| Gemini 4 Argon: our next era of frontier intelligence | ✓ Link | 91.7 | Gemini 4 Argon | 2026-09-30 |
| Introducing Gemini 3.8 Flash and 3.8 Flash Cyber | ✓ Link | 87.1 | Gemini 3.8 Flash | 2026-09-02 |
| Introducing Gemini 3.7 Flash | ✓ Link | 85.4 | Gemini 3.7 Flash | 2026-08-13 |
| Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities | | 78.7 | Gemini 2.5 Pro | 2025-07-07 |
| Qwen3.8-Flash-Next | ✓ Link | 76.6 | Qwen3.8-Flash-Next | 2026-08-24 |
| Kimi K2.5: Visual Agentic Intelligence | ✓ Link | 75.9 | Kimi K2.5 | 2026-02-02 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 75.5 | Qwen3.5-397B-A17B | 2026-02-16 |
| Active Video Perception: Iterative Evidence Seeking for Agentic Long Video Understanding | | 74.8 | Active Video Perception (Gemini-2.5-Pro) | 2025-12-05 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 74.4 | Qwen3.5-122B-A10B | 2026-02-16 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 73.6 | Qwen3.5-27B | 2026-02-16 |
| VideoJev: System One Model for Video Understanding and Interaction | ✓ Link | 73.4 | VideoJev-35B-A3B | 2026-10-13 |
| Keep It Simple: Multi-Key Episodic Memory Retrieval for Ultra-Long Video Understanding | ✓ Link | 71.8 | MERIT (GPT-5) | 2026-08-07 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 71.4 | Qwen3.5-35B-A3B | 2026-02-16 |
| Qwen3.5-Omni Technical Report | | 71.2 | Qwen3.5-Omni-Plus | 2026-04-21 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 70.0 | Qwen3.5-9B | 2026-02-16 |
| ReflectWorld-MM: An Entity-Oriented Multimodal Memory System for Open-Ended Video Streams | ✓ Link | 69.4 | ReflectWorld-MM | 2026-07-14 |
| VideoJev: System One Model for Video Understanding and Interaction | ✓ Link | 67.9 | VideoJev-4B | 2026-10-13 |
| Qwen3-VL Technical Report | ✓ Link | 67.7 | Qwen3-VL-235B-A22B-Instruct | 2025-11-26 |
| Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities | | 62.7 | Gemini 2.5 Flash | 2025-07-07 |
| WorldMM: Dynamic Multimodal Memory Agent for Long Video Reasoning | ✓ Link | 61.9 | WorldMM-GPT | 2025-12-02 |
| Towards Temporal Compositional Reasoning in Long-Form Sports Videos | | 59.9 | CoTR (Qwen3-VL-4B) | 2026-04-24 |
| GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning | ✓ Link | 59.5 | GLM-4.6V | 2025-07-01 |
| HopChain: Multi-Hop Data Synthesis for Generalizable Vision-Language Reasoning | | 59.07 | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | 2026-03-17 |
| Think While Watching: Online Streaming Segment-Level Memory for Multi-Turn Video Reasoning in Multimodal Large Language Models | ✓ Link | 57.68 | TWW_single-turn (Qwen3-VL-4B, S3) | 2026-03-12 |
| FlexiVideo: Variation-Aware Temporal Dynamics Modeling for Efficient Video Understanding | ✓ Link | 57.3 | FlexiVideo-3B (extra data) | 2026-06-03 |
| VideoChat3: Fully Open Video MLLM for Efficient and Generalist Video Understanding | ✓ Link | 56.7 | VideoChat3-4B | 2026-07-16 |
| EgoVITA: Learning to Plan and Verify for Egocentric Video Reasoning | ✓ Link | 56.5 | EgoVITA (Qwen3-VL) | 2025-11-23 |
| InternVideo3: Agentify Foundation Models with Multimodal Contextual Reasoning | ✓ Link | 55.7 | InternVideo3-8B | 2026-06-10 |
| Towards Effective Long Video Understanding: Dynamic MAS Construction via Meta-Agent | | 55.6 | DyMAC | 2026-09-08 |
| LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligence | ✓ Link | 55.5 | LLaVA-OneVision-2-8B | 2026-05-25 |
| AdaptToken: Entropy-based Adaptive Token Selection for MLLM Long Video Understanding | ✓ Link | 54.8 | Qwen2.5-VL + AdaptToken | 2026-03-30 |
| Reasoning with Memory: A Temporal Granularity-Adaptive Framework for Training-Free Long Video Understanding | | 54.5 | ReMem (LLaVA-Video-7B-Qwen2, 64 frames) | 2026-06-30 |
| CoEvoWhen: Policy-Tool Coevolution for Ultra-Long Video Temporal Grounding | ✓ Link | 54.1 | Qwen3.5-27B + Evolved Skill | 2026-09-30 |
| Molmo2: Open Weights and Data for Vision-Language Models with Video Understanding and Grounding | ✓ Link | 53.9 | Molmo2-4B | 2026-01-15 |
| Open-o3-Video-v2: Seek, Ground, and Reason with Verifiable Spatio-Temporal Evidence | ✓ Link | 53.9 | Open-o3-Video-v2-8B | 2026-10-13 |
| QCA: Query- and Content-Aware Keyframe Selection for Long Video Understanding | ✓ Link | 53.2 | QCA | 2026-07-01 |
| VideoTIR: Accurate Understanding for Long Videos with Efficient Tool-Integrated Reasoning | | 53.1 | VideoTIR (w/ Tool + TAGPO) | 2026-03-26 |
| When Vision Speaks for Sound | ✓ Link | 52.1 | Qwen3-Omni-30B + Thud | 2026-05-13 |
| OmniVideo-R1: Reinforcing Audio-visual Reasoning with Query Intention and Modality Attention | | 51.9 | OmniVideo-R1 (Qwen3-Omni-30B-A3B) | 2026-02-05 |
| OmniSeek: Native Tool Integration for Multi-turn Audio-Visual Reasoning | | 51.4 | OmniSeek (ours) | 2026-10-01 |
| Select, Compress, Reinvest: A Controlled Study of Visual-Token Allocation in Long-Video MLLMs | ✓ Link | 51.39 | GPT-5-mini + OMP (16 compressed frames) | 2026-09-03 |
| VideoDetective: Clue Hunting via both Extrinsic Query and Intrinsic Relevance for Long Video Understanding | ✓ Link | 51.3 | VideoDetective (SeedVL-1.5, 20B) | 2026-03-23 |
| Qwen3-Omni Technical Report | ✓ Link | 51.1 | Qwen3-Omni-Flash-Instruct | 2025-09-22 |
| MOSS-VL Technical Report | ✓ Link | 51.1 | MOSS-VL-Instruct-0708 | 2026-08-15 |
| MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction | ✓ Link | 50.9 | MiniCPM-o 4.5-Instruct | 2026-04-30 |
| Native Active Perception as Reasoning for Omni-Modal Understanding | ✓ Link | 50.5 | OmniAgent-RL-7B | 2026-06-17 |
| Training-Free Speech-Centric Omni Understanding with Frozen VLMs | ✓ Link | 50.5 | MiniCPM4.5-V-TFO-9B | 2026-08-07 |
| MiniCPM-V 4.5: Cooking Efficient MLLMs via Architecture, Data, and Training Recipe | ✓ Link | 50.4 | MiniCPM-V 4.5 | 2025-09-16 |
| LongVideo-R1: Smart Navigation for Low-cost Long Video Understanding | ✓ Link | 50.0 | LongVideo-R1 | 2026-02-24 |
| Light-Omni: Reflex over Reasoning in Agentic Video Understanding with Long-Term Memory | ✓ Link | 49.9 | Light-Omni | 2026-07-06 |
| QSVideo: Query-Conditioned Semantic Temporal Retrieval for Video Understanding | | 49.1 | QSVideo (8B, 32 frames) | 2026-07-06 |
| Zoom-Zero: Reinforced Coarse-to-Fine Video Understanding via Temporal Zoom-in | ✓ Link | 48.1 | Zoom-Zero + Divide-and-conquer | 2025-12-16 |
| Qwen2.5-VL Technical Report | ✓ Link | 47.3 | Qwen2.5-VL-72B | 2025-02-19 |
| CoPE-VideoLM: Codec Primitives For Efficient Video Language Models | | 46.4 | CoPE 7B | 2026-02-13 |
| From Pixels to Words -- Towards Native One-Vision Models at Scale | ✓ Link | 46.4 | NEO-ov 9B | 2026-05-27 |
| Reinforced Attention Learning | | 45.9 | RAL-zero | 2026-02-04 |
| Rethinking Long-Video Efficiency: A Joint Allocation Perspective on Frames, Pixels, and Front-End Latency | | 45.9 | LoHi-SemDiv (Qwen3-VL-4B) | 2026-10-03 |
| VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding | ✓ Link | 45.3 | VideoLLaMA3 7B | 2025-01-22 |
| Qwen2.5-VL Technical Report | ✓ Link | 45.3 | Qwen2.5-VL-7B | 2025-02-19 |
| Incentivizing Versatile Video Reasoning in MLLMs via Data-Efficient Reinforcement Learning | ✓ Link | 44.6 | VideoReasoner | 2026-06-03 |
| LiteFrame: Efficient Vision Encoders Unlock Frame Scaling in Video LLMs | ✓ Link | 43.9 | LiteFrame | 2026-05-17 |
| Qwen2.5-VL Technical Report | ✓ Link | 43.3 | Qwen2.5-VL-3B | 2025-02-19 |
| Mage-VL: An Efficient Codec-Native Streaming Multimodal Foundation Model | ✓ Link | 41.8 | Mage-VL-4B (tc32) | 2026-07-27 |
| Less Data, Faster Convergence: Goal-Driven Data Optimization for Multimodal Instruction Tuning | ✓ Link | 41.06 | GDO (Temp+) | 2026-03-12 |
| PixelUMM: Encoder-Free Unified Image and Video Understanding and Generation | ✓ Link | 40.41 | PixelUMM 8B MoT | 2026-09-29 |
| Rethinking Chain-of-Thought Reasoning for Videos | ✓ Link | 38.9 | Qwen2.5-VL-7B (GRPO, concise reason, trained with token compression & 6x frames) | 2025-12-10 |
| TimeViper: A Hybrid Mamba-Transformer Vision-Language Model for Efficient Long Video Understanding | ✓ Link | 35.6 | TimeViper (w/ TransV) | 2025-11-20 |
| TUNA: Taming Unified Visual Representations for Native Unified Multimodal Models | ✓ Link | 27.4 | Tuna | 2025-12-01 |