# Long-term Video Consistency on VBench

Source: [Long-term Video Consistency on VBench](<https://www.sota2.com/research/sota/long-term-video-consistency-on-vbench>)

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

Data updated: 2026-06-19T08:20:40.968Z

Evaluation of model performance on maintaining Long-term Video Consistency using the VBench dataset, assessed via Sub., Back., and Arcf. metrics.

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 | Long-term Video Consistency |
| Dataset (including any stated split) | VBench |
| Metric | Sub. Consistency Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| 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

- [Sub. Consistency Score](<https://www.sota2.com/research/sota/long-term-video-consistency-on-vbench.md?metric=Sub.%20Consistency%20Score>): Higher is better; unit: PERCENT.

- [Back. Consistency Score](<https://www.sota2.com/research/sota/long-term-video-consistency-on-vbench.md?metric=Back.%20Consistency%20Score>): Higher is better; unit: PERCENT.

- [Arcf. Consistency Score](<https://www.sota2.com/research/sota/long-term-video-consistency-on-vbench.md?metric=Arcf.%20Consistency%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Sub. Consistency Score | Result date | Report paper |
| --- | --- | --- | --- |
| Gloria | 96 | Not recorded | 69cca6a90636e5bf4afcbcbb |
| InfiniteTalk | 95.2 | Not recorded | 69cca6a90636e5bf4afcbcbb |
| HunyuanA. | 91.4 | Not recorded | 69cca6a90636e5bf4afcbcbb |
| WanS2V | 87.2 | Not recorded | 69cca6a90636e5bf4afcbcbb |
| MagicDrive-V2 | 79.66 | Not recorded | 6a34cd4a206542251d7c6179 |
| FrozenDrive | 77.24 | Not recorded | 6a34cd4a206542251d7c6179 |
| SD + CA | 75.86 | Not recorded | 6a34cd4a206542251d7c6179 |
| DriveArena | 75.73 | Not recorded | 6a34cd4a206542251d7c6179 |

### Gloria

Sub. Consistency Score: 96. Result ID: 69ccdcbe47f34ce924b6aab3.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6a90636e5bf4afcbcbb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cca6a90636e5bf4afcbcbb.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### InfiniteTalk

Sub. Consistency Score: 95.2. Result ID: 69ccdcbe47f34ce924b6aab2.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6a90636e5bf4afcbcbb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ac11e209131cea85021.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### HunyuanA.

Sub. Consistency Score: 91.4. Result ID: 69ccdcbe47f34ce924b6aab0.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6a90636e5bf4afcbcbb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79161e209131cea7c113.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### WanS2V

Sub. Consistency Score: 87.2. Result ID: 69ccdcbe47f34ce924b6aab1.

[Reporting paper](<https://www.sota2.com/research/paper/69cca6a90636e5bf4afcbcbb.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ae61e209131cea85cba.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### MagicDrive-V2

Sub. Consistency Score: 79.66. Result ID: 6a34fbd1fbc373592c3248cc.

[Reporting paper](<https://www.sota2.com/research/paper/6a34cd4a206542251d7c6179.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a361e209131cea81d0b.md>)

Recorded evaluation configuration: {"model\_base":"video diffusion model"}. This record does not assert independent reproduction.

### FrozenDrive

Sub. Consistency Score: 77.24. Result ID: 6a34fbd1fbc373592c3248cd.

[Reporting paper](<https://www.sota2.com/research/paper/6a34cd4a206542251d7c6179.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a34cd4a206542251d7c6179.md>)

Recorded evaluation configuration: {"training\_mode":"zero-shot","parameters":"parameter-free"}. This record does not assert independent reproduction.

### SD + CA

Sub. Consistency Score: 75.86. Result ID: 6a34fbd1fbc373592c3248ca.

[Reporting paper](<https://www.sota2.com/research/paper/6a34cd4a206542251d7c6179.md>)

Recorded evaluation configuration: {"training\_mode":"fine-tuned","mechanism":"learnable cross-attention layers"}. This record does not assert independent reproduction.

### DriveArena

Sub. Consistency Score: 75.73. Result ID: 6a34fbd1fbc373592c3248cb.

[Reporting paper](<https://www.sota2.com/research/paper/6a34cd4a206542251d7c6179.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa744f1e209131cea60f2a.md>)

Recorded evaluation configuration: {"training\_mode":"fine-tuned"}. This record does not assert independent reproduction.

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