Visual Understanding on HR-Bench 4K (Avg@8 Exact Match)
90.2Avg@8 Exact MatchSenseNova-MARS-32B
Evaluation results
| Method | Links | |
|---|---|---|
| SenseNova-MARS-32BMethod Category=Agentic Model2025.12 | 90.2 | |
| Gemini-2.5-ProMethod Category=Direct Answer2025.12 | 87.3 | |
| Qwen3-VL-235B-A22B-InstructMethod Category=Agentic Model2025.12 | 85.4 | |
| Qwen3-VL-32B-InstructMethod Category=Agentic Model2025.12 | 84.6 | |
| SenseNova-MARS-8BMethod Category=Agentic Model2025.12 | 83.1 | |
| Skywork-R1V4Method Category=Agentic Model2025.12 | 82.8 | |
| Qwen3-VL-8B-InstructMethod Category=Direct Answer2025.12 | 78.9 | |
| DeepEyesV2Method Category=Agentic Model2025.12 | 77.9 | |
| Mini o3Method Category=Agentic Model2025.12 | 77.5 | |
| ThymeMethod Category=Agentic Model2025.12 | 77 | |
| DeepEyesMethod Category=Agentic Model2025.12 | 73.2 | |
| Pixel-ReasonerMethod Category=Agentic Model2025.12 | 72.6 | |
| MonetMethod Category=Agentic Model2025.12 | 71 | |
| Qwen2.5-VL-32B-InstructMethod Category=Direct Answer2025.12 | 69.3 | |
| Qwen2.5-VL-7B-InstructMethod Category=Direct Answer2025.12 | 65.5 | |
| GPT-4oMethod Category=Direct Answer2025.12 | 65 | |
| LLaVA-onevisonMethod Category=Direct Answer2025.12 | 63 |