| Benchmark | Model | Rank | Results |
|---|---|---|---|
| chart-question-answering-on-chartqa | Pixel+RF | – | 1:1 Accuracy: 81.3 |
| multimodal-reasoning-on-mmmu | Pixel+RF | – | Accuracy: 54.2 |
| text-to-image-generation-on-dpg-bench | RF-Pixel (Pixel+RF) | – | DPG-Bench Score: 84.15 |
| text-to-image-generation-on-geneval | RF-Pixel (original prompts) | – | Overall: 0.84 |
| visual-question-answering-vqa-on-blink | Pixel+RF | – | Accuracy: 53.0 |
| visual-question-answering-vqa-on-realworldqa | VAE+RF | – | Accuracy: 66.6 |