Document Understanding on AI2D
0.977AccuracyGemini 3-Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 3-Pro2026.02 | 0.977 | |
| Gemini 2.5-Pro2026.02 | 0.9709 | |
| Qwen3-VLmode=Thinking2026.02 | 0.9696 | |
| ERNIE 5.02026.02 | 0.9689 | |
| GPT-5tier=High2026.02 | 0.9563 | |
| Qwen3-VLParameters=8B2026.04 | 0.835 | |
| BARD-VLParameters=8B, B=42026.04 | 0.832 | |
| BARD-VLParameters=4B, B=322026.04 | 0.828 | |
| InternVL3.5Parameters=8B2026.04 | 0.821 | |
| Qwen3-VLParameters=4B2026.04 | 0.81 | |
| InternVL3.5Parameters=4B2026.04 | 0.806 | |
| Dream-VLParameters=7B2026.04 | 0.804 | |
| SDAR-VLParameters=8B2026.04 | 0.796 | |
| LLaDA-VParameters=8B2026.04 | 0.778 | |
| Dimple-VLParameters=7B2026.04 | 0.742 | |
| BARD-VLParameters=2B, B=322026.04 | 0.726 | |
| LaviDaParameters=8B2026.04 | 0.701 | |
| GenLIPArch=g/16, Data=8.0B2026.05 | 0.689 | |
| GenLIPArch=So/16, Data=8.0B2026.05 | 0.672 | |
| SigLIP2Arch=So/16, Data=40.0B2026.05 | 0.67 | |
| SigLIP2Arch=L/16, Data=40.0B2026.05 | 0.667 | |
| SigLIP2Arch=g/16, Data=40.0B2026.05 | 0.667 | |
| GenLIPArch=L/16, Data=8.0B2026.05 | 0.666 | |
| SigLIPArch=L/16, Data=40.0B2026.05 | 0.664 | |
| OVision2Arch=L/16, Data=12.8B2026.05 | 0.656 | |
| CLIPArch=L/14, Data=12.8B2026.05 | 0.645 | |
| AIMv2Arch=L/14, Data=12.0B2026.05 | 0.642 | |
| MMaDAParameters=8B2026.04 | 0.549 |