Real-world Multimodal Understanding on RealWorldQA (Accuracy)
73.99AccuracyInternVL3.5-38B
Evaluation results
| Method | Links | |
|---|---|---|
| InternVL3.5-38BRatio=100%, Backbone=InternVL3.5-38B2026.05 | 73.99 | |
| DivPruneRatio=60%, Backbone=InternVL3.5-38B2026.05 | 73.05 | |
| FastVRatio=60%, Backbone=InternVL3.5-38B2026.05 | 72.88 | |
| F3ARatio=60%, Backbone=InternVL3.5-38B2026.05 | 72.74 | |
| CDPrunerRatio=60%, Backbone=InternVL3.5-38B2026.05 | 72.66 | |
| FastVRatio=40%, Backbone=InternVL3.5-38B2026.05 | 71.25 | |
| InterSketch2026.05 | 70.5 | |
| Thyme-7B2026.05 | 70.2 | |
| DivPruneRatio=40%, Backbone=InternVL3.5-38B2026.05 | 70.07 | |
| CDPrunerRatio=40%, Backbone=InternVL3.5-38B2026.05 | 69.36 | |
| F3ARatio=40%, Backbone=InternVL3.5-38B2026.05 | 69.36 | |
| VisionZipRatio=60%, Backbone=InternVL3.5-38B2026.05 | 69.28 | |
| F3ARatio=20%, Backbone=InternVL3.5-38B2026.05 | 68.52 | |
| Qwen2.5-VL-7B2026.05 | 68.2 | |
| DivPruneRatio=20%, Backbone=InternVL3.5-38B2026.05 | 67.77 | |
| VisionZipRatio=40%, Backbone=InternVL3.5-38B2026.05 | 67.58 | |
| FastVRatio=20%, Backbone=InternVL3.5-38B2026.05 | 66.22 | |
| CDPrunerRatio=20%, Backbone=InternVL3.5-38B2026.05 | 65.04 | |
| Qwen3-VL-8B2026.05 | 61.7 | |
| VisionZipRatio=20%, Backbone=InternVL3.5-38B2026.05 | 60.92 | |
| ThinkMorph-7B2026.05 | 44.5 |