| Qwen3.8-Max: A New Bar for Coding and Cowork | ✓ Link | 82.4 | Qwen3.8-Max | 2026-08-03 |
| Kimi K3: Open Frontier Intelligence | ✓ Link | 82.1 | Kimi K3 Max | 2026-07-27 |
| GLM-5.3-Flash: Frontier Intelligence, Flash Cost | ✓ Link | 80.5 | GLM-5.3-Flash | 2026-08-26 |
| Kimi K2.5: Visual Agentic Intelligence | ✓ Link | 80.4 | Kimi K2.5 | 2026-02-02 |
| dots3-note Preview: A Small but Mighty Step Toward Long-Horizon Agency in Real Life | ✓ Link | 79.9 | dots3-note Preview | 2026-08-13 |
| Seed2.0 Model Card: Towards Intelligence Frontier for Real-World Complexity | | 78.2 | Seed2.0 Pro | 2026-06-30 |
| Annotations as Rollouts: Efficient and Scalable Reinforcement Learning for Video MLLMs | ✓ Link | 77.0 | Video-ORA-9B | 2026-08-20 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 75.4 | Qwen3.5-397B-A17B | 2026-02-16 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 74.7 | Qwen3.5-122B-A10B | 2026-02-16 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 73.3 | Qwen3.5-27B | 2026-02-16 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 72.3 | Qwen3.5-35B-A3B | 2026-02-16 |
| ERNIE 5.0 Technical Report | | 72.24 | ERNIE 5.0 | 2026-02-04 |
| Qwen3-VL Technical Report | ✓ Link | 71.1 | Qwen3-VL-235B-A22B-Thinking | 2025-11-26 |
| VideoAuto-R1: Video Auto Reasoning via Thinking Once, Answering Twice | ✓ Link | 71.1 | VideoAuto-R1 (Qwen3-VL-8B) | 2026-01-08 |
| Less from More: Reinforcing Sparse Video Reasoning from Dense References | ✓ Link | 70.9 | SAVER-4B (extra data) | 2026-10-07 |
| HopChain: Multi-Hop Data Synthesis for Generalizable Vision-Language Reasoning | | 68.9 | Qwen3.5-397B-A17B (RLVR w/ Multi-Hop) | 2026-03-17 |
| ParaVT: Taming the Tool Prior Paradox for Parallel Tool Use in Agentic Video Reinforcement Learning | ✓ Link | 68.6 | ParaVT (Qwen3-VL-8B) | 2026-05-19 |
| GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learning | ✓ Link | 68.4 | GLM-4.6V | 2025-07-01 |
| Kwai Keye-VL 1.5 Technical Report | ✓ Link | 68.3 | Keye-VL-1.5-8B | 2025-09-01 |
| SiLVR: A Simple Language-based Video Reasoning Framework | ✓ Link | 68.2 | SiLVR | 2025-05-30 |
| Incentivizing Versatile Video Reasoning in MLLMs via Data-Efficient Reinforcement Learning | ✓ Link | 68.0 | VideoReasoner (Qwen2.5-VL-7B-Instruct + RL) | 2026-06-03 |
| Qwen3.5: Towards Native Multimodal Agents | ✓ Link | 67.8 | Qwen3.5-9B | 2026-02-16 |
| Qwen3.5-Omni Technical Report | | 67.5 | Qwen3.5-Omni-Plus | 2026-04-21 |
| Let VLMs Grade Their Own Thoughts: A Self-Quantification Approach to Reasoning-Aware Reward Modeling | | 66.9 | Video-RAISE-7B | 2026-06-03 |
| Kwai Keye-VL Technical Report | ✓ Link | 66.1 | Keye-VL-8B | 2025-07-02 |
| PointThinker: Point-Incentivized Parallel Thinking for Multimodal Large Language Model | | 66.1 | PointThinker-7B (extra data) | 2026-06-03 |
| Watch Before You Answer: Learning from Visually Grounded Post-Training | ✓ Link | 65.8 | VidGround (32 frames) | 2026-04-06 |
| Insight-V++: Towards Advanced Long-Chain Visual Reasoning with Multimodal Large Language Models | | 65.0 | Insight-V++ | 2026-03-18 |
| VideoKR: Towards Knowledge- and Reasoning-Intensive Video Understanding | ✓ Link | 64.8 | VideoKR (SFT + RL) on Qwen3-VL-8B-Instruct (Frames=128) | 2026-06-03 |
| Rethinking Chain-of-Thought Reasoning for Videos | ✓ Link | 63.6 | Qwen2.5-VL-7B (GRPO, concise reason, trained with token compression & 6x frames) | 2025-12-10 |
| Qwen2.5-VL Technical Report | ✓ Link | 62.9 | Qwen2.5-VL-72B | 2025-02-19 |
| CogniRoute: Learning to Route Social Evidence in Omni-Modal Models | | 60.7 | CogniRoute | 2026-06-18 |
| Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients | | 59.0 | ZPPO (Qwen3.5-2B) | 2026-06-16 |
| Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoders | ✓ Link | 58.7 | Penguin-VL 8B | 2026-03-06 |
| Kimi-VL Technical Report | ✓ Link | 57.5 | Kimi-VL-A3B-Thinking-2506 | 2025-04-10 |
| VideoChat3: Fully Open Video MLLM for Efficient and Generalist Video Understanding | ✓ Link | 56.4 | VideoChat3-4B | 2026-07-16 |
| ResAdapt: Adaptive Resolution for Efficient Multimodal Reasoning | ✓ Link | 56.3 | Qwen3-VL-8B + ResAdapt (128 frames, 22.9% retention) | 2026-03-30 |
| LLaVA-OneVision-2: Towards Next-Generation Perceptual Intelligence | ✓ Link | 56.2 | LLaVA-OneVision-2-8B | 2026-05-25 |
| Kimi-VL Technical Report | ✓ Link | 52.2 | Kimi-VL-A3B-Instruct | 2025-04-10 |
| VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding | ✓ Link | 44.1 | VideoLLaMA3 7B | 2025-01-22 |
| Penguin-VL: Exploring the Efficiency Limits of VLM with LLM-based Vision Encoders | ✓ Link | 41.7 | Penguin-VL 2B | 2026-03-06 |