Spatial Reasoning on CV-Bench (2D/3D/Avg Metrics)
79Average Spatial ScoreQwen2.5-VL-7B
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen2.5-VL-7BModel Category=Open-Source General Models2026.05 | 79 | 75 | 83.1 | |
| Qwen2.5-VL-7BParameters=7B2026.05 | 79 | 75 | 83.1 | |
| SpatialForgeQwen3-VL-2BModel Category=Ours2026.05 | 78.7 | 72.2 | 85.2 | |
| Qwen3-VL-2BModel Category=Open-Source General Models2026.05 | 78.6 | 73.7 | 83.4 | |
| CaMo-3BParameters=3B2026.05 | 76.3 | 73.5 | 79.2 | |
| Spa3-VLMModel=Spa3-VLM2026.02 | 75.6 | 72.9 | 78.3 | |
| GPT-4oModel Category=Proprietary Models2026.05 | 75.4 | 69.4 | 81.3 | |
| GPT-4oAccess=Proprietary2026.05 | 75.4 | 69.4 | 81.3 | |
| InternVL-2.5-4BParameters=4B2026.05 | 74.4 | 73.5 | 75.1 | |
| Spatial-MLLM-4BModel=Spatial-MLLM-4B2026.02 | 73.8 | — | — | |
| SpatialLadder-3BModel=SpatialLadder-3B2026.02 | 73.7 | 72.4 | 74.9 | |
| SpatialLadder-3BModel Category=Open-sourced Specialized Models2026.05 | 73.7 | 72.4 | 74.9 | |
| SpatialThinker-3BModel=SpatialThinker-3B2026.02 | 73.6 | 71 | 76.3 | |
| Qwen2.5-VL-3BModel=Qwen2.5-VL-3B2026.02 | 70.6 | 69.1 | 72.2 | |
| Qwen2.5-VL-3BModel Category=Open-Source General Models2026.05 | 70.6 | 69.1 | 72.2 | |
| Qwen2.5-VL-3BParameters=3B2026.05 | 70.6 | 69.1 | 72.2 | |
| LLaVA-OneVision-7BModel Category=Open-Source General Models2026.05 | 58.3 | 53.2 | 63.5 | |
| LLaVA-OneVision-7BParameters=7B2026.05 | 58.3 | 50.6 | 63.5 | |
| SpaceQwen2.5-VL-3B-InstructModel Category=Open-sourced Specialized Models2026.05 | 57.8 | 54.9 | 60.7 | |
| Kimi-VL-A3BParameters=3B2026.05 | 48.3 | 41.9 | 54.7 | |
| SpaceR-7BModel Category=Open-sourced Specialized Models2026.05 | 43.2 | 49.9 | 36.4 | |
| RandomModel Category=Baseline2026.05 | 25 | 25 | 25 |