Multi-image Visual Question Answering on MUIRBench
76.4AccuracyQwen3-VL-32B-Instruct + SpatialBoost
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen3-VL-32B-Instruct + SpatialBoostEncoder Enhancement=SpatialBoost2026.03 | 76.4 | |
| Qwen3-VL-32B-Instruct2026.03 | 72.8 | |
| InternVL3-38B + SpatialBoostEncoder Enhancement=SpatialBoost2026.03 | 70.7 | |
| InternVL3-38B2026.03 | 63.8 |