Real-World Understanding on HRBench 4K (Score)
86.9ScoreGemini 2.5 Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 2.5 ProTool=Code, Param Size=-2025.11 | 86.9 | |
| DeepEyesV2Tool=General, Param Size=7B2025.11 | 77.9 | |
| ThymeTool=Code, Param Size=7B2025.11 | 77 | |
| DeepEyesTool=Crop, Param Size=7B2025.11 | 75.1 | |
| Qwen2.5-VLTool=✗, Param Size=32B2025.11 | 74.1 | |
| Pixel-ReasonerTool=Crop, Param Size=7B2025.11 | 74 | |
| Qwen2.5-VLTool=✗, Param Size=7B2025.11 | 71.6 | |
| InternVL3Tool=✗, Param Size=8B2025.11 | 70 | |
| LLaVA-OVTool=✗, Param Size=7B2025.11 | 63 | |
| GPT-4oTool=Code, Param Size=-2025.11 | 60.6 |