Computer Vision Perception on CV-Bench
89ScoreKimi
Evaluation results
| Method | Links | |
|---|---|---|
| KimiModel Version/Scale=K2.52026.04 | 89 | |
| HY-EmbodiedModel Version/Scale=0.5 MoE A32B2026.04 | 88.8 | |
| QwenModel Version/Scale=3.5 A17B2026.04 | 88.6 | |
| SeedModel Version/Scale=2.02026.04 | 88.5 | |
| Vero Q3I-8BRL=Qwen3VL, Initial Model=8B Inst2026.04 | 87.9 | |
| Vero Q3T-8BRL=Qwen3VL, Initial Model=8B Think2026.04 | 87.4 | |
| Q3VL 8B-ThkRL=N/A, Initial Model=N/A2026.04 | 86 | |
| Q3VL 8B-InsRL=N/A, Initial Model=N/A2026.04 | 85.5 | |
| GeminiModel Version/Scale=3.0 Pro2026.04 | 85.4 | |
| Vero Mi-7BRL=MiMoVL, Initial Model=7B-SFT2026.04 | 84.3 | |
| MiMoVL 7B-RLRL=MiMoVL, Initial Model=7B-SFT2026.04 | 83.5 | |
| LLaVA OV1.5-RLRL=LLaVA, Initial Model=OV1.5-Ins2026.04 | 82.9 | |
| GPT-5 NanoRL=N/A, Initial Model=N/A2026.04 | 82.5 | |
| Vero Q25-7BRL=Qwen25VL, Initial Model=7B Inst2026.04 | 82 | |
| Mo2-O 7BRL=SFT Only, Initial Model=N/A2026.04 | 81.7 | |
| Q25VL 7B-InsRL=N/A, Initial Model=N/A2026.04 | 80.4 | |
| Qwen3-VL-8B-Thinking + DiGParameter Scale=8B, Thinking Mode=true, Grounding Strategy=DiG2025.12 | 0.858 | |
| Qwen3-VL-8B-ThinkingParameter Scale=8B, Thinking Mode=true2025.12 | 0.85 | |
| Qwen3-VL-4B-Thinking + DiGParameter Scale=4B, Thinking Mode=true, Grounding Strategy=DiG2025.12 | 0.838 | |
| Qwen2.5-VL-72BParameter Scale=72B2025.12 | 0.83 | |
| Qwen3-VL-4B-ThinkingParameter Scale=4B, Thinking Mode=true2025.12 | 0.828 | |
| Qwen2.5-VL-7BParameter Scale=7B2025.12 | 0.649 |