Text-to-Image Generation on DPG-Bench (test)
88.32Overall FidelityQwen-Image
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Qwen-Imagezero-shot=true2026.02 | 88.32 | 91.32 | 91.56 | 92.02 | 94.31 | 92.73 | — | — | |
| Seedream 3.0zero-shot=true2026.02 | 88.27 | 94.31 | 92.65 | 91.36 | 92.78 | 88.24 | — | — | |
| FAIL-PDbase_model=FLUX.1 Dev, resolution=512x512, CFG=optimal, zero-shot=true2026.02 | 87.32 | 85 | 91.65 | 90.22 | 92.52 | 91.62 | — | — | |
| SparVAR#Scales=11, Skip Strategy=w/ skip last 2 scales, Model Scale=8B2026.02 | 86.468 | 91.729 | 91.068 | 91.031 | 90.354 | 89.072 | 2.28 | — | |
| Infinity#Scales=13, Skip Strategy=w/o skip scales, Model Scale=8B2026.02 | 86.44 | 92.144 | 89.741 | 90.376 | 92.084 | 91.862 | 1 | — | |
| FAIL-PGbase_model=FLUX.1 Dev, resolution=512x512, CFG=optimal, zero-shot=true2026.02 | 86.42 | 89.19 | 91.68 | 90.18 | 92.36 | 88.43 | — | — | |
| SparVAR#Scales=13, Skip Strategy=w/o skip scales, Model Scale=8B2026.02 | 86.411 | 87.306 | 91.51 | 90.862 | 91.565 | 90.193 | 1.57 | — | |
| SparVAR#Scales=10~13, Skip Strategy=w/ dynamic skip last scales, Model Scale=8B2026.02 | 86.409 | 88.227 | 91.863 | 90.703 | 92.239 | 86.758 | 2.05 | — | |
| ZImageCFG=Yes2026.04 | 86.35 | 90.55 | 91.71 | 91.49 | 92.43 | 87.94 | — | — | |
| FastVAR#Scales=13, Skip Strategy=w/o skip scales, Model Scale=8B2026.02 | 86.343 | 81.784 | 90.195 | 89.311 | 94.084 | 84.496 | 1.14 | — | |
| ScaleKV#Scales=13, Skip Strategy=w/o skip scales, Model Scale=8B2026.02 | 86.333 | 81.876 | 90.041 | 89.666 | 94.464 | 92.675 | 0.67 | — | |
| FastVAR#Scales=11, Skip Strategy=w/ skip last 2 scales, Model Scale=8B2026.02 | 86.332 | 87.295 | 91.466 | 90.048 | 91.113 | 90.308 | 1.79 | — | |
| SkipVAR#Scales=10~13, Skip Strategy=w/ dynamic skip last scales, Model Scale=8B2026.02 | 86.293 | 92.648 | 90.591 | 91.032 | 90.088 | 86.008 | 1.71 | — | |
| NPC_FLUXbackbone=FLUX2025.12 | 85.38 | 83.88 | 91.12 | 87.69 | 93.34 | 83.03 | — | — | |
| ZImage+CAFMCFG=Yes2026.04 | 85.21 | 89.55 | 89.83 | 89.99 | 91.2 | 91.88 | — | — | |
| GPT-Image-1 Highzero-shot=true2026.02 | 85.15 | 88.89 | 88.94 | 89.84 | 92.63 | 90.96 | — | — | |
| BAGEL2025.12 | 85.07 | 88.94 | 90.37 | 91.29 | 90.82 | 88.67 | — | — | |
| Show-o2#Params.=1.5B, #Data=177M, Capability Group=Understanding & Generation2026.03 | 85.02 | 87.53 | 90.38 | 91.34 | 90.3 | 91.21 | — | — | |
| Z-Image-Turbozero-shot=true2026.02 | 84.86 | 91.29 | 89.59 | 90.14 | 92.16 | 88.68 | — | — | |
| T2I-R12025.12 | 84.76 | 91.79 | 90.23 | 89.05 | 90.13 | 89.48 | — | — | |
| Janus-Pro-7Bzero-shot=true2026.02 | 84.19 | 86.9 | 88.9 | 89.4 | 89.32 | 89.48 | — | — | |
| Janus Pro2025.12 | 84.19 | 86.9 | 88.9 | 89.4 | 89.32 | 89.48 | — | — | |
| SD3 Mediumzero-shot=true2026.02 | 84.08 | 87.9 | 91.01 | 88.83 | 80.7 | 88.68 | — | — | |
| SD3-mediumsize=medium2025.12 | 84.08 | 87.9 | 91.01 | 88.83 | 80.7 | 88.68 | — | — | |
| SD3-Medium#Params.=2B, #Data=-, Capability Group=Generation Only2026.03 | 84.08 | 87.9 | 91.01 | 88.83 | 80.7 | 88.68 | — | — | |
| FLUX.1-devversion=dev2025.12 | 84 | 82.1 | 89.5 | 88.7 | 91.1 | 89.4 | — | — | |
| FLUX.1 Devzero-shot=true2026.02 | 83.84 | 74.35 | 90 | 88.96 | 90.87 | 88.33 | — | — | |
| ZImage+FMCFG=Yes2026.04 | 83.67 | 90.34 | 90.56 | 88.98 | 88.17 | 90.71 | — | — | |
| OmniGen2version=22025.12 | 83.57 | 88.81 | 88.83 | 90.18 | 89.37 | 90.27 | — | — | |
| DALLE32025.12 | 83.5 | 90.97 | 89.61 | 88.39 | 90.58 | 89.83 | — | — | |
| DALL-E 3#Params.=-, #Data=-, Capability Group=Generation Only2026.03 | 83.5 | 90.97 | 89.61 | 88.39 | 90.58 | 89.83 | — | — | |
| DALL-E 3Model Type=Latent Diffusion Models, Resolution=512x5122026.05 | 83.5 | 90.97 | 89.61 | 88.39 | 90.58 | 89.83 | — | — | |
| CHEERS#Params.=1.5B, #Data=83M, Capability Group=Understanding & Generation2026.03 | 83.48 | 90.84 | 90.24 | 89.42 | 89.71 | 87.37 | — | — | |
| Tar#Params.=1.5B, #Data=403M, Capability Group=Understanding & Generation2026.03 | 82.96 | 83.59 | 89.35 | 86.91 | 93.5 | 80.8 | — | — | |
| SparVAR#Scales=10~13, Skip Strategy=w/ dynamic skip last scales, Model Scale=2B2026.02 | 82.891 | 89.216 | 88.896 | 88.004 | 89.279 | 89.366 | 1.58 | — | |
| ScaleKV#Scales=13, Skip Strategy=w/o skip scales, Model Scale=2B2026.02 | 82.885 | 78.088 | 88.55 | 87.495 | 90.229 | 89.468 | 0.7 | — | |
| SkipVAR#Scales=10~13, Skip Strategy=w/ dynamic skip last scales, Model Scale=2B2026.02 | 82.877 | 85.273 | 89.115 | 89.289 | 89.519 | 84.836 | 1.33 | — | |
| SparVAR#Scales=13, Skip Strategy=w/o skip scales, Model Scale=2B2026.02 | 82.858 | 89.021 | 89.174 | 87.274 | 91.015 | 84.405 | 1.38 | — | |
| Infinity#Scales=13, Skip Strategy=w/o skip scales, Model Scale=2B2026.02 | 82.839 | 88.499 | 88.12 | 88.957 | 88.767 | 84.521 | 1 | — | |
| FastVAR#Scales=13, Skip Strategy=w/o skip scales, Model Scale=2B2026.02 | 82.723 | 77.291 | 89.168 | 88.537 | 91.932 | 84.637 | 1.01 | — | |
| SparVAR#Scales=11, Skip Strategy=w/ skip last 2 scales, Model Scale=2B2026.02 | 82.663 | 83.864 | 88.945 | 88.274 | 90.883 | 81.888 | 1.7 | — | |
| FastVAR#Scales=11, Skip Strategy=w/ skip last 2 scales, Model Scale=2B2026.02 | 82.651 | 81.423 | 88.267 | 88.184 | 90.127 | 89.351 | 1.33 | — | |
| Janus-Pro#Params.=1.5B, #Data=162M, Capability Group=Understanding & Generation2026.03 | 82.63 | 87.58 | 88.63 | 88.17 | 88.98 | 88.3 | — | — | |
| RaPD#Params=1.1B, Model Type=Pixel Diffusion Models, Resolution=512x5122026.05 | 81.6 | 85.1 | 87.7 | 88.9 | 89.6 | 84.1 | — | — | |
| PixNerd-XXL/16#Params=1.2B, Model Type=Pixel Diffusion Models, Resolution=512x512, Same random seed and environment as RaPD=true2026.05 | 81.5 | 87.2 | 87.1 | 88.9 | 87.1 | 89.8 | — | — | |
| DeCo-XXL/16#Params=1.1B, Model Type=Pixel Diffusion Models, Resolution=512x5122026.05 | 81.4 | — | — | — | — | — | — | — | |
| UniWorld-V1version=V12025.12 | 81.38 | 83.64 | 88.39 | 88.44 | 89.27 | 87.22 | — | — | |
| DeCo-XXL/16#Params=1.1B, Model Type=Pixel Diffusion Models, Resolution=512x512, Same random seed and environment as RaPD=true2026.05 | 81.2 | 88.6 | 88.4 | 88.9 | 87.6 | 86 | — | — | |
| OmniGen2025.12 | 81.16 | 87.9 | 88.97 | 88.47 | 87.95 | 83.56 | — | — | |
| PixNerd-XXL/16#Params=1.2B, Model Type=Pixel Diffusion Models, Resolution=512x5122026.05 | 80.9 | 80.5 | 87.9 | 87.2 | 91.3 | 72.8 | — | — | |
| Emu3-Genzero-shot=true2026.02 | 80.6 | 85.21 | 86.68 | 86.84 | 90.22 | 83.15 | — | — | |
| EMU32025.12 | 80.6 | 85.21 | 86.68 | 86.84 | 90.22 | 83.15 | — | — | |
| Emu3#Params.=8B, #Data=-, Capability Group=Understanding & Generation2026.03 | 80.6 | 85.21 | 86.68 | 86.84 | 90.22 | 83.15 | — | — | |
| PixelGen-XXL/16#Params=1.1B, Model Type=Pixel Diffusion Models, Resolution=512x512, Same random seed and environment as RaPD=true2026.05 | 80.4 | 81.4 | 88.8 | 86.7 | 88.9 | 88.9 | — | — | |
| JanusFlow#Params.=1.3B, #Data=-, Capability Group=Understanding & Generation2026.03 | 80.09 | 87.03 | 87.31 | 87.39 | 89.79 | 88.1 | — | — | |
| ZImageCFG=No2026.04 | 79.83 | 87.27 | 87.22 | 86.44 | 88.97 | 89.21 | — | — | |
| Januszero-shot=true2026.02 | 79.68 | 82.33 | 87.38 | 87.7 | 85.46 | 86.41 | — | — | |
| Hunyuan-DiTzero-shot=true2026.02 | 78.87 | 84.59 | 80.59 | 88.01 | 74.36 | 86.41 | — | — | |
| Hunyuan-DiT2025.12 | 78.87 | 84.59 | 80.59 | 88.01 | 74.36 | 86.41 | — | — | |
| PixelFlow-XL/4#Params=0.9B, Model Type=Pixel Diffusion Models, Resolution=512x5122026.05 | 77.9 | — | — | — | — | — | — | — | |
| ZImage+CAFMCFG=No2026.04 | 77.21 | 87.82 | 86.65 | 86.33 | 86.49 | 84.85 | — | — | |
| SDXL2025.12 | 74.65 | 83.27 | 82.43 | 80.91 | 86.76 | 80.41 | — | — | |
| SDXL#Params.=2.6B, #Data=-, Capability Group=Generation Only2026.03 | 74.65 | 83.27 | 82.43 | 80.91 | 86.76 | 80.41 | — | — | |
| TokenFlow-XLsize=XL2025.12 | 73.38 | 78.72 | 79.22 | 81.29 | 85.22 | 71.2 | — | — | |
| ZImage+FMCFG=No2026.04 | 72.25 | 81.34 | 82.96 | 81.71 | 83.17 | 85.07 | — | — | |
| PixArt-α#Params=0.6B, Model Type=Latent Diffusion Models, Resolution=512x5122026.05 | 71.1 | 74.97 | 79.32 | 78.6 | 82.57 | 76.96 | — | — | |
| SD v2#Params=0.9B, Model Type=Latent Diffusion Models, Resolution=512x5122026.05 | 68.1 | 77.67 | 78.13 | 74.91 | 80.72 | 80.66 | — | — | |
| Show-o2025.12 | 67.27 | 79.33 | 75.44 | 78.02 | 84.45 | 60.8 | — | — | |
| Flux.1-devArch.=DiT, Param.=12B, FPS=0.04, Latency=OOM2026.01 | — | — | — | — | — | — | — | 83.8 | |
| IF-XLArch.=U-Net, Param.=5.5B, FPS=0.06, Latency=OOM2026.01 | — | — | — | — | — | — | — | 75.6 | |
| LUMINA-NextArch.=DiT, Param.=2.0B, FPS=0.06, Latency=OOM2026.01 | — | — | — | — | — | — | — | 74.6 | |
| Ours-fullArch.=DiT, Param.=1.6B, FPS=0.28, Latency=15802026.01 | — | — | — | — | — | — | — | 87.2 | |
| Ours-smallArch.=DiT, Param.=0.4B, FPS=0.62, Latency=3602026.01 | — | — | — | — | — | — | — | 85.2 | |
| Ours-tinyArch.=DiT, Param.=0.3B, FPS=0.81, Latency=2802026.01 | — | — | — | — | — | — | — | 84.6 | |
| PixArt-αArch.=DiT, Param.=0.6B, FPS=0.42, Latency=OOM2026.01 | — | — | — | — | — | — | — | 71.1 | |
| PixArt-ΣArch.=DiT, Param.=0.6B, FPS=0.46, Latency=OOM2026.01 | — | — | — | — | — | — | — | 80.5 | |
| Playgroundv2.5Arch.=DiT, Param.=2.6B, FPS=0.18, Latency=OOM2026.01 | — | — | — | — | — | — | — | 75.5 | |
| SANAArch.=Hybrid, Param.=1.6B, FPS=0.91, Latency=OOM2026.01 | — | — | — | — | — | — | — | 84.8 | |
| SD3-MediumArch.=DiT, Param.=2.0B, FPS=0.28, Latency=OOM2026.01 | — | — | — | — | — | — | — | 84.1 | |
| SD3.5-LargeArch.=DiT, Param.=8.1B, FPS=0.08, Latency=OOM2026.01 | — | — | — | — | — | — | — | 85.6 | |
| SDXLArch.=U-Net, Param.=2.6B, FPS=0.18, Latency=OOM2026.01 | — | — | — | — | — | — | — | 74.7 | |
| SnapGenArch.=U-Net, Param.=0.4B, FPS=0.51, Latency=2742026.01 | — | — | — | — | — | — | — | 81.1 |