| Benchmark | Model | Rank | Results |
|---|---|---|---|
| multimodal-reasoning-on-mmmu | Hydra-X | – | Accuracy: 51.5 |
| text-to-image-generation-on-geneval | Hydra-X | – | Overall: 0.88 |
| text-to-image-generation-on-geneval | Hydra-XTok (img distill; Decompressor distilled with video teacher) | – | Overall: 0.72 |