Text-to-Video Generation on T2V-CompBench
84.58Consistency Attribute ScoreVQQA (Gemini-3-Pro)
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| VQQA (Gemini-3-Pro)Method Category=VQQA, Backbone=CogVideoX-5B, Refinement Model=Gemini-3-Pro, Optimization rounds=42026.03 | 84.58 | 7.92 | 66.03 | 27.73 | 50.91 | 68.5 | 68.54 | 53.46 | |
| VQQA (GPT-4o)Method Category=VQQA, Backbone=CogVideoX-5B, Refinement Model=GPT-4o, Optimization rounds=42026.03 | 82.96 | 4.88 | 58.73 | 28.51 | 47.75 | 66.72 | 69.57 | 51.3 | |
| VQQA (Gemini-3-Pro)Generation Strategy=VQQA, N=5, Optimization rounds=42026.03 | 82 | 26.44 | 65.73 | 59.7 | 56.47 | 74.31 | 68.02 | 61.81 | |
| VideoScore2Method Category=Best-of-N, Backbone=CogVideoX-5B, Scoring Model=VideoScore2, N=52026.03 | 80.86 | 1.99 | 56.9 | 26.82 | 39.53 | 58.24 | 61.54 | 46.55 | |
| VQAScoreMethod Category=Best-of-N, Backbone=CogVideoX-5B, Scoring Model=VQAScore, N=52026.03 | 80.83 | 2.48 | 56.62 | 26.24 | 44.48 | 64.31 | 65.96 | 48.7 | |
| GPT-4o (Best-of-N)Method Category=Best-of-N & VLM-Rating, Backbone=CogVideoX-5B, Scoring Model=GPT-4o, N=52026.03 | 80.17 | 2 | 56.36 | 27.29 | 42.19 | 60.59 | 66.06 | 47.81 | |
| Gemini-3-Pro (Best-of-N)Method Category=Best-of-N & VLM-Rating, Backbone=CogVideoX-5B, Scoring Model=Gemini-3-Pro, N=52026.03 | 79 | 2.34 | 58.36 | 28.63 | 43.42 | 60.31 | 58.97 | 47.29 | |
| Gemini-3-ProGeneration Strategy=Best-of-N & VLM-Rating, N=52026.03 | 78.7 | 27.56 | 62.98 | 55.61 | 49.25 | 73.63 | 65.33 | 59.01 | |
| VQAScoreGeneration Strategy=Best-of-N, N=52026.03 | 78.21 | 26.13 | 63.76 | 52.03 | 54.59 | 74.83 | 66.99 | 59.51 | |
| VideoScore2Generation Strategy=Best-of-N, N=52026.03 | 78.19 | 24.76 | 60.99 | 54.49 | 46.7 | 73.52 | 65.98 | 57.8 | |
| Wan2.1-1.3B + STASInference Time (s)=234.532026.03 | 77.36 | 3.4 | — | — | — | — | — | — | |
| Veo3.1Generation Strategy=Vanilla Generation2026.03 | 76.45 | 25.7 | 59.76 | 51.18 | 43.77 | 71.81 | 62.87 | 55.93 | |
| VPOMethod Category=Prompt Optimizer, Backbone=CogVideoX-5B2026.03 | 76.43 | 4.52 | 54.18 | 26.07 | 45.72 | 64.09 | 68.83 | 48.55 | |
| Wan2.1-1.3BInference Time (s)=234.472026.03 | 76.27 | 2.72 | — | — | — | — | — | — | |
| CogVideoX-5BMethod Category=Vanilla Generation, Backbone=CogVideoX-5B2026.03 | 61.64 | 2.19 | 51.72 | 26.58 | 37.06 | 53.33 | 60.69 | 41.89 | |
| WAN 2.2 (5B)Mode=Factorized T2V (Ours)2025.12 | 0.943 | 0.393 | 0.71 | 0.282 | 0.74 | 0.831 | 0.873 | 0.682 | |
| WAN 2.2 (14B)Mode=Factorized T2V (Ours)2025.12 | 0.929 | 0.183 | 0.696 | 0.328 | 0.757 | 0.839 | 0.891 | 0.661 | |
| WAN 2.2 (14B)Mode=Factorized I2V + Text2025.12 | 0.927 | 0.217 | 0.68 | 0.32 | 0.781 | 0.85 | 0.886 | 0.666 | |
| WAN 2.1 (1B)Mode=Factorized T2V (Ours)2025.12 | 0.92 | 0.384 | 0.656 | 0.262 | 0.402 | 0.79 | 0.844 | 0.608 | |
| CogVideo 1.5 (5B)Mode=Factorized T2V (Ours)2025.12 | 0.916 | 0.316 | 0.645 | 0.255 | 0.55 | 0.77 | 0.838 | 0.613 | |
| Wan2.1-14B + TTOMModel Type=DiT based, Enhancement=TTOM2025.10 | 0.8982 | 0.1182 | 0.6275 | 0.4922 | 0.5881 | 0.8152 | 0.7691 | 0.6155 | |
| WAN 2.2 (14B)Mode=T2V (upsampled)2025.12 | 0.897 | 0.084 | 0.643 | 0.291 | 0.605 | 0.675 | 0.762 | 0.565 | |
| CogVideoX-5B + TTOMModel Type=DiT based, Enhancement=TTOM2025.10 | 0.8782 | 0.0341 | 0.6173 | 0.4351 | 0.5081 | 0.7191 | 0.7502 | 0.5632 | |
| WAN 2.1 (1B)Mode=T2V (upsampled)2025.12 | 0.871 | 0.117 | 0.59 | 0.244 | 0.47 | 0.572 | 0.684 | 0.507 | |
| CogVideoX-5B + DyST-XLModel Type=DiT based, Enhancement=DyST-XL2025.10 | 0.8696 | 0.0221 | 0.611 | 0.2712 | 0.3969 | 0.7321 | 0.6536 | 0.5081 | |
| WAN 2.2 (5B)Mode=T2V (upsampled)2025.12 | 0.866 | 0.137 | 0.606 | 0.245 | 0.523 | 0.509 | 0.69 | 0.511 | |
| Wan2.1-14B + LVDModel Type=DiT based, Enhancement=LVD2025.10 | 0.861 | 0.0829 | 0.5753 | 0.2864 | 0.4707 | 0.8107 | 0.7201 | 0.5439 | |
| WAN 2.2 (14B)Mode=T2V2025.12 | 0.844 | 0.127 | 0.639 | 0.299 | 0.553 | 0.624 | 0.765 | 0.55 | |
| Wan2.1-14BModel Type=DiT based2025.10 | 0.8369 | 0.057 | 0.5709 | 0.2696 | 0.5113 | 0.7504 | 0.7239 | 0.5314 | |
| WAN 2.2 (5B)Mode=T2V2025.12 | 0.814 | 0.177 | 0.577 | 0.257 | 0.419 | 0.487 | 0.629 | 0.48 | |
| Kling 1.02025.01 | 0.8045 | 0.2256 | 0.615 | 0.2448 | 0.3044 | — | — | — | |
| CogVideo 1.5 (5B)Mode=T2V (upsampled)2025.12 | 0.794 | 0.144 | 0.526 | 0.24 | 0.338 | 0.414 | 0.573 | 0.433 | |
| WAN 2.1 (1B)Mode=T2V2025.12 | 0.793 | 0.15 | 0.541 | 0.232 | 0.383 | 0.461 | 0.648 | 0.458 | |
| VIDEOREPAIRBase Model=T2V-turbo2024.11 | 0.7675 | — | 0.5807 | — | 0.2709 | — | — | 0.5439 | |
| CogVideo 1.5 (5B)Mode=T2V2025.12 | 0.762 | 0.191 | 0.491 | 0.252 | 0.305 | 0.308 | 0.491 | 0.4 | |
| Wan + SCMAPRBackbone=Wan, Prompt Refinement Method=SCMAPR2026.04 | 0.756 | 0.297 | — | 0.346 | — | 0.691 | — | 0.523 | |
| CogVideoX-5B + LVDModel Type=DiT based, Enhancement=LVD2025.10 | 0.7534 | 0.0219 | 0.5274 | 0.3291 | 0.3825 | 0.6826 | 0.6204 | 0.4739 | |
| LaVie-RAPO++Backbone=LaVie2025.10 | 0.742 | 0.294 | — | — | — | 0.632 | 0.849 | — | |
| BlobGEN-VidLayout Generator=GPT-4o2025.01 | 0.74 | 0.265 | 0.6725 | 0.388 | 0.391 | — | — | — | |
| VIDEOREPAIRBase Model=VideoCrafter22024.11 | 0.7275 | — | 0.569 | — | 0.3278 | — | — | 0.5383 | |
| Latte-RAPO++Backbone=Latte2025.10 | 0.727 | 0.283 | — | — | — | 0.595 | 0.856 | — | |
| VIDEOREPAIRBase Model=Wan2.1-1.3B2024.11 | 0.7262 | — | 0.5841 | — | 0.3837 | — | — | 0.5646 | |
| Wan + RAPOBackbone=Wan, Prompt Refinement Method=RAPO2026.04 | 0.721 | 0.279 | — | 0.309 | — | 0.672 | — | 0.495 | |
| VideoTetrisLayout Generator=LLM2025.01 | 0.7125 | 0.2066 | 0.5148 | 0.2204 | 0.2609 | — | — | — | |
| PixVerse-V3Mode=SOTA2025.12 | 0.706 | 0.062 | 0.598 | 0.287 | 0.607 | 0.872 | 0.831 | 0.566 | |
| Latte-RAPOBackbone=Latte2025.10 | 0.706 | 0.258 | — | — | — | 0.591 | 0.847 | — | |
| Gen-3 Alpha2025.01 | 0.7045 | 0.2078 | 0.5533 | 0.3111 | 0.2169 | — | — | — | |
| LaVie + SCMAPRBackbone=LaVie, Prompt Refinement Method=SCMAPR2026.04 | 0.704 | 0.273 | — | 0.285 | — | 0.64 | — | 0.476 | |
| T2V-turbo2024.11 | 0.7025 | — | 0.5492 | — | 0.2496 | — | — | 0.5004 | |
| WanBackbone=Wan, Prompt Refinement Method=User Input2026.04 | 0.694 | 0.263 | — | 0.269 | — | 0.591 | — | 0.454 | |
| Kling-1.0Model Type=Commercial2025.10 | 0.6931 | 0.0098 | 0.569 | 0.2562 | 0.4413 | 0.5787 | 0.7128 | 0.463 | |
| LaVie + RAPOBackbone=LaVie, Prompt Refinement Method=RAPO2026.04 | 0.692 | 0.267 | — | 0.243 | — | 0.635 | — | 0.46 | |
| LaVie-RAPOBackbone=LaVie2025.10 | 0.692 | 0.267 | — | — | — | 0.635 | 0.839 | — | |
| Dreamina 1.2Model Type=Commercial2025.10 | 0.6913 | 0.0051 | 0.5773 | 0.2361 | 0.438 | 0.5924 | 0.6824 | 0.4689 | |
| Dream Machine2025.01 | 0.69 | 0.2002 | 0.5387 | 0.2713 | 0.2109 | — | — | — | |
| Wan2.1-1.3B2024.11 | 0.687 | — | 0.569 | — | 0.3516 | — | — | 0.5358 | |
| VideoCrafter22024.11 | 0.6812 | — | 0.5214 | — | 0.2906 | — | — | 0.4977 | |
| VIDEOREPAIRBase Model=CogVideoX-5B2024.11 | 0.6725 | — | 0.5811 | — | 0.3034 | — | — | 0.519 | |
| T2V-Turbo-V2Model Type=Diffusion Unet based2025.10 | 0.6723 | 0.0127 | 0.5025 | 0.2556 | 0.3261 | 0.6087 | 0.6439 | 0.4317 | |
| Wan + Open-SoraBackbone=Wan, Prompt Refinement Method=Open-Sora2026.04 | 0.672 | 0.258 | — | 0.274 | — | 0.583 | — | 0.446 | |
| Open-Sora v1.22025.01 | 0.66 | 0.1714 | 0.5406 | 0.2388 | 0.2556 | — | — | — | |
| Pika 1.02025.01 | 0.6513 | 0.1744 | 0.5043 | 0.2221 | 0.2613 | — | — | — | |
| Vico2024.11 | 0.647 | — | 0.5425 | — | 0.2762 | — | — | 0.4886 | |
| LatteBackbone=Latte2025.10 | 0.633 | 0.227 | — | — | — | 0.476 | 0.792 | — | |
| CogVideoX-5B2024.11 | 0.622 | — | 0.4988 | — | 0.2228 | — | — | 0.4479 | |
| VideoTetrisModel Type=Diffusion Unet based2025.10 | 0.6211 | 0.0104 | 0.4832 | 0.2249 | 0.3467 | 0.4839 | 0.6578 | 0.4097 | |
| VideoTetris2024.11 | 0.6211 | — | 0.4832 | — | 0.3467 | — | — | 0.4836 | |
| LaVieBackbone=LaVie, Prompt Refinement Method=User Input2026.04 | 0.62 | 0.232 | — | 0.215 | — | 0.483 | — | 0.388 | |
| LaVieBackbone=LaVie2025.10 | 0.62 | 0.232 | — | — | — | 0.483 | 0.76 | — | |
| CogVideoX-5BModel Type=DiT based2025.10 | 0.6164 | 0.0219 | 0.5172 | 0.2658 | 0.3706 | 0.5333 | 0.6069 | 0.4189 | |
| Open-Sora-Plan v1.3.0Model Type=DiT based2025.10 | 0.6076 | 0.0119 | 0.5162 | 0.2377 | 0.2952 | 0.4524 | 0.4483 | 0.367 | |
| Gen-3Model Type=Commercial2025.10 | 0.598 | 0.0687 | 0.5194 | 0.2754 | 0.2306 | 0.5233 | 0.5906 | 0.4094 | |
| Latte-GPT4Backbone=Latte2025.10 | 0.598 | 0.21 | — | — | — | 0.405 | 0.688 | — | |
| Latte-PromptistBackbone=Latte2025.10 | 0.583 | 0.205 | — | — | — | 0.521 | 0.687 | — | |
| Latte-PAEBackbone=Latte2025.10 | 0.576 | 0.208 | — | — | — | 0.536 | 0.695 | — | |
| LaVie-PAEBackbone=LaVie2025.10 | 0.571 | 0.21 | — | — | — | 0.432 | 0.631 | — | |
| Show-1Model Type=Diffusion Unet based2025.10 | 0.567 | 0.0115 | 0.4544 | 0.2291 | 0.3086 | 0.3881 | 0.6244 | 0.3676 | |
| Show-12024.11 | 0.567 | — | 0.4544 | — | 0.3086 | — | — | 0.4433 | |
| Open-Sora 1.2Model Type=DiT based2025.10 | 0.5639 | 0.0189 | 0.5063 | 0.2468 | 0.3719 | 0.4839 | 0.5039 | 0.3851 | |
| LaVie-GPT4Backbone=LaVie2025.10 | 0.561 | 0.218 | — | — | — | 0.428 | 0.62 | — | |
| LVDLayout Generator=GPT-42025.01 | 0.5595 | 0.1499 | 0.5469 | 0.2699 | 0.0991 | — | — | — | |
| Pika-1.0Model Type=Commercial2025.10 | 0.5536 | 0.0128 | 0.465 | 0.2234 | 0.387 | 0.425 | 0.5198 | 0.3752 | |
| LaVie-PromptistBackbone=LaVie2025.10 | 0.552 | 0.203 | — | — | — | 0.412 | 0.615 | — | |
| Latte-Prompt RefinerBackbone=Latte2025.10 | 0.549 | 0.203 | — | — | — | 0.487 | 0.743 | — | |
| ModelScope + LVDModel Type=Diffusion Unet based, Enhancement=LVD2025.10 | 0.5439 | 0.0171 | 0.5405 | 0.2457 | 0.2008 | 0.3802 | 0.4502 | 0.3912 | |
| LaVie + Open-SoraBackbone=LaVie, Prompt Refinement Method=Open-Sora2026.04 | 0.532 | 0.214 | — | 0.226 | — | 0.47 | — | 0.361 | |
| LaVie-Prompt RefinerBackbone=LaVie2025.10 | 0.532 | 0.214 | — | — | — | 0.47 | 0.698 | — | |
| ModelScopeModel Type=Diffusion Unet based2025.10 | 0.5148 | 0.0161 | 0.4118 | 0.2408 | 0.1986 | 0.3639 | 0.4613 | 0.3468 | |
| ModelScope2024.11 | 0.5148 | — | 0.4118 | — | 0.1986 | — | — | 0.375 | |
| Latte2024.11 | 0.4713 | — | 0.434 | — | 0.232 | — | — | 0.3791 | |
| Open-Sora-Plan2024.11 | 0.4246 | — | 0.452 | — | 0.2331 | — | — | 0.3699 | |
| ZeroScope2024.11 | 0.4011 | — | 0.4287 | — | 0.2408 | — | — | 0.3568 | |
| Base Model-720pNFE=50×2, Speedup=1.0×, Backbone=Wan2.1-T2V-14B2026.03 | — | — | — | — | — | — | — | 54.17 | |
| DMD2NFE=6, Speedup=16.7×, Backbone=Wan2.1-T2V-14B2026.03 | — | — | — | — | — | — | — | 52.81 | |
| RMDNFE=3 + 3, Speedup=25.6×, Backbone=Wan2.1-T2V-14B2026.03 | — | — | — | — | — | — | — | 54 | |
| TDMNFE=6, Speedup=16.7×, Backbone=Wan2.1-T2V-14B2026.03 | — | — | — | — | — | — | — | 52.27 |