Text-to-Image Generation on MJHQ-30K
4.48Overall FIDPlayground v2.5
Evaluation Results
| Method | Links | |||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Playground v2.5Resolution=1024x10242024.02 | 4.48 | 22.63 | 23.19 | 23.65 | — | 22.49 | 25.94 | 23.9 | 24.15 | 22.87 | 21.13 | — | — | — | — | — | — | — | — | — | — | — | — | |
| PlaygroundType=Diffusion, Resolution=10242025.03 | 4.48 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LiquidType=Autoregressive, Resolution=5122025.03 | 5.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DIM-4.6B-T2IParams=3.0B (frozen) | 1.6B (trainable), LLM rewriter=true2025.09 | 5.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| STARType=NSP, 1D Token=false, Resolution=2562026.05 | 5.67 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PixArt-aParams.=0.6B, Type=Diffusion2024.12 | 6.14 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PixArtType=Gen. Only2024.10 | 6.14 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PixArt-αParams=0.6B (trainable), LLM rewriter=false2025.09 | 6.14 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PixArtType=Diffusion, Resolution=10242025.03 | 6.14 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PixArtType=Diff., 1D Token=false, Resolution=5122026.05 | 6.14 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MetaQuery-LParams=3.0B (frozen) | 3.2B (trainable), LLM rewriter=true2025.09 | 6.35 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CARType=NTP, 1D Token=true, Resolution=2562026.05 | 6.42 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NextStep-1 + OursSpeedup=3.812026.03 | 6.66 | — | — | — | — | — | — | — | — | — | — | 29.02 | — | — | — | — | — | — | — | — | — | — | — | |
| NextStep-1 + Fast-ARDiffSpeedup=2.972025.12 | 6.68 | — | — | — | — | — | — | — | — | — | — | 28.88 | — | — | — | — | — | — | — | — | — | — | — | |
| NextStep-1Speedup=1.002025.12 | 6.71 | — | — | — | — | — | — | — | — | — | — | 28.67 | — | — | — | — | — | — | — | — | — | — | — | |
| NextStep-1Speedup=1.002026.03 | 6.71 | — | — | — | — | — | — | — | — | — | — | 28.67 | — | — | — | — | — | — | — | — | — | — | — | |
| NextStep-1 + SDSpeedup=2.032025.12 | 6.9 | — | — | — | — | — | — | — | — | — | — | 27.96 | — | — | — | — | — | — | — | — | — | — | — | |
| NextStep-1 + SDSpeedup=2.032026.03 | 6.9 | — | — | — | — | — | — | — | — | — | — | 27.96 | — | — | — | — | — | — | — | — | — | — | — | |
| Playground v2Resolution=1024x10242024.02 | 7.07 | 25.9 | 28.93 | 28.9 | — | 28.83 | 26.96 | 35.82 | 25.05 | 24.66 | 22.08 | — | — | — | — | — | — | — | — | — | — | — | — | |
| FlowTok-HType=Diff., 1D Token=true, Resolution=2562026.05 | 7.15 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| UniTokType=NTP, 1D Token=false, Resolution=2562026.05 | 7.46 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TA-TiTokType=Mask., 1D Token=true, Resolution=2562026.05 | 7.51 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-UType=Und. and Gen., #Params=7B, Resolution=3842024.10 | 7.69 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-U (384)Params.=7B, Type=Autoregressive2024.12 | 7.69 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-U 384Params=7B2024.11 | 7.69 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MUSE-VLType=NTP, 1D Token=false, Resolution=2562026.05 | 7.73 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ILLUMEParams.=7B, Type=Autoregressive2024.12 | 7.76 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DualToken 3BType=Autoregressive, Resolution=2562025.03 | 7.88 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ECADModel=PixArt-α, Caching=Ours, Setting=fast2025.06 | 8.02 | — | — | — | — | — | — | — | — | — | — | 32.78 | — | — | — | — | — | — | — | — | — | — | — | |
| SWITTIType=NSP, 1D Token=false, Resolution=10242026.05 | 8.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NextStep-1 + DMDSpeedup=1.662025.12 | 8.33 | — | — | — | — | — | — | — | — | — | — | 25.19 | — | — | — | — | — | — | — | — | — | — | — | |
| NextStep-1 + DMDSpeedup=1.662026.03 | 8.33 | — | — | — | — | — | — | — | — | — | — | 25.19 | — | — | — | — | — | — | — | — | — | — | — | |
| TiTokType=Mask., 1D Token=true, Resolution=2562026.05 | 8.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ECADModel=PixArt-α, Caching=Ours, Setting=fastest2025.06 | 8.67 | — | — | — | — | — | — | — | — | — | — | 32.24 | — | — | — | — | — | — | — | — | — | — | — | |
| SDXLParams=2.6B (trainable), LLM rewriter=false2025.09 | 8.76 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ECADModel=PixArt-Σ, Caching=Ours, Setting=fast2025.06 | 8.91 | — | — | — | — | — | — | — | — | — | — | 32.52 | — | — | — | — | — | — | — | — | — | — | — | |
| LCDM+LC+DCDDBackbone=Janus-Pro 7B2026.03 | 9.416 | 25.106 | 28.9806 | 26.0182 | 28.158 | 28.5172 | 26.7104 | 38.2032 | 30.2217 | 33.6617 | 25.0277 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SWITTIType=NSP, 1D Token=false, Resolution=5122026.05 | 9.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| JanusFlowParams=1.3B, CFG factor=2, sampling steps=302024.11 | 9.51 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SDXL 1.0 + refinerResolution=1024x10242024.02 | 9.55 | 36.67 | 35.56 | 31.05 | — | 31.49 | 30.78 | 29.95 | 28.42 | 26.27 | 23.19 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SDXLParams.=2.6B, Type=Diffusion2024.12 | 9.55 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SDXLType=Gen. Only2024.10 | 9.55 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SD-XLType=Diffusion, Resolution=10242025.03 | 9.55 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SD-XLType=Diff., 1D Token=false, Resolution=10242026.05 | 9.55 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NoneModel=PixArt-α, Caching=None2025.06 | 9.75 | — | — | — | — | — | — | — | — | — | — | 32.77 | — | — | — | — | — | — | — | — | — | — | — | |
| StepVAR (Infinity)SpeedUp=2.0×2026.03 | 9.85 | 24.78 | — | — | 32.09 | — | — | — | — | 30.35 | — | 0.278 | — | — | — | — | 0.27 | 0.27 | 0.3 | — | — | — | — | |
| ECADModel=PixArt-α, Caching=Ours, Setting=faster2025.06 | 9.92 | — | — | — | — | — | — | — | — | — | — | 32.34 | — | — | — | — | — | — | — | — | — | — | — | |
| Infinity + FastVARSpeedUp=2.0×2026.03 | 10.08 | 25.24 | — | — | 31.96 | — | — | — | — | 30.16 | — | 0.278 | — | — | — | — | 0.27 | 0.27 | 0.3 | — | — | — | — | |
| InfinitySpeedUp=1.0×2026.03 | 10.09 | 25.12 | — | — | 32.13 | — | — | — | — | 30.49 | — | 0.275 | — | — | — | — | 0.27 | 0.26 | 0.29 | — | — | — | — | |
| JanusType=Und. and Gen., #Params=1.3B2024.10 | 10.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| JanusParams.=1.3B, Type=Autoregressive2024.12 | 10.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| JanusParams=1.3B2024.11 | 10.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| JanusParams=1.3B (trainable), LLM rewriter=false2025.09 | 10.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| JanusType=Autoregressive, Resolution=3842025.03 | 10.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| JanusType=NTP, 1D Token=false, Resolution=2562026.05 | 10.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| StepVAR (HART)SpeedUp=1.4×2026.03 | 10.28 | 22 | — | — | 30.28 | — | — | — | — | 30.95 | — | 0.283 | — | — | — | — | 0.28 | 0.28 | 0.3 | — | — | — | — | |
| FORAModel=PixArt-α, Caching=FORA, Setting=N = 22025.06 | 10.33 | — | — | — | — | — | — | — | — | — | — | 32.74 | — | — | — | — | — | — | — | — | — | — | — | |
| TGATEModel=PixArt-α, Caching=TGATE, Setting=m = 15, k = 12025.06 | 10.38 | — | — | — | — | — | — | — | — | — | — | 32.33 | — | — | — | — | — | — | — | — | — | — | — | |
| HART + FastVARSpeedUp=1.4×2026.03 | 10.52 | 23.24 | — | — | 30.92 | — | — | — | — | 36.35 | — | 0.28 | — | — | — | — | 0.28 | 0.27 | 0.3 | — | — | — | — | |
| NoneModel=PixArt-Σ, Caching=None2025.06 | 10.53 | — | — | — | — | — | — | — | — | — | — | 32.65 | — | — | — | — | — | — | — | — | — | — | — | |
| HARTSpeedUp=1.0×2026.03 | 11 | 22.35 | — | — | 30.39 | — | — | — | — | 31.14 | — | 0.283 | — | — | — | — | 0.28 | 0.28 | 0.3 | — | — | — | — | |
| FORAModel=PixArt-α, Caching=FORA, Setting=N = 32025.06 | 11.11 | — | — | — | — | — | — | — | — | — | — | 32.63 | — | — | — | — | — | — | — | — | — | — | — | |
| DuCaModel=PixArt-α, Caching=DuCa, Setting=N = 3, R = 60%2025.06 | 11.69 | — | — | — | — | — | — | — | — | — | — | 32.48 | — | — | — | — | — | — | — | — | — | — | — | |
| ToCaModel=PixArt-α, Caching=ToCa, Setting=N = 3, R = 90%2025.06 | 11.8 | — | — | — | — | — | — | — | — | — | — | 32.35 | — | — | — | — | — | — | — | — | — | — | — | |
| SD3-MediumParams=2.0B (trainable), LLM rewriter=false2025.09 | 11.92 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ToCaModel=PixArt-α, Caching=ToCa, Setting=N = 3, R = 60%2025.06 | 12.01 | — | — | — | — | — | — | — | — | — | — | 32.37 | — | — | — | — | — | — | — | — | — | — | — | |
| DuCaModel=PixArt-α, Caching=DuCa, Setting=N = 3, R = 90%2025.06 | 12.53 | — | — | — | — | — | — | — | — | — | — | 32.39 | — | — | — | — | — | — | — | — | — | — | — | |
| FORAModel=PixArt-Σ, Caching=FORA, Setting=N = 32025.06 | 12.7 | — | — | — | — | — | — | — | — | — | — | 32.28 | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-UType=Und. and Gen., #Params=7B, Resolution=2562024.10 | 12.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-U (256)Params.=7B, Type=Autoregressive2024.12 | 12.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-U 256Params=7B2024.11 | 12.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-UType=Joint Prob. Models2024.10 | 12.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-U 7BType=Autoregressive, Resolution=2562025.03 | 12.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-UType=NTP, 1D Token=false, Resolution=2562026.05 | 12.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Janus-Pro-7BParams=7.0B (trainable), LLM rewriter=false2025.09 | 13.48 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VQ-VAE DecoderBackbone=Janus-Pro 7B2026.03 | 14.2888 | 31.6866 | 35.6408 | 32.141 | 33.1787 | 32.4016 | 30.502 | 42.3565 | 37.6049 | 33.1228 | 27.4424 | — | — | — | — | — | — | — | — | — | — | — | — | |
| NoneCaching=None, Setting=None, TMACs=1190.25, s/img (speedup)=18.30 (1.00x)2025.06 | 14.63 | — | — | — | — | — | — | — | — | — | — | 31.99 | — | — | — | — | — | — | — | — | — | — | — | |
| Show-oType=Discrete Diff., 1D Token=false, Resolution=2562026.05 | 14.99 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VILA-U 3BType=Autoregressive, Resolution=2562025.03 | 15.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Show-oType=Und. and Gen., #Params=1.3B2024.10 | 15.18 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Show-oParams.=1.5B, Type=Autoregressive2024.12 | 15.18 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Show-oParams=1.3B2024.11 | 15.18 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Show-oType=Joint Prob. Models2024.10 | 15.18 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Show-oParams=1.3B (trainable), LLM rewriter=false2025.09 | 15.18 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Show-oType=Discrete Diff., Resolution=2562025.03 | 15.18 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ToCaCaching=ToCa, Setting=N = 4, R = 90%, TMACs=300.41*, s/img (speedup)=7.42 (2.47x)*2025.06 | 15.39 | — | — | — | — | — | — | — | — | — | — | 31.93 | — | — | — | — | — | — | — | — | — | — | — | |
| MMAR-7BType=Joint Prob. Models2024.10 | 15.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Q-DriftModel=PixArt-Sigma, Quantization Framework=ViDiT-Q, Precision=W8A82026.03 | 15.62 | — | — | — | — | — | — | — | — | — | — | 26.67 | — | 0.176 | 21.08 | 0.775 | — | — | — | — | — | — | — | |
| QuantizedModel=PixArt-Sigma, Quantization Framework=ViDiT-Q, Precision=W8A82026.03 | 15.67 | — | — | — | — | — | — | — | — | — | — | 26.65 | — | 0.176 | 21.11 | 0.774 | — | — | — | — | — | — | — | |
| Q-DriftModel=Sana, Quantization Framework=SVDQuant, Precision=W4A42026.03 | 15.68 | — | — | — | — | — | — | — | — | — | — | 27.24 | — | 0.27 | 17.16 | 0.626 | — | — | — | — | — | — | — | |
| QuantizedModel=Sana, Quantization Framework=SVDQuant, Precision=W4A42026.03 | 15.74 | — | — | — | — | — | — | — | — | — | — | 27.23 | — | 0.26 | 17.27 | 0.637 | — | — | — | — | — | — | — | |
| DiRotQModel=PixArt-Σ, Precision=INT W4A4, Inference Steps=202026.05 | 15.9 | — | — | — | — | — | — | — | — | — | — | — | 0.941 | 0.236 | 19.1 | — | — | — | — | — | — | — | — | |
| SVDQuantModel=SANA-1.6B, Precision=NVFP W4A4, Inference Steps=202026.05 | 15.9 | — | — | — | — | — | — | — | — | — | — | — | 1.09 | 0.162 | 19.8 | — | — | — | — | — | — | — | — | |
| ECADCaching=Ours, Setting=slow 256→1024, TMACs=644.05, s/img (speedup)=10.59 (1.73x)2025.06 | 15.98 | — | — | — | — | — | — | — | — | — | — | 31.79 | — | — | — | — | — | — | — | — | — | — | — | |
| Sana (FP)Model=Sana, Precision=FP2026.03 | 15.98 | — | — | — | — | — | — | — | — | — | — | 27.3 | — | — | — | — | — | — | — | — | — | — | — | |
| FPModel=Sana, Quantization Setting=None (Full Precision)2026.03 | 15.98 | — | — | — | — | — | — | — | — | — | — | 27.3 | — | — | — | — | — | — | — | — | — | — | — | |
| DiRotQModel=PixArt-Σ, Precision=NVFP W4A4, Inference Steps=202026.05 | 16 | — | — | — | — | — | — | — | — | — | — | — | 0.946 | 0.224 | 19.6 | — | — | — | — | — | — | — | — | |
| BF16Model=SANA-1.6B, Precision=BF16, Inference Steps=202026.05 | 16 | — | — | — | — | — | — | — | — | — | — | — | 1.1 | — | — | — | — | — | — | — | — | — | — | |
| DiRotQModel=SANA-1.6B, Precision=INT W4A8, Inference Steps=202026.05 | 16 | — | — | — | — | — | — | — | — | — | — | — | 1.1 | 0.078 | 24.1 | — | — | — | — | — | — | — | — | |
| DiRotQModel=SANA-1.6B, Precision=INT W4A4, Inference Steps=202026.05 | 16 | — | — | — | — | — | — | — | — | — | — | — | 1.1 | 0.086 | 23.5 | — | — | — | — | — | — | — | — |