Spatial Understanding on MMSI-Bench
48ScoreGemini
Evaluation Results
| Method | Links | |
|---|---|---|
| GeminiModel Version/Scale=3.0 Pro2026.04 | 48 | |
| SeedModel Version/Scale=2.02026.04 | 47.6 | |
| QwenModel Version/Scale=3.5 A17B2026.04 | 43.8 | |
| GPT-5Model Category=Proprietary Models2026.06 | 41.9 | |
| HY-EmbodiedModel Version/Scale=0.5 MoE A32B2026.04 | 39.2 | |
| Gemini 2.5 ProModel Category=Proprietary Models2026.06 | 37.6 | |
| KimiModel Version/Scale=K2.52026.04 | 36.5 | |
| GPT-5 miniModel Category=Proprietary Models2026.06 | 34.9 | |
| Gemini 2.5 FlashModel Category=Proprietary Models2026.06 | 33.1 | |
| InternVL3.5-8BModel Category=Open-Weight Models, Parameter count=8B2026.06 | 30.5 | |
| MiniCPM-V-4.5-8BModel Category=Open-Weight Models, Parameter count=4.5-8B2026.06 | 28.1 | |
| VideoChat-Flash-7BModel Category=Open-Weight Models, Parameter count=7B2026.06 | 27.6 | |
| InternVideo3Model Category=Open-Weight Models2026.06 | 27.6 | |
| Qwen3-VL-8BModel Category=Open-Weight Models, Parameter count=8B2026.06 | 27 | |
| InternVideo2.5-7BModel Category=Open-Weight Models, Parameter count=7B2026.06 | 26.9 | |
| Keye-VL-1.5-8BModel Category=Open-Weight Models, Parameter count=1.5-8B2026.06 | 26.6 |