# Video Generation on VBench (Total, Quality, Semantic, HPSv3, MQ, Throughput)

Source: [Video Generation on VBench (Total, Quality, Semantic, HPSv3, MQ, Throughput)](<https://www.sota2.com/research/sota/video-generation-on-vbench-total-quality-semantic-hpsv3-mq-throughput>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-03-19T07:52:21.041Z

Evaluation of Video Generation models on the VBench dataset, reporting scores across Total performance, Quality, Semantic alignment, HPSv3, MQ, and inference Throughput.

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Video Generation |
| Dataset (including any stated split) | VBench |
| Metric | Total Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 14 of 14 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Total Score](<https://www.sota2.com/research/sota/video-generation-on-vbench-total-quality-semantic-hpsv3-mq-throughput.md?metric=Total%20Score>): Higher is better; unit: PERCENT.

- [Quality Score](<https://www.sota2.com/research/sota/video-generation-on-vbench-total-quality-semantic-hpsv3-mq-throughput.md?metric=Quality%20Score>): Higher is better; unit: PERCENT.

- [Semantic Alignment Score](<https://www.sota2.com/research/sota/video-generation-on-vbench-total-quality-semantic-hpsv3-mq-throughput.md?metric=Semantic%20Alignment%20Score>): Higher is better; unit: PERCENT.

- [HPSv3 Score](<https://www.sota2.com/research/sota/video-generation-on-vbench-total-quality-semantic-hpsv3-mq-throughput.md?metric=HPSv3%20Score>): Higher is better; unit: SCALAR.

- [MQ Score](<https://www.sota2.com/research/sota/video-generation-on-vbench-total-quality-semantic-hpsv3-mq-throughput.md?metric=MQ%20Score>): Higher is better; unit: SCALAR.

- [Throughput](<https://www.sota2.com/research/sota/video-generation-on-vbench-total-quality-semantic-hpsv3-mq-throughput.md?metric=Throughput>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Total Score | Result date | Report paper |
| --- | --- | --- | --- |
| LongLive + Astrolabe | 84.93 | Not recorded | 69bb82eb0636e5bf4afc188b |
| Causal Forcing + Astrolabe | 84.46 | Not recorded | 69bb82eb0636e5bf4afc188b |
| Wan2.1 | 84.26 | Not recorded | 69bb82eb0636e5bf4afc188b |
| Reward Forcing | 84.13 | Not recorded | 69bb82eb0636e5bf4afc188b |
| Causal Forcing | 84.04 | Not recorded | 69bb82eb0636e5bf4afc188b |
| Self-Forcing + Astrolabe | 83.79 | Not recorded | 69bb82eb0636e5bf4afc188b |
| Self-Forcing | 83.74 | Not recorded | 69bb82eb0636e5bf4afc188b |
| LongLive | 83.22 | Not recorded | 69bb82eb0636e5bf4afc188b |
| SkyReels-V2 | 82.67 | Not recorded | 69bb82eb0636e5bf4afc188b |
| PyramidFlow | 81.72 | Not recorded | 69bb82eb0636e5bf4afc188b |
| CausVid | 81.2 | Not recorded | 69bb82eb0636e5bf4afc188b |
| NOVA | 80.12 | Not recorded | 69bb82eb0636e5bf4afc188b |
| LTX-Video | 80 | Not recorded | 69bb82eb0636e5bf4afc188b |
| MAGI-1 | 79.18 | Not recorded | 69bb82eb0636e5bf4afc188b |

### LongLive + Astrolabe

Total Score: 84.93. Result ID: 69bbaadf47f34ce924a546dc.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Implementation](<https://github.com/franklinz233/Astrolabe>)

Recorded evaluation configuration: {"Model Type":"Distilled AR Models"}. This record does not assert independent reproduction.

### Causal Forcing + Astrolabe

Total Score: 84.46. Result ID: 69bbaadf47f34ce924a546de.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Implementation](<https://github.com/franklinz233/Astrolabe>)

Recorded evaluation configuration: {"Model Type":"Distilled AR Models"}. This record does not assert independent reproduction.

### Wan2.1

Total Score: 84.26. Result ID: 69bbaadf47f34ce924a546d2.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78241e209131cea76f7d.md>)

Recorded evaluation configuration: {"Model Type":"Diffusion Models"}. This record does not assert independent reproduction.

### Reward Forcing

Total Score: 84.13. Result ID: 69bbaadf47f34ce924a546d8.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693266f9d6b0dd33429341ff.md>)

[Implementation](<https://github.com/JaydenLyh/Reward-Forcing>)

Recorded evaluation configuration: {"Model Type":"Distilled AR Models"}. This record does not assert independent reproduction.

### Causal Forcing

Total Score: 84.04. Result ID: 69bbaadf47f34ce924a546dd.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698181ada8b01e76cc8e27d4.md>)

[Implementation](<https://github.com/thu-ml/Causal-Forcing>)

Recorded evaluation configuration: {"Model Type":"Distilled AR Models"}. This record does not assert independent reproduction.

### Self-Forcing + Astrolabe

Total Score: 83.79. Result ID: 69bbaadf47f34ce924a546da.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Implementation](<https://github.com/franklinz233/Astrolabe>)

Recorded evaluation configuration: {"Model Type":"Distilled AR Models"}. This record does not assert independent reproduction.

### Self-Forcing

Total Score: 83.74. Result ID: 69bbaadf47f34ce924a546d9.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa793a1e209131cea7cb5a.md>)

Recorded evaluation configuration: {"Model Type":"Distilled AR Models"}. This record does not assert independent reproduction.

### LongLive

Total Score: 83.22. Result ID: 69bbaadf47f34ce924a546db.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c251e209131cea8cb13.md>)

Recorded evaluation configuration: {"Model Type":"Distilled AR Models"}. This record does not assert independent reproduction.

### SkyReels-V2

Total Score: 82.67. Result ID: 69bbaadf47f34ce924a546d3.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78151e209131cea76abc.md>)

Recorded evaluation configuration: {"Model Type":"AR Models"}. This record does not assert independent reproduction.

### PyramidFlow

Total Score: 81.72. Result ID: 69bbaadf47f34ce924a546d6.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa777b1e209131cea737d3.md>)

Recorded evaluation configuration: {"Model Type":"AR Models"}. This record does not assert independent reproduction.

### CausVid

Total Score: 81.2. Result ID: 69bbaadf47f34ce924a546d7.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b831e209131cea88cde.md>)

Recorded evaluation configuration: {"Model Type":"Distilled AR Models"}. This record does not assert independent reproduction.

### NOVA

Total Score: 80.12. Result ID: 69bbaadf47f34ce924a546d5.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77401e209131cea71feb.md>)

Recorded evaluation configuration: {"Model Type":"AR Models"}. This record does not assert independent reproduction.

### LTX-Video

Total Score: 80. Result ID: 69bbaadf47f34ce924a546d1.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa766c1e209131cea6cdee.md>)

Recorded evaluation configuration: {"Model Type":"Diffusion Models"}. This record does not assert independent reproduction.

### MAGI-1

Total Score: 79.18. Result ID: 69bbaadf47f34ce924a546d4.

[Reporting paper](<https://www.sota2.com/research/paper/69bb82eb0636e5bf4afc188b.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78ab1e209131cea79c32.md>)

Recorded evaluation configuration: {"Model Type":"AR Models"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
