Fine-grained VQA on V*
93.2AccuracyiVGR-Qwen3-VL-32B
Evaluation results
| Method | Links | |
|---|---|---|
| iVGR-Qwen3-VL-32BTools=No2026.05 | 93.2 | |
| iVGR-Qwen3-VL-8BTools=No2026.05 | 90.1 | |
| GPT-5.4Tools=No2026.05 | 88 | |
| Gemini-3.1-Pro-PreviewTools=No2026.05 | 87.4 | |
| iVGR-Qwen2.5-VL-7BTools=No2026.05 | 86.4 | |
| Qwen2.5-VL-72BTools=No2026.05 | 85.9 | |
| Qwen3-VL-32BTools=No2026.05 | 83.8 | |
| TreeVGR-7BTools=No2026.05 | 83.8 | |
| Qwen3-VL-8BTools=No2026.05 | 82.7 | |
| DeepEyes-7BTools=Yes2026.05 | 82.7 | |
| Thyme-7BTools=Yes2026.05 | 82.2 | |
| DeepEyesV2-7BTools=Yes2026.05 | 81.8 | |
| Qwen2.5-VL-32BTools=No2026.05 | 80.1 | |
| Qwen2.5-VL-7BTools=No2026.05 | 78.5 | |
| Qwen3-VL-4BTools=No2026.05 | 78.5 | |
| LLaVA-OneVision-7BTools=No2026.05 | 72.8 | |
| InternVL3-8BTools=No2026.05 | 70.2 | |
| GRIT-3BTools=No2026.05 | 54.5 |