Multiple Choice Answering on VIEW2SPACE v1
64.93AccuracyOurs (Grounded CoT)
Evaluation Results
| Method | Links | |
|---|---|---|
| Ours (Grounded CoT)Category=Fine-tuned, Backbone=Qwen3VL-4B2026.03 | 64.93 | |
| GPT-5Category=Closed-source MLLMs2026.03 | 59.86 | |
| Ours (CoT)Category=Fine-tuned, Backbone=Qwen3VL-4B2026.03 | 59.57 | |
| GPT-5-miniCategory=Closed-source MLLMs2026.03 | 49 | |
| Qwen3-VL-8BCategory=Open-source MLLMs2026.03 | 37.41 | |
| GPT-5.2Category=Closed-source MLLMs2026.03 | 37.07 | |
| Ours (Instruct-tuning)Category=Fine-tuned, Backbone=Qwen3VL-4B2026.03 | 36.5 | |
| Qwen3-VL-4BCategory=Open-source MLLMs2026.03 | 35.19 | |
| Molmo2-8BCategory=Open-source Spatial Models2026.03 | 34.57 | |
| Ours (Grounded CoT, Direct Answer at Test)Category=Fine-tuned, Backbone=Qwen3VL-4B2026.03 | 33.95 | |
| GPT-4oCategory=Closed-source MLLMs2026.03 | 33.93 | |
| GPT-5-nanoCategory=Closed-source MLLMs2026.03 | 33.36 | |
| Qwen3-VL-2BCategory=Open-source MLLMs2026.03 | 31.43 | |
| Molmo2-4BCategory=Open-source Spatial Models2026.03 | 31.36 | |
| Gemma-3-12B-itCategory=Open-source MLLMs2026.03 | 31.14 | |
| Internvl2.5-8BCategory=Open-source MLLMs2026.03 | 30.93 | |
| RoBobrain2.0Category=Open-source Spatial Models2026.03 | 30.79 | |
| OpenviewCategory=Open-source Spatial Models2026.03 | 30.72 | |
| Qwen2.5-VL-7BCategory=Open-source MLLMs2026.03 | 30.29 | |
| MINDCUBE-PlainCategory=Open-source Spatial Models2026.03 | 30.21 | |
| Mantis-8B (SigLip)Category=Open-source MLLMs2026.03 | 30.07 | |
| DeepSeek-VL2-SmallCategory=Open-source MLLMs2026.03 | 29.43 | |
| MINDCUBE-AugCategory=Open-source Spatial Models2026.03 | 29 | |
| Random (chance)Category=Baseline2026.03 | 28.59 | |
| Random (frequency)Category=Baseline2026.03 | 28.59 | |
| Spatial-MLLMCategory=Open-source Spatial Models2026.03 | 27.93 | |
| Idefics2-8BCategory=Open-source MLLMs2026.03 | 26.64 |