Visual Spatial Intelligence Reasoning on VSI
73.4AccuracyGeoThinker Qwen3vl-8B-32frame
Evaluation Results
| Method | Links | |
|---|---|---|
| GeoThinker Qwen3vl-8B-32frame# Frames=642026.02 | 73.4 | |
| GeoThinker Qwen3vl-8B-32frame# Frames=1282026.02 | 73.4 | |
| GeoThinker Qwen3vl-8B-32frame# Frames=322026.02 | 72.6 | |
| GeoThinker Qwen3vl-8B-8frame# Frames=1282026.02 | 71.2 | |
| GeoThinker Qwen3vl-8B-8frame# Frames=642026.02 | 70.3 | |
| GeoThinker Qwen3vl-8B-8frame# Frames=322026.02 | 69.8 | |
| GeoThinker Qwen3vl-8B-32frame# Frames=162026.02 | 69.2 | |
| SenseNova-SI InternVL3-8B# Frames=642026.02 | 68.8 | |
| SenseNova-SI InternVL3-8B# Frames=322026.02 | 68.7 | |
| Cambrian-S-7B# Frames=1282026.02 | 67.5 | |
| GeoThinker Qwen3vl-8B-8frame# Frames=162026.02 | 67.1 | |
| Cambrian-S-7B# Frames=642026.02 | 66.4 | |
| SenseNova-SI InternVL3-8B# Frames=1282026.02 | 66.3 | |
| SenseNova-SI InternVL3-8B# Frames=162026.02 | 64.6 | |
| VG-LLM-8B*# Frames=642026.02 | 63.7 | |
| Cambrian-S-7B# Frames=322026.02 | 63.6 | |
| VG-LLM-8B*# Frames=1282026.02 | 63.1 | |
| VG-LLM-8B*# Frames=322026.02 | 62.2 | |
| VG-LLM-8B*# Frames=162026.02 | 60.5 | |
| Cambrian-S-7B# Frames=162026.02 | 58.6 |