 2025-02-19 |
| LongVU: Spatiotemporal Adaptive Compression for Long Video-Language Understanding | ✓ Link | 66.9 | LongVU (7B) | 2024-10-22 |
| MOSS-VL Technical Report | ✓ Link | 66.7 | MOSS-VL-Instruct-0708 | 2026-08-15 |
| VAST: Video Ability-Stratified Taxonomy for Data-Efficient Video Reasoning | | 66.3 | VideoVAST-7B (extra data) | 2026-06-03 |
| LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligence | ✓ Link | 66.2 | LLaVA-OneVision-2-8B | 2026-05-25 |
| PointThinker: Point-Incentivized Parallel Thinking for Multimodal Large Language Model | | 66.2 | PointThinker-7B (extra data) | 2026-06-03 |
| CurveStream: Boosting Streaming Video Understanding in MLLMs via Curvature-Aware Hierarchical Visual Memory Management | ✓ Link | 66.03 | Qwen2.5-VL-7B (with CurveStream) | 2026-03-20 |
| Let VLMs Grade Their Own Thoughts: A Self-Quantification Approach to Reasoning-Aware Reward Modeling | | 65.9 | Video-RAISE-7B | 2026-06-03 |
| Rethinking Chain-of-Thought Reasoning for Videos | ✓ Link | 65.6 | Qwen2.5-VL-7B (GRPO, concise reason, trained with token compression & 6x frames) | 2025-12-10 |
| HERMES: KV Cache as Hierarchical Memory for Efficient Streaming Video Understanding | ✓ Link | 65.53 | HERMES (Qwen2.5-VL-7B, 4K tokens) | 2026-01-21 |
| Reinforced Attention Learning | | 65.5 | RAL | 2026-02-04 |
| Mage-VL: An Efficient Codec-Native Streaming Multimodal Foundation Model | ✓ Link | 65.5 | Mage-VL-4B (tc16) | 2026-07-27 |
| EventLens: Event-Structure Reinforcement Learning for Video Understanding | ✓ Link | 65.42 | EventLens-8B | 2026-10-12 |
| Less from More: Reinforcing Sparse Video Reasoning from Dense References | ✓ Link | 65.3 | SAVER-4B (extra data) | 2026-10-07 |
| Oryx MLLM: On-Demand Spatial-Temporal Understanding at Arbitrary Resolution | ✓ Link | 64.7 | Oryx(34B) | 2024-09-19 |
| Taming Hallucinations: Boosting MLLMs' Video Understanding via Counterfactual Video Generation | ✓ Link | 63.8 | DNA-Train-7B | 2025-12-30 |
| Less Data, Faster Convergence: Goal-Driven Data Optimization for Multimodal Instruction Tuning | ✓ Link | 63.65 | GDO (Temp+) | 2026-03-12 |
| Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution | ✓ Link | 63.2 | Qwen2-VL-2B | 2024-09-18 |
| Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients | | 62.7 | ZPPO (Qwen3.5-2B) | 2026-06-16 |
| Unified Multimodal Autoregressive Modeling with Shared Context-Visual Tokenizer is Key to Unification | ✓ Link | 62.3 | UniAR | 2026-06-16 |
| See What I Mean: Aligning Vision and Language Representations for Video Fine-grained Object Understanding | ✓ Link | 62.1 | SWIM | 2026-05-18 |
| VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs | ✓ Link | 62.0 | VideoLLaMA2 (72B) | 2024-06-11 |
| Dynin-Omni: Omnimodal Unified Large Diffusion Language Model | ✓ Link | 62.0 | Dynin-Omni (8B) | 2026-03-09 |
| Lance: Unified Multimodal Modeling by Multi-Task Synergy | ✓ Link | 62.0 | Lance (3B) | 2026-05-18 |
| CoPE-VideoLM: Codec Primitives For Efficient Video Language Models | | 61.9 | CoPE 7B | 2026-02-13 |
| Intern Lumina U2: A Multi-Codebook Diffusion Large Language Model for Omni-Visual Understanding and Image Generation | ✓ Link | 61.8 | Intern Lumina U2 (Ascend) | 2026-10-14 |
| Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoders | ✓ Link | 61.7 | Penguin-VL 2B | 2026-03-06 |
| Puzzle Curriculum GRPO for Vision-Centric Reasoning | ✓ Link | 61.0 | Qwen2.5-VL-7B (Mix) | 2025-12-16 |
| CASA: Cross-Attention via Self-Attention for Efficient Vision-Language Fusion | ✓ Link | 60.6 | CAt+v (Qwen2.5-VL, 3B) | 2025-12-22 |
| TimeSuite: Improving MLLMs for Long Video Understanding via Grounded Tuning | ✓ Link | 59.9 | VideoChat-T (7B) | 2024-10-25 |
| mPLUG-Owl3: Towards Long Image-Sequence Understanding in Multi-Modal Large Language Models | ✓ Link | 59.5 | mPLUG-Owl3(7B) | 2024-08-09 |
| LLaVA-OneVision: Easy Visual Task Transfer | ✓ Link | 59.4 | LLaVA-OneVision-72B | 2024-08-06 |
| PPLLaVA: Varied Video Sequence Understanding With Prompt Guidance | ✓ Link | 59.2 | PPLLaVA (7b) | 2024-11-04 |
| Token Reduction via Local and Global Contexts Optimization for Efficient Video Large Language Models | ✓ Link | 59.2 | AOT (LLaVA-Video-7B) | 2026-03-02 |
| VideoGPT+: Integrating Image and Video Encoders for Enhanced Video Understanding | ✓ Link | 58.7 | VideoGPT+ | 2024-06-13 |
| PLLaVA : Parameter-free LLaVA Extension from Images to Videos for Video Dense Captioning | ✓ Link | 58.1 | PLLaVA | 2024-04-25 |
| EarlyTom: Early Token Compression Completes Fast Video Understanding | ✓ Link | 57.8 | EarlyTom (LLaVA-OneVision-7B) | 2026-05-28 |
| MASH-VLM: Mitigating Action-Scene Hallucination in Video-LLMs through Disentangled Spatial-Temporal Representations | | 57.6 | MASH-VLM | 2025-03-20 |
| TimeViper: A Hybrid Mamba-Transformer Vision-Language Model for Efficient Long Video Understanding | ✓ Link | 57.2 | TimeViper | 2025-11-20 |
| DynImg: Key Frames with Visual Prompts are Good Representation for Multi-Modal Video Understanding | | 55.8 | DynImg | 2025-07-21 |
| ST-LLM: Large Language Models Are Effective Temporal Learners | ✓ Link | 54.9 | ST-LLM | 2024-03-30 |
| LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Models | ✓ Link | 54.9 | LLaVA-NeXT-Interleave 14B | 2024-07-10 |
| Generation as Auxiliary Supervision: Enhancing Visual Understanding at Zero Inference Overhead via Decoupled Embedding Prediction | | 54.6 | GAS (2B after pretraining) | 2026-08-12 |
| TUNA: Taming Unified Visual Representations for Native Unified Multimodal Models | ✓ Link | 54.4 | Tuna | 2025-12-01 |
| COLT: Enhancing Video Large Language Models with Continual Tool Usage | | 53.4 | COLT joint | 2025-09-23 |
| MVBench: A Comprehensive Multi-modal Video Understanding Benchmark | ✓ Link | 51.9 | VideoChat2 | 2023-11-28 |
| COLT: Enhancing Video Large Language Models with Continual Tool Usage | | 51.8 | COLT 5x2 | 2025-09-23 |
| COLT: Enhancing Video Large Language Models with Continual Tool Usage | | 50.6 | COLT 10x1 | 2025-09-23 |
| InternVideo-Next: Towards General Video Foundation Models without Video-Text Supervision | | 50.6 | InternVideo-Next (Large) | 2025-12-01 |
| HawkEye: Training Video-Text LLMs for Grounding Text in Videos | ✓ Link | 47.55 | HawkEye | 2024-03-15 |
| SmolVLM: Redefining small and efficient multimodal models | ✓ Link | 46.3 | SmolVLM 2.2B | 2025-04-07 |
| SPHINX-X: Scaling Data and Parameters for a Family of Multi-modal Large Language Models | ✓ Link | 39.7 | SPHINX-Plus | 2024-02-08 |
| TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding | ✓ Link | 38.5 | TimeChat | 2023-12-04 |
| Visual Instruction Tuning | ✓ Link | 36.0 | LLaVa | 2023-04-17 |
| VideoChat: Chat-Centric Video Understanding | ✓ Link | 35.5 | VideoChat | 2023-05-10 |
| Video-LLaMA: An Instruction-tuned Audio-Visual Language Model for Video Understanding | ✓ Link | 34.1 | VideoLLaMA | 2023-06-05 |
| Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models | ✓ Link | 32.7 | Video-ChatGPT | 2023-06-08 |
| InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning | ✓ Link | 32.5 | InstructBLIP | 2023-05-11 |
| MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models | ✓ Link | 18.8 | MiniGPT4 | 2023-04-20 |