Text-to-Image Generation on MSCOCO zero-shot 30k
5.76FID (MSCOCO 30k, zero-shot)Sana 1.0 1.6B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Sana 1.0 1.6BParameters=1.6B2025.05 | 5.76 | — | |
| PG 2.5Parameters=2.6B2025.05 | 6.09 | — | |
| PixArt-ΣParameters=0.6B, Training Data=46M2025.05 | 6.15 | — | |
| SDXLParameters=2.6B, Training Data=1.6B2025.05 | 6.63 | — | |
| DeepFloydSize=2562024.10 | 6.7 | — | |
| SiD2 16-step distilledSize=512, NFE=162024.10 | 6.7 | — | |
| eDiff-ISize=2562024.10 | 6.9 | — | |
| PartiSize=2562024.10 | 7.2 | — | |
| SDv1.5Size=512, NFE=200, evaluation_source=DMD22024.10 | 7.2 | — | |
| ImagenSize=2562024.10 | 7.3 | — | |
| FuseDiTParameters=2B, Training Data=26M2025.05 | 7.54 | — | |
| Lumina NextParameters=2B2025.05 | 7.58 | — | |
| SiD2Size=512, NFE=256, sigmoid_bias=-32024.10 | 8.1 | — | |
| simple diffusionSize=2562024.10 | 8.3 | — | |
| DMD2Size=512, NFE=42024.10 | 8.3 | — | |
| simple diffusionSize=512, NFE=2562024.10 | 9.6 | — | |
| FLUX.1 [dev]Parameters=12B2025.05 | 10.15 | — | |
| DREAM2026.03 | 10.4 | 31.5 | |
| SD 3 MParameters=2B, Training Data=1B2025.05 | 11.92 | — | |
| SCAD2026.03 | 12.3 | 27.6 | |
| GALIP2026.03 | 14.6 | 29.3 | |
| 4M-LExtra Labels=Yes2026.03 | 30.1 | 23.1 | |
| mmDiT-ImpCS Evaluation samples=40K2026.03 | — | 26.6 | |
| mmDiT-SD3CS Evaluation samples=40K2026.03 | — | 24.8 | |
| SDXLCS Evaluation samples=40K2026.03 | — | 25.4 |