Text-to-Image Generation on CompBench (Numeracy, 3D Spatial, Overall)
88.5Overall ScoreUniCorn
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| UniCornModel Category=Unified Multimodal Models2026.01 | 88.5 | 83.5 | 84.1 | |
| SeGroSUMM=Harmon-1.5B, Fine-Tuning Dataset=BLIP3o-60k Dataset2026.03 | 88.08 | — | — | |
| SFT*UMM=Harmon-1.5B, Fine-Tuning Dataset=BLIP3o-60k Dataset2026.03 | 87.32 | — | — | |
| RecaUMM=Harmon-1.5B, Fine-Tuning Dataset=BLIP3o-60k Dataset2026.03 | 87.2 | — | — | |
| SeGroSUMM=Harmon-0.5B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 86.9 | — | — | |
| SeGroSUMM=OpenUni-3.6B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 86.14 | — | — | |
| RecaUMM=OpenUni-3.6B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 86 | — | — | |
| OmniGen2Model Category=Unified Multimodal Models2026.01 | 85.8 | 72 | 82.2 | |
| RecaUMM=Harmon-0.5B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 85.7 | — | — | |
| BLIP3-oModel Category=Unified Multimodal Models2026.01 | 84.7 | 71.7 | 81.7 | |
| RecaUMM=Show-o-512, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 84.47 | — | — | |
| SeGroSUMM=Show-o-512, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 84.35 | — | — | |
| SD3 MediumModel Category=Generation Only Models2026.01 | 84.3 | 72.8 | 77.8 | |
| SeGroSUMM=OpenUni-1.6B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 83.83 | — | — | |
| SFT*UMM=Harmon-0.5B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 83.5 | — | — | |
| FLUX.1 devModel Category=Generation Only Models2026.01 | 83.1 | 75.3 | 76.4 | |
| RecaUMM=OpenUni-1.6B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 83.1 | — | — | |
| show-o2Model Category=Unified Multimodal Models2026.01 | 82.8 | 69.7 | 88.6 | |
| SFT*UMM=OpenUni-3.6B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 82.6 | — | — | |
| BAGELModel Category=Unified Multimodal Models2026.01 | 82.2 | 70.4 | 78 | |
| T2I-R1Model Category=Unified Multimodal Models2026.01 | 81.9 | 83.3 | 79.4 | |
| w/o SFTUMM=Harmon-1.5B, Fine-Tuning Dataset=BLIP3o-60k Dataset2026.03 | 81.36 | — | — | |
| SFT*UMM=OpenUni-1.6B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 81.25 | — | — | |
| w/o SFTUMM=Show-o-512, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 80.53 | — | — | |
| w/o SFTUMM=Harmon-0.5B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 80.34 | — | — | |
| SFT*UMM=Show-o-512, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 80.02 | — | — | |
| w/o SFTUMM=OpenUni-3.6B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 78.84 | — | — | |
| SeGroSUMM=Show-o-256, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 78.3 | — | — | |
| RecaUMM=Show-o-256, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 77.5 | — | — | |
| w/o SFTUMM=OpenUni-1.6B, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 75.47 | — | — | |
| SFT*UMM=Show-o-256, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 75.37 | — | — | |
| Janus-ProModel Category=Unified Multimodal Models2026.01 | 74 | 56.4 | 76.2 | |
| w/o SFTUMM=Show-o-256, Fine-Tuning Dataset=MidjourneyV6 Dataset2026.03 | 73.58 | — | — | |
| TwiGModel Category=Unified Multimodal Models2026.01 | — | 61.9 | 38.9 |