| Benchmark | Model | Rank | Results |
|---|---|---|---|
| multimodal-reasoning-on-mmmu | Gemma3 + AuditDM | – | Accuracy: 45.2 |
| vision-language-understanding-on-mmbench-v1-1 | Gemma3 + AuditDM | – | Accuracy: 75.0 |
| vision-language-understanding-on-mmstar | Gemma3 + AuditDM | – | Accuracy: 52.4 |
| visual-question-answering-vqa-on-mmt-bench | Gemma3 + AuditDM | – | Accuracy: 58.9 |
| visual-question-answering-vqa-on-realworldqa | Gemma3 + AuditDM | – | Accuracy: 61.4 |