| SenseNova-MARS: Empowering Multimodal Agentic Reasoning and Search via Reinforcement Learning | | 90.2 | | | SenseNova-MARS-32B | 2025-12-30 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 89.63 | | | Gemini-3.1-Pro | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 89.38 | | | Qwen3.5-397B-A17B | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 89.12 | | | Gemini-3.5-Flash | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 88.13 | | | Vision-OPD-9B | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 86.13 | | | Qwen3-VL-235B-A22B-Instruct | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 85.75 | | | Qwen3.5-9B | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 84.50 | | | Vision-OPD-4B | 2026-05-18 |
| Zooming without Zooming: Region-to-Image Distillation for Fine-Grained Multimodal Perception | | 84.38 | | | ZwZ-8B | 2026-02-12 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 84.38 | | | Qwen3.5-4B | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 84.00 | | | GPT-5.4 | 2026-05-18 |
| SenseNova-MARS: Empowering Multimodal Agentic Reasoning and Search via Reinforcement Learning | | 83.1 | | | SenseNova-MARS-8B | 2025-12-30 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 82.13 | | | GLM-4.6V | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 81.88 | | | Kimi-K2.6 | 2026-05-18 |
| Zooming without Zooming: Region-to-Image Distillation for Fine-Grained Multimodal Perception | | 81.75 | | | ZwZ-4B | 2026-02-12 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 81.12 | | | GPT-5.2 | 2026-05-18 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 79.63 | | | Qwen3-VL-8B-Instruct | 2026-05-18 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 79.4 | 88.8 | 70.0 | Qwen2.5-VL-72B | 2026-03-04 |
| Region-Level Policy Optimization for Fine-grained MLLM Perception | ✓ Link | 78.8 | | | Vision-RL 2 | 2026-09-17 |
| DeepEyesV2: Toward Agentic Multimodal Model | | 77.9 | | | DeepEyesV2 | 2025-11-07 |
| DeepEyesV2: Toward Agentic Multimodal Model | | 77.0 | | | Thyme | 2025-11-07 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 76.3 | 83.5 | 69.0 | InternVL3-38B | 2026-03-04 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 75.5 | 84.5 | 66.5 | InternVL3-78B | 2026-03-04 |
| Zooming without Zooming: Region-to-Image Distillation for Fine-Grained Multimodal Perception | | 75.38 | | | ZwZ-7B | 2026-02-12 |
| DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning | ✓ Link | 75.1 | 91.3 | 59.0 | DeepEyes | 2025-05-20 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 75.0 | 90.1 | 59.7 | DeepScan | 2026-03-04 |
| DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning | ✓ Link | 73.9 | 89.8 | 58.0 | Qwen2.5-VL-32B | 2025-05-20 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 73.50 | | | MiMo-VL-7B-RL | 2026-05-18 |
| DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning | ✓ Link | 72.9 | 86.0 | 60.3 | Pixel-Reasoner | 2025-05-20 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 72.7 | 89.5 | 61.5 | TreeVGR | 2026-03-04 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 71.5 | 88.5 | 55.3 | ZoomRefine | 2026-03-04 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 71.3 | 89.2 | 53.5 | DyFo | 2026-03-04 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 71.0 | 82.0 | 60.0 | InternVL-2-llama3-76B | 2024-08-28 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 70.8 | 79.3 | 62.3 | InternVL3-8B | 2026-03-04 |
| Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillation | | 69.63 | | | MiniCPM-V-4.5 | 2026-05-18 |
| DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning | ✓ Link | 68.8 | 85.2 | 52.2 | Qwen2.5-VL-7B | 2025-05-20 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 66.8 | 76.8 | 56.8 | Gemini 1.5 Flash | 2024-08-28 |
| DeepScan: A Training-Free Framework for Visually Grounded Reasoning in Large Vision-Language Models | | 66.3 | 76.5 | 56.0 | LLaVA-OneVision-72B | 2026-03-04 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 63.4 | 73.3 | 53.5 | DC² (InternVL-1.5-26B) | 2024-08-28 |
| DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning | ✓ Link | 63.0 | 72.0 | 54.0 | LLaVA-OneVision-7B | 2025-05-20 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 60.6 | 69.5 | 51.8 | InternVL-1.5-26B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 59.0 | 70.0 | 48.0 | GPT-4o | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 58.5 | 65.0 | 52.0 | Qwen-VL-Max | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 57.8 | 63.8 | 51.8 | Xcomposer2-4kHD-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 53.6 | 61.3 | 46.0 | LLaVA-HR-X-13B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 53.0 | 65.0 | 41.0 | Qwen-VL-Plus | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 52.9 | 55.3 | 50.5 | LLaVA-v1.6-34B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 52.0 | 57.8 | 46.3 | LLaVA-HR-X-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 51.5 | 52.0 | 51.0 | DC² (LLaVA-v1.5-13B) | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 50.0 | 53.0 | 47.0 | DC² (LLaVA-v1.6-7B) | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 48.1 | 54.3 | 42.0 | Phi3-Vision-4.2B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 47.9 | 49.0 | 46.8 | LLaVA-v1.6-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 46.0 | 45.5 | 46.5 | InternLM-Xcomposer2-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 45.5 | 49.5 | 41.5 | CogVLM-Chat-17B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 45.5 | 49.8 | 41.3 | LLaVA-v1.6-13B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 44.4 | 46.0 | 42.8 | Yi-VL-34B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 44.0 | 44.0 | 44.0 | DC² (Yi-VL-6B) | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 43.3 | 45.2 | 41.3 | LLaVA-v1.5-13B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 42.6 | 42.8 | 42.5 | Yi-VL-6B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 39.3 | 37.3 | 41.3 | InternLM-Xcomposer-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 38.1 | 47.0 | 29.3 | SEAL-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 36.9 | 38.0 | 35.8 | MPLUG2-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 36.1 | 38.5 | 33.8 | LLaVA-v1.5-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 35.5 | 36.8 | 34.3 | Deepseek-VL-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 31.8 | 30.5 | 33.0 | Qwen-VL-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 30.6 | 36.3 | 25.0 | idefics-9B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 25.8 | 25.5 | 26.0 | fuyu-8B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 25.5 | 25.8 | 25.3 | minigptv2-7B | 2024-08-28 |
| Divide, Conquer and Combine: A Training-Free Framework for High-Resolution Image Perception in Multimodal Large Language Models | | 20.5 | 22.5 | 18.5 | VisualGLM-6B | 2024-08-28 |