Visual Counting on CountBenchQA
93.1AccuracyVero Q3T-8B
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Vero Q3T-8BRL=Qwen3VL, Initial Model=8B Think2026.04 | 93.1 | — | — | — | |
| Vero Q3I-8BRL=Qwen3VL, Initial Model=8B Inst2026.04 | 90.4 | — | — | — | |
| VAPModel=Qwen3-VL-8B2026.03 | 90.22 | 0.2 | 0.85 | 95.93 | |
| Q3VL 8B-ThkRL=N/A, Initial Model=N/A2026.04 | 89.6 | — | — | — | |
| Mo2-O 7BRL=SFT Only, Initial Model=N/A2026.04 | 89.4 | — | — | — | |
| Q3VL 8B-InsRL=N/A, Initial Model=N/A2026.04 | 88.8 | — | — | — | |
| BaselineModel=Qwen3-VL-8B2026.03 | 87.98 | 0.23 | 0.86 | 95.52 | |
| LLaVA OV1.5-RLRL=LLaVA, Initial Model=OV1.5-Ins2026.04 | 86.8 | — | — | — | |
| MiMoVL 7B-RLRL=MiMoVL, Initial Model=7B-SFT2026.04 | 86.4 | — | — | — | |
| Q25VL 7B-InsRL=N/A, Initial Model=N/A2026.04 | 85.9 | — | — | — | |
| Vero Mi-7BRL=MiMoVL, Initial Model=7B-SFT2026.04 | 84.9 | — | — | — | |
| Vero Q25-7BRL=Qwen25VL, Initial Model=7B Inst2026.04 | 83.7 | — | — | — | |
| VAPModel=Qwen2.5-VL-7B2026.03 | 82.08 | 0.58 | 4.28 | 91.24 | |
| BaselineModel=Qwen2.5-VL-7B2026.03 | 80.44 | 0.66 | 4.34 | 89.61 | |
| GPT-5 NanoRL=N/A, Initial Model=N/A2026.04 | 75.4 | — | — | — | |
| VAPModel=LLaVA-1.5-7B2026.03 | 47.86 | 1.15 | 1.78 | 67.01 | |
| BaselineModel=LLaVA-1.5-7B2026.03 | 42.16 | 1.38 | 2.08 | 59.67 |