Visual Perception on HR-8K (test)
82AccuracyZwZ-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| ZwZ-8BInference Strategy=Single Forward Pass, Model Scale=8B2026.02 | 82 | |
| Skywork-R1V4Inference Strategy=Agentic, Model Scale=30B2026.02 | 79.8 | |
| ZwZ-4BInference Strategy=Single Forward Pass, Model Scale=4B2026.02 | 79.5 | |
| SenseNova-MARS-8BInference Strategy=Agentic, Model Scale=8B2026.02 | 78.4 | |
| Qwen3-VL-8B+toolInference Strategy=Tool Use, Model Scale=8B2026.02 | 78 | |
| Qwen3-VL-8BInference Strategy=Single Forward Pass, Model Scale=8B2026.02 | 74.6 | |
| Qwen3-VL-4B+toolInference Strategy=Tool Use, Model Scale=4B2026.02 | 74.4 | |
| DeepEyesV2Inference Strategy=Agentic, Model Scale=7B2026.02 | 73.8 | |
| Mini o3Inference Strategy=Agentic, Model Scale=7B2026.02 | 73.3 | |
| ZwZ-7BInference Strategy=Single Forward Pass, Model Scale=7B2026.02 | 73.3 | |
| Qwen3-VL-4BInference Strategy=Single Forward Pass, Model Scale=4B2026.02 | 72.9 | |
| DeepEyesInference Strategy=Agentic, Model Scale=7B2026.02 | 72.6 | |
| ThymeInference Strategy=Agentic, Model Scale=7B2026.02 | 72 | |
| Qwen2.5-VL-7BInference Strategy=Single Forward Pass, Model Scale=7B2026.02 | 67.9 | |
| Pixel-ReasonerInference Strategy=Agentic, Model Scale=7B2026.02 | 66.1 |