List your company

Text-to-Video Generation on VBench (test)

84.26Total ScoreWan2.1
0.893622.536844.1865.8232Jul 17, 2025Sep 6, 2025Oct 28, 2025Dec 18, 2025Feb 8, 2026Mar 31, 2026May 22, 2026

Evaluation results

MethodLinks
Wan2.1#Params=1.3B, Resolution=832 × 480, NFE=50, First-Frame Enhancement (FFE)=false2026.0584.26————————————85.380.09—
One-Forcing (framewise)#Params=1.3B, Resolution=832 × 480, NFE=1*, First-Frame Enhancement (FFE)=true2026.0583.76————————————85.2277.91—
Self Forcing#Params=1.3B, Resolution=832 × 480, NFE=4, First-Frame Enhancement (FFE)=false2026.0583.46————————————84.7778.24—
SkyReels-V2#Params=1.3B, Resolution=960 × 540, NFE=30, First-Frame Enhancement (FFE)=false2026.0582.67————————————84.774.53—
Pyramid Flow#Params=2B, Resolution=640 × 384, NFE=20, First-Frame Enhancement (FFE)=false2026.0581.72————————————84.7469.62—
One-Forcing (chunkwise)#Params=1.3B, Resolution=832 × 480, NFE=1*, First-Frame Enhancement (FFE)=true2026.0581.6————————————83.6573.41—
Full-ComputeModel=HunyuanVideo, NFC=50, FLOPs (T)=75202026.0281.4———————————————
CausVid#Params=1.3B, Resolution=832 × 480, NFE=4, First-Frame Enhancement (FFE)=false2026.0581.18————————————84.4168.3—
50 stepsLatency(s)=332.92, Speed=1.00×, FLOPs(T)=14050.37, Ratio=1.00×, Mem(GB)=60.222026.0280.93———————————————
RFCModel=HunyuanVideo, m=1, NFC=8.96, FLOPs (T)=1354.652026.0280.83———————18.54080.63520.1329—————
TaylorSeerModel=HunyuanVideo, N=5, m=1, NFC=10, FLOPs (T)=1510.212026.0280.77———————16.36060.51260.1985—————
RFCModel=HunyuanVideo, m=1, NFC=7.09, FLOPs (T)=1072.652026.0280.49———————18.25210.61550.1436—————
DPCache (K = 11)Latency(s)=83.22, Speed=4.05×, FLOPs(T)=3100.33, Ratio=4.53×, Mem(GB)=60.582026.0280.35———————23.110.74620.189—————
TaylorSeer¹Latency(s)=86.11, Speed=3.87×, FLOPs(T)=2821.20, Ratio=4.98×, Mem(GB)=84.472026.0280.33———————18.530.60250.3722—————
SpeCa¹Latency(s)=82.15, Speed=4.05×, FLOPs(T)=2731.61, Ratio=5.14×, Mem(GB)=84.472026.0280.26———————20.090.6370.3554—————
DPCache (K = 9)Latency(s)=70.09, Speed=4.75×, FLOPs(T)=2538.79, Ratio=5.53×, Mem(GB)=60.582026.0280.23———————21.040.68520.2504—————
NOVA#Params=0.6B, Resolution=768 × 480, NFE=25, First-Frame Enhancement (FFE)=false2026.0580.12————————————80.3979.05—
LTX-Video#Params=1.9B, Resolution=768 × 512, NFE=20, First-Frame Enhancement (FFE)=false2026.0580————————————82.370.79—
TaylorSeer²Latency(s)=79.84, Speed=4.17×, FLOPs(T)=2540.46, Ratio=5.53×, Mem(GB)=84.472026.0279.99———————18.230.58990.3967—————
TaylorSeerModel=HunyuanVideo, N=8, m=1, NFC=7, FLOPs (T)=1058.452026.0279.59———————15.19990.44080.2624—————
TaylorSeerModel=HunyuanVideo, N=6, m=1, NFC=9, FLOPs (T)=1359.192026.0279.52———————15.53210.46060.2448—————
TeaCache¹Latency(s)=79.56, Speed=4.18×, FLOPs(T)=2972.71, Ratio=4.73×, Mem(GB)=60.582026.0279.25———————17.550.58010.4081—————
MAGI-1#Params=4.5B, Resolution=832 × 480, NFE=64, First-Frame Enhancement (FFE)=false2026.0579.18————————————82.0467.74—
ASD#Params=1.3B, Resolution=832 × 480, NFE=1*, First-Frame Enhancement (FFE)=true2026.0579.12————————————81.3570.19—
22% stepsLatency(s)=82.47, Speed=4.04×, FLOPs(T)=3095.22, Ratio=4.54×, Mem(GB)=60.222026.0278.89———————16.970.57530.4654—————
FORAModel=HunyuanVideo, N=6, NFC=9, FLOPs (T)=1359.192026.0278.51———————15.52210.43490.2743—————
SpeCa²Latency(s)=72.09, Speed=4.62×, FLOPs(T)=2254.75, Ratio=6.23×, Mem(GB)=84.472026.0278.5———————17.830.58670.4085—————
Causal-Forcing#Params=1.3B, Resolution=832 × 480, NFE=1, First-Frame Enhancement (FFE)=false2026.0578.39————————————80.6769.25—
TeaCache²Latency(s)=70.05, Speed=4.75×, FLOPs(T)=2538.79, Ratio=5.53×, Mem(GB)=60.582026.0278.17———————17.750.58610.4137—————
FORAModel=HunyuanVideo, N=8, NFC=7, FLOPs (T)=1058.452026.0277.5———————14.88780.41140.2959—————
Self Forcing#Params=1.3B, Resolution=832 × 480, NFE=1, First-Frame Enhancement (FFE)=false2026.0577.18————————————79.468.34—
CausVidModel size=1.3B, Guidance strategy=Modulation guidance2026.0265.4398.4586.5957.6519.02———————————
CausVidModel size=1.3B, Guidance strategy=Normalized attention guidance2026.0263.5898.3974.2262.0819.61———————————
CausVidModel size=1.3B, Guidance strategy=Original + CLIP2026.0262.8298.6376.3857.7718.49———————————
CausVidModel size=1.3B, Guidance strategy=Original2026.0262.7298.7675.2557.8519.01———————————
HunyuanModel size=13B, Guidance strategy=Modulation guidance2026.0257.5699.0353.6156.521.09———————————
Baseline vs NABLAJudge=Both are good2025.0757.164.8————————————66.740
HunyuanModel size=13B, Guidance strategy=Original2026.0256.6899.2350.5155.8821.08———————————
BaselineJudge=Baseline is better2025.0720.310.5————————————19.931.4
NABLAJudge=NABLA(0.7) is better, Sparsity=80%2025.0718.415.2————————————13.326.7
Baseline vs NABLAJudge=Both are bad2025.074.19.5————————————0.91.9
BaseModel=Wan2.2-TI2V-5B2026.01—97.54244.4—94.996.465.5————————
Base modelModel=Wan2.2-T2V-A14B, Inference steps=40, Function evaluations=802025.10—97.67—63.62—90.0694.03—————95.7———
DMD2Inference steps=4, Function evaluations=42025.10—98.57—67.02—92.9395.29—————97.08———
FPModel=CogVideoX-2B, Quantization=Full Precision2025.08—97.955055.2525.9194.396.458.69———33.79————
FPModel=CogVideoX-5B, Quantization=Full Precision2025.08—97.6172.2258.8826.4694.6395.5661.8———45.28————
Full fine-tuningModel=Wan2.2-TI2V-5B2026.01—97.545.344.8—95.396.566.2————————
MotiveModel=Wan2.2-TI2V-5B, Selection percentage=10%, Aggregation=majority vote2026.01—97.648.345.6—95.196.665.5————————
Motive w/o MMModel=Wan2.2-TI2V-5B, Motion Masking=false2026.01—97.443.845.2—94.996.564.8————————
Phased DMDInference steps=4, Function evaluations=42025.10—97.74—65.73—91.1594.4—————95.26———
PTQ4DiTModel=CogVideoX-2B, Quantization=W4A42025.08—99.3028.511.1195.3397.6120.66———2.11————
PTQ4DiTModel=CogVideoX-5B, Quantization=W4A42025.08—96.38028.637.3795.0997.2941.48———0————
Q-DiTModel=CogVideoX-2B, Quantization=W4A42025.08—99.14027.668.4396.5298.0926.26———1.16————
Q-DiTModel=CogVideoX-5B, Quantization=W4A42025.08—95.7136.11331.9196.9998.2640.8———0.22————
QuaRotModel=CogVideoX-2B, Quantization=W4A42025.08—97.5413.8939.5919.5792.3596.1843.22———12.21————
QuaRotModel=CogVideoX-5B, Quantization=W4A42025.08—97.7715.2835.3615.3192.7197.2329.41———8.36————
Random selectionModel=Wan2.2-TI2V-5B, Selection percentage=10%2026.01—97.341.644.6—94.796.265.2————————
S2Q-VDiTModel=CogVideoX-2B, Quantization=W4A42025.08—98.0936.1152.3124.993.9996.1553.71———34.23————
S2Q-VDiTModel=CogVideoX-5B, Quantization=W4A42025.08—97.1847.2255.3526.0293.6996.2558.76———36.56————
SmoothQuantModel=CogVideoX-2B, Quantization=W4A42025.08—98.95028.319.7494.6597.6229.76———5.31————
SmoothQuantModel=CogVideoX-5B, Quantization=W4A42025.08—95.761.3929.996.5196.0296.5440.3———0.44————
ViDiT-QModel=CogVideoX-2B, Quantization=W4A42025.08—97.5712.542.0319.6192.4396.0845.56———11.91————
ViDiT-QModel=CogVideoX-5B, Quantization=W4A42025.08—97.0915.2836.7116.9193.0196.3731.95———10.85————