Interactive Visual and Physical Grounding on MM-CreativityBench (test)
24Gold Correct RateQwen3-VL-32B-Instruct
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Qwen3-VL-32B-InstructTemperature=0, Context Length=Maximum, Parameters=32B2026.05 | 24 | 44.7 | 8.766 | 4.802 | 2.309 | 100 | 75 | 100 | 11.1 | |
| GPT-5.4Temperature=0, Context Length=Maximum2026.05 | 19.2 | 43.5 | 4.177 | 1.661 | 1.492 | 51 | 21.8 | 42.2 | 5.9 | |
| Qwen3-VL-8B-InstructTemperature=0, Context Length=Maximum, Parameters=8B2026.05 | 19.2 | 44.1 | 13.45 | 4.979 | 3.766 | 99.3 | 74.7 | 95.3 | 20.1 | |
| GPT-5.4 MiniTemperature=0, Context Length=Maximum2026.05 | 18.3 | 40.8 | 4.072 | 1.36 | 0.706 | 66.2 | 19.3 | 27.9 | 3.3 | |
| Gemma-4-26B-A4B-itTemperature=0, Context Length=Maximum, Parameters=26B, Architecture=A4B2026.05 | 18.3 | 40.2 | 5.33 | 2.679 | 1.574 | 100 | 47.7 | 90.2 | 4 | |
| Gemma-4-31B-itTemperature=0, Context Length=Maximum, Parameters=31B2026.05 | 16.5 | 35.4 | 3.802 | 1.982 | 0.796 | 100 | 25.6 | 54.5 | 1.8 | |
| InternVL3.5-38BTemperature=0, Context Length=Maximum, Parameters=38B2026.05 | 15.6 | 42.6 | 6.991 | 3.565 | 1.775 | 100 | 52.4 | 94.2 | 6.8 | |
| InternVL3.5-14BTemperature=0, Context Length=Maximum, Parameters=14B2026.05 | 15 | 34.5 | 4.847 | 1.811 | 1.7 | 100 | 21.1 | 96 | 3.2 |