Spatial Reasoning on VSI-Bench 1.0 (test)
79.2Average ScoreHuman
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Human2026.05 | 79.2 | 94.3 | 47 | 60.4 | 45.9 | 94.7 | 95.8 | 95.8 | 100 | |
| GeoWeaver-10BScale=10B2026.05 | 75.05 | 84.3 | 58.8 | 83.8 | 68.8 | 73.5 | 92 | 55 | 84.3 | |
| GeoWeaver-5BScale=5B2026.05 | 72.8 | 83.2 | 61.8 | 76.8 | 62.4 | 72.8 | 83.5 | 56.3 | 84.7 | |
| GeoAlignCategory=Spatial-Enhanced Models, Model Size=4B2026.04 | 71.4 | 71.2 | 59.8 | 74.1 | 75 | 72 | 87.1 | 50.5 | 81.7 | |
| SpatialStack-5BScale=5B2026.05 | 67.5 | 71 | 55.6 | 69.1 | 68.2 | 67.3 | 84.1 | 41.2 | 83.5 | |
| GUIDE-9B2026.04 | 64.2 | 73.6 | 58.9 | 71.8 | 72.1 | 70.6 | 86.7 | 47.8 | 32.2 | |
| Cambrian-S-7BScale=7B2026.05 | 62.9 | 68.2 | 45.8 | 72.5 | 67.6 | 66.8 | 69.6 | 39.2 | 73.8 | |
| VG LLM-8B*Training dataset=S1+S2 dataset configuration from VG-LLM2026.04 | 62.2 | 71.4 | 56.8 | 69 | 69.1 | 67.9 | 83.2 | 47.4 | 32.5 | |
| Ours (100% dataset)Category=Open-sourced VLMs, Rank=1, Training Data=100%2026.01 | 61 | 70.8 | 55 | 70.1 | 69.4 | 56.9 | 69.8 | 38.1 | 57.4 | |
| VLM-3R (100% dataset)Category=Open-sourced VLMs, Rank=2, Training Data=100%2026.01 | 60.9 | 70.2 | 49.4 | 69.2 | 67.1 | 65.4 | 80.5 | 45.4 | 40.1 | |
| VLM-3R2026.04 | 60.9 | 70.2 | 49.4 | 69.2 | 67.1 | 65.4 | 80.5 | 45.4 | 40.1 | |
| VLM-3RCategory=Spatial-Enhanced Models, Model Size=7B2026.04 | 60.9 | 70.2 | 49.4 | 69.2 | 67.1 | 65.4 | 80.5 | 45.4 | 40.1 | |
| VLM-3R-7BScale=7B2026.05 | 60.9 | 70.2 | 49.4 | 69.2 | 67.1 | 65.4 | 80.5 | 45.4 | 40.1 | |
| VST-7B-SFT2026.04 | 60.6 | 72 | 44.4 | 74.3 | 68.3 | 59.7 | 55.8 | 44.9 | 65.2 | |
| Ours (25% dataset)Category=Open-sourced VLMs, Rank=3, Training Data=25%2026.01 | 58.8 | 68.2 | 53 | 64.4 | 69.1 | 52.1 | 70.9 | 38.1 | 54.5 | |
| Spa3-VLM-4BModel Category=Spatial Models, Parameters=4B2026.02 | 58.6 | 69 | 39.3 | 70.6 | 62.8 | 57.9 | 59.5 | 36.1 | 73.6 | |
| VG-LLM-8BCategory=Open-sourced VLMs, Rank=4, Parameters=8B2026.01 | 58.2 | 71.7 | 53.8 | 68.8 | 62.1 | 63.8 | 83 | 44.3 | 18.4 | |
| Qwen3-VL-8B-Instruct2026.04 | 57.9 | 67.6 | 47 | 76.3 | 61.9 | 58 | 51 | 35.1 | 66.3 | |
| Qwen3-VL-8B-InstructScale=8B2026.05 | 57.9 | 67.6 | 47 | 76.3 | 61.9 | 58 | 51 | 35 | 66.3 | |
| Cambrian-S-3BModel Category=Spatial Models, Parameters=3B2026.02 | 57.3 | 70.7 | 40.6 | 68 | 46.3 | 64.8 | 61.9 | 27.3 | 78.8 | |
| Cambrian-S-3B2026.04 | 57.3 | 70.7 | 40.6 | 68 | 46.3 | 64.8 | 61.9 | 27.3 | 78.8 | |
| Cambrian-SCategory=Spatial-Enhanced Models, Model Size=3B2026.04 | 57.3 | 70.7 | 40.6 | 68 | 46.3 | 64.8 | 61.9 | 27.3 | 78.8 | |
| GUIDE-5B2026.04 | 55.6 | 70 | 43 | 64.3 | 67 | 58.8 | 50.5 | 35.6 | 56 | |
| VST-7B-SFTScale=7B2026.05 | 55.5 | 68.8 | 37.3 | 74.5 | 62.2 | 55.2 | 48.7 | 41.8 | 55.5 | |
| GPT-52026.04 | 55 | 53.3 | 34.5 | 73.3 | 47.5 | 63.7 | 48.7 | 50.3 | 68.9 | |
| GPT-52026.05 | 55 | 53.3 | 34.5 | 73.3 | 47.5 | 63.7 | 48.7 | 50.3 | 68.9 | |
| VLM-3R (25% dataset)Category=Open-sourced VLMs, Rank=5, Training Data=25%2026.01 | 54 | 69.2 | 41.7 | 67.6 | 62.4 | 58.4 | 47.3 | 43.3 | 41.7 | |
| Gemini-2.5-pro2026.04 | 53.6 | 46 | 37.4 | 68.7 | 54.4 | 62 | 43.9 | 47.4 | 68.8 | |
| Gemini-2.5-ProCategory=Proprietary Models2026.04 | 53.6 | 46 | 37.4 | 68.7 | 54.4 | 62 | 43.9 | 47.4 | 68.8 | |
| Gemini-3-Pro2026.05 | 52.5 | 38 | 37.8 | 72.7 | 44.1 | 59.9 | 55.7 | 45.9 | 66 | |
| VG-LLM-8BModel Category=Spatial Models, Parameters=8B2026.02 | 50.7 | 67.9 | 37.7 | 58.6 | 62 | 46.6 | 40.7 | 32.4 | 59.2 | |
| VG LLM-8B2026.04 | 50.7 | 67.9 | 37.7 | 58.6 | 62 | 46.6 | 40.7 | 32.4 | 59.2 | |
| VG-LLMCategory=Spatial-Enhanced Models, Model Size=8B2026.04 | 50.7 | 67.9 | 37.7 | 58.6 | 62 | 46.6 | 40.7 | 32.4 | 59.2 | |
| Seed-1.62026.04 | 49.9 | 43.5 | 34.4 | 66.1 | 52.8 | 55.1 | 35.7 | 44.3 | 68 | |
| Seed-1.62026.05 | 49.9 | 43.5 | 34.4 | 66.1 | 52.8 | 55.1 | 35.7 | 44.3 | 68 | |
| Spatial-MLLM-4BCategory=Open-sourced VLMs, Rank=6, Parameters=4B2026.01 | 48.4 | 65.3 | 34.8 | 63.1 | 45.1 | 41.3 | 46.2 | 33.5 | 46.3 | |
| Spatial-MLLM-4BModel Category=Spatial Models, Parameters=4B2026.02 | 48.4 | 65.3 | 34.8 | 63.1 | 45.1 | 41.3 | 46.2 | 33.5 | 46.3 | |
| Spatial-MLLM-4B2026.04 | 48.4 | 65.3 | 34.8 | 63.1 | 45.1 | 41.3 | 46.2 | 33.5 | 46.3 | |
| InternVL3Category=Open-Sourced Models, Model Size=78B2026.04 | 48.4 | 71.2 | 53.7 | 44.4 | 39.5 | 55.9 | 39.5 | 28.9 | 54.5 | |
| Spatial-MLLMCategory=Spatial-Enhanced Models, Model Size=4B2026.04 | 48.4 | 65.3 | 34.8 | 63.1 | 45.1 | 41.3 | 46.2 | 33.5 | 46.3 | |
| Grok-42026.05 | 47.9 | 37.2 | 33 | 60.8 | 45.4 | 53.1 | 39.7 | 47.4 | 66.8 | |
| VG-LLM-4BModel Category=Spatial Models, Parameters=4B2026.02 | 47.3 | 66 | 37.8 | 55.2 | 59.2 | 44.6 | 45.6 | 33.5 | 36.4 | |
| VG LLM-4B2026.04 | 47.3 | 66 | 37.8 | 55.2 | 59.2 | 44.6 | 45.6 | 33.5 | 36.4 | |
| VG-LLMCategory=Spatial-Enhanced Models, Model Size=4B2026.04 | 47.3 | 66 | 37.8 | 55.2 | 59.2 | 44.6 | 45.6 | 33.5 | 36.4 | |
| Spatial-MLLM-4BScale=4B2026.05 | 46.3 | 66.7 | 38.1 | 63.6 | 35.5 | 40.4 | 48.2 | 33 | 44.3 | |
| Gemini-1.5 ProCategory=Proprietary Models (API), Rank=12026.01 | 45.4 | 56.2 | 30.9 | 64.1 | 43.6 | 51.3 | 46.3 | 36 | 34.6 | |
| Gemini-1.5-ProModel Category=Proprietary Models2026.02 | 45.4 | 56.2 | 30.9 | 64.1 | 43.6 | 51.3 | 46.3 | 36 | 34.6 | |
| Gemini-1.5-Pro2026.04 | 45.4 | 56.2 | 30.9 | 64.1 | 43.6 | 51.3 | 46.3 | 36 | 34.6 | |
| Gemini-1.5-ProCategory=Proprietary Models2026.04 | 45.4 | 56.2 | 30.9 | 64.1 | 43.6 | 51.3 | 46.3 | 36 | 34.6 | |
| SpatialLadder-3B2026.04 | 44.9 | 62.2 | 35.4 | 62 | 41.4 | 45.6 | 46.5 | 27.3 | 38.5 | |
| SpatialLadder-3BScale=3B2026.05 | 44.9 | 62.2 | 35.4 | 62 | 41.4 | 45.6 | 46.5 | 27.3 | 38.5 | |
| ViLaSR-7B2026.04 | 44.6 | 58.1 | 33.9 | 61.4 | 28.9 | 45.1 | 46.5 | 29.9 | 53.2 | |
| ViLaSR-7BScale=7B2026.05 | 44.6 | 58.1 | 33.9 | 61.4 | 28.9 | 45.1 | 46.5 | 29.9 | 53.2 | |
| Gemini-1.5 FlashCategory=Proprietary Models (API), Rank=22026.01 | 42.1 | 49.8 | 30.8 | 53.5 | 54.4 | 37.7 | 41 | 31.5 | 37.8 | |
| Gemini-1.5-FlashModel Category=Proprietary Models2026.02 | 42.1 | 49.8 | 30.8 | 53.5 | 54.4 | 37.7 | 41 | 31.5 | 37.8 | |
| InternVL3-8B2026.04 | 42.1 | 66.1 | 34.9 | 43.6 | 47.5 | 48 | 39.3 | 26.3 | 31.4 | |
| InternVL3Category=Open-Sourced Models, Model Size=8B2026.04 | 42.1 | 68.1 | 39 | 48.4 | 33.6 | 48.3 | 36.4 | 27.3 | 35.4 | |
| InternVL3-8BScale=8B2026.05 | 42.1 | 66 | 34.9 | 43.6 | 47.5 | 48 | 39.3 | 26.3 | 31.4 | |
| SpaceR-7BScale=7B2026.05 | 41.5 | 44.5 | 24.7 | 53.5 | 37.3 | 41.9 | 46.1 | 29.3 | 54.8 | |
| SPAR-8B2026.04 | 41.1 | — | — | — | — | — | — | — | — | |
| LLaVA-NeXT-Video-72BCategory=Open-sourced VLMs, Rank=7, Parameters=72B2026.01 | 40.9 | 48.9 | 22.8 | 57.4 | 35.3 | 42.4 | 36.7 | 35 | 48.6 | |
| LLaVA-NeXT-Video-72B2026.04 | 40.9 | 48.9 | 22.8 | 57.4 | 35.3 | 42.4 | 36.7 | 35 | 48.6 | |
| LLaVA-NeXT-VideoCategory=Open-Sourced Models, Model Size=72B2026.04 | 40.9 | 48.9 | 22.8 | 57.4 | 35.3 | 42.4 | 36.7 | 35 | 48.6 | |
| LLaVA-OneVision-72BCategory=Open-sourced VLMs, Rank=8, Parameters=72B2026.01 | 40.2 | 43.5 | 23.9 | 57.6 | 37.5 | 42.5 | 39.9 | 32.5 | 44.6 | |
| LLaVA-OneVisionCategory=Open-Sourced Models, Model Size=72B2026.04 | 40.2 | 43.5 | 23.9 | 57.6 | 37.5 | 42.5 | 39.9 | 32.5 | 44.6 | |
| InternVL2-8BModel Category=Open-source Models, Parameters=8B2026.02 | 37.5 | 31.3 | 29 | 48.9 | 44.2 | 38 | 33.4 | 28.9 | 46.4 | |
| Qwen2.5-VLCategory=Open-Sourced Models, Model Size=72B2026.04 | 37 | 25.1 | 29.3 | 54.5 | 38.8 | 38.2 | 37 | 34 | 28.9 | |
| InternVL2-40BCategory=Open-sourced VLMs, Rank=9, Parameters=40B2026.01 | 36 | 34.9 | 26.9 | 46.5 | 31.8 | 42.1 | 32.2 | 34 | 39.6 | |
| LLaVA-NeXT-Video-7BCategory=Open-sourced VLMs, Rank=10, Parameters=7B2026.01 | 35.6 | 48.5 | 14 | 47.8 | 24.2 | 43.5 | 42.4 | 34 | 30.6 | |
| LLaVA-Video-7BModel Category=Open-source Models, Parameters=7B2026.02 | 35.6 | 48.5 | 14 | 47.8 | 24.2 | 43.5 | 42.4 | 34 | 30.6 | |
| LLaVA-NeXT-VideoCategory=Open-Sourced Models, Model Size=7B2026.04 | 35.6 | 48.5 | 14 | 47.8 | 24.2 | 43.5 | 42.4 | 34 | 30.6 | |
| InternVL2-8BCategory=Open-sourced VLMs, Rank=11, Parameters=8B2026.01 | 34.6 | 23.1 | 28.7 | 48.2 | 39.8 | 36.7 | 30.7 | 29.9 | 39.6 | |
| Chance Level (Frequency)Category=Baseline2026.01 | 34 | 62.1 | 32 | 29.9 | 33.1 | 25.1 | 47.9 | 28.4 | 25.2 | |
| GPT-4oCategory=Proprietary Models (API), Rank=32026.01 | 34 | 46.2 | 5.3 | 43.8 | 38.2 | 37 | 41.3 | 31.5 | 28.5 | |
| GPT-4oModel Category=Proprietary Models2026.02 | 34 | 46.2 | 5.3 | 43.8 | 38.2 | 37 | 41.3 | 31.5 | 28.5 | |
| Random Choice (Frequency)2026.04 | 34 | 62.1 | 32 | 29.9 | 33.1 | 25.1 | 47.9 | 28.4 | 25.2 | |
| GPT-4o2026.04 | 34 | 46.2 | 5.3 | 43.8 | 38.2 | 37 | 41.3 | 31.5 | 28.5 | |
| GPT-4oCategory=Proprietary Models2026.04 | 34 | 46.2 | 5.3 | 43.8 | 38.2 | 37 | 41.3 | 31.5 | 28.5 | |
| Random Choice(Frequency)2026.05 | 34 | 62.1 | 32 | 29.9 | 33.1 | 25.1 | 47.9 | 28.4 | 25.2 | |
| Qwen2.5VL-7BModel Category=Open-source Models, Parameters=7B2026.02 | 33 | 40.9 | 14.8 | 43.4 | 10.7 | 38.6 | 38.5 | 33 | 29.8 | |
| Qwen2.5-VLCategory=Open-Sourced Models, Model Size=7B2026.04 | 33 | 40.9 | 14.8 | 43.4 | 10.7 | 38.6 | 38.5 | 33 | 29.8 | |
| LLaVA-OneVision-7BCategory=Open-sourced VLMs, Rank=12, Parameters=7B2026.01 | 32.4 | 47.7 | 20.2 | 47.4 | 12.3 | 42.5 | 35.2 | 29.4 | 24.4 | |
| LLaVA-OneVision-7BModel Category=Open-source Models, Parameters=7B2026.02 | 32.4 | 47.7 | 20.2 | 47.4 | 12.3 | 42.5 | 35.2 | 29.4 | 24.4 | |
| LLaVA-OneVisionCategory=Open-Sourced Models, Model Size=7B2026.04 | 32.4 | 47.7 | 20.2 | 47.4 | 12.3 | 42.5 | 35.2 | 29.4 | 24.4 | |
| Qwen2.5-VL-7B-InstructScale=7B2026.05 | 32.3 | 32.9 | 18.2 | 43.9 | 31.7 | 38 | 37.4 | 28.4 | 28 | |
| Bagel-7B-MoT2026.04 | 31.4 | 30.1 | 29.2 | 35.5 | 25.8 | 34.9 | 41.4 | 30.4 | 24.1 | |
| Bagel-7B-MoTScale=7B2026.05 | 31.4 | 30.1 | 29.2 | 35.5 | 25.8 | 34.9 | 41.4 | 30.4 | 24.1 | |
| VILA-1.5-40BCategory=Open-sourced VLMs, Rank=13, Parameters=40B2026.01 | 31.2 | 22.4 | 24.8 | 48.7 | 22.7 | 40.5 | 25.7 | 31.5 | 32.9 | |
| VILA-1.5-40B2026.04 | 31.2 | 22.4 | 24.8 | 48.7 | 22.7 | 40.5 | 25.7 | 31.5 | 32.9 | |
| VILA-1.5Category=Open-Sourced Models, Model Size=40B2026.04 | 31.2 | 22.4 | 24.8 | 48.7 | 22.7 | 40.5 | 25.7 | 31.5 | 32.9 | |
| Qwen2.5VL-3BModel Category=Open-source Models, Parameters=3B2026.02 | 30.6 | 24.3 | 24.7 | 31.7 | 22.6 | 38.3 | 41.6 | 26.3 | 21.2 | |
| LongVA-7BCategory=Open-sourced VLMs, Rank=14, Parameters=7B2026.01 | 29.2 | 38 | 16.6 | 38.9 | 22.2 | 33.1 | 43.3 | 25.4 | 15.7 | |
| LongVA-7B2026.04 | 29.2 | 38 | 16.6 | 38.9 | 22.2 | 33.1 | 43.3 | 25.4 | 15.7 | |
| LongVACategory=Open-Sourced Models, Model Size=7B2026.04 | 29.2 | 38 | 16.6 | 38.9 | 22.2 | 33.1 | 43.3 | 25.4 | 15.7 | |
| VILA-1.5-8BCategory=Open-sourced VLMs, Rank=15, Parameters=8B2026.01 | 28.9 | 17.4 | 21.8 | 50.3 | 18.8 | 32.1 | 34.8 | 31 | 24.8 | |
| VILA-1.5-8BModel Category=Open-source Models, Parameters=8B2026.02 | 28.9 | 17.4 | 21.8 | 50.3 | 18.8 | 32.1 | 34.8 | 31 | 24.8 | |
| VILA-1.5Category=Open-Sourced Models, Model Size=8B2026.04 | 28.9 | 17.4 | 21.8 | 50.3 | 18.8 | 32.1 | 34.8 | 31 | 24.8 | |
| LLaVA-OneVision-0.5BCategory=Open-sourced VLMs, Rank=16, Parameters=0.5B2026.01 | 28 | 46.1 | 28.4 | 15.4 | 28.3 | 28.9 | 36.9 | 34.5 | 5.8 | |
| InternVL2-2BCategory=Open-sourced VLMs, Rank=17, Parameters=2B2026.01 | 27.4 | 21.8 | 24.9 | 22 | 35 | 33.8 | 44.2 | 30.5 | 7.1 | |
| LongVILA-8BCategory=Open-sourced VLMs, Rank=18, Parameters=8B2026.01 | 21.6 | 29.1 | 9.1 | 16.7 | 0 | 29.6 | 30.7 | 32.5 | 25.5 |