# Video Understanding on SEED-Bench Video Understanding

Source: [Video Understanding on SEED-Bench Video Understanding](<https://www.sota2.com/research/sota/video-understanding-on-seed-bench-video-understanding>)

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

Data updated: 2026-02-26T05:32:52.170Z

Evaluation of general video understanding capabilities using the SEED-Bench Video Understanding dataset, measured by Accuracy.

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 Understanding |
| Dataset (including any stated split) | SEED-Bench Video Understanding |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 33 of 33 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/video-understanding-on-seed-bench-video-understanding.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Baseline | 74.12 | Not recorded | 6993f62a9ebc499c97409b88 |
| IDPruner | 72.68 | Not recorded | 6993f62a9ebc499c97409b88 |
| SCOPE | 72.63 | Not recorded | 6993f62a9ebc499c97409b88 |
| DivPrune | 72.11 | Not recorded | 6993f62a9ebc499c97409b88 |
| VisionZip | 72.11 | Not recorded | 6993f62a9ebc499c97409b88 |
| HiPrune | 71.97 | Not recorded | 6993f62a9ebc499c97409b88 |
| VisPruner | 71.89 | Not recorded | 6993f62a9ebc499c97409b88 |
| InternVL-2.5-26B (Reproduction) | 71.41 | Not recorded | 693a50186cfd488761a54ac3 |
| InternVL-2.5-26B + LUVC | 71.28 | Not recorded | 693a50186cfd488761a54ac3 |
| DART | 71.27 | Not recorded | 6993f62a9ebc499c97409b88 |
| InternVL-2.5-26B + VTW | 70.96 | Not recorded | 693a50186cfd488761a54ac3 |
| VisionSelector | 70.75 | Not recorded | 6993f62a9ebc499c97409b88 |
| InternVL-2.5-26B + PACT | 70.24 | Not recorded | 693a50186cfd488761a54ac3 |
| FastV | 69.22 | Not recorded | 6993f62a9ebc499c97409b88 |
| InternVL-2.5-8B (Reproduction) | 66.41 | Not recorded | 693a50186cfd488761a54ac3 |
| InternVL-2.5-8B + LUVC | 65.85 | Not recorded | 693a50186cfd488761a54ac3 |
| InternVL-2.5-8B + VTW | 65.45 | Not recorded | 693a50186cfd488761a54ac3 |
| InternVL-2.5-8B + PACT | 65.18 | Not recorded | 693a50186cfd488761a54ac3 |
| Dream-VL | 58.1 | Not recorded | 69535c8653fdd5c36173d9b8 |
| LlavaOV-7B + LUVC | 57.36 | Not recorded | 693a50186cfd488761a54ac3 |
| MAmmoTH-VL | 57.1 | Not recorded | 69535c8653fdd5c36173d9b8 |
| LLaVA-OV | 56.9 | Not recorded | 69535c8653fdd5c36173d9b8 |
| LlavaOV-7B | 56.9 | Not recorded | 693a50186cfd488761a54ac3 |
| LlavaOV-7B + PACT | 56.8 | Not recorded | 693a50186cfd488761a54ac3 |
| LlavaOV-7B + FastV | 56.72 | Not recorded | 693a50186cfd488761a54ac3 |
| LLaDA-V | 53.7 | Not recorded | 69535c8653fdd5c36173d9b8 |
| LlavaOV-7B + VTW | 48.44 | Not recorded | 693a50186cfd488761a54ac3 |
| LlavaOV-0.5B + LUVC | 44.96 | Not recorded | 693a50186cfd488761a54ac3 |
| LlavaOV-0.5B + PACT | 44.8 | Not recorded | 693a50186cfd488761a54ac3 |
| LlavaOV-0.5B | 44.2 | Not recorded | 693a50186cfd488761a54ac3 |
| LlavaOV-0.5B (Reproduction) | 44.16 | Not recorded | 693a50186cfd488761a54ac3 |
| LlavaOV-0.5B + VTW | 43.86 | Not recorded | 693a50186cfd488761a54ac3 |
| LlavaOV-0.5B + FastV | 43.55 | Not recorded | 693a50186cfd488761a54ac3 |

### Baseline

Accuracy: 74.12. Result ID: 699a74d65698665ec4fddb58.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### IDPruner

Accuracy: 72.68. Result ID: 699a74d65698665ec4fddb61.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### SCOPE

Accuracy: 72.63. Result ID: 699a74d65698665ec4fddb60.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69019f753ef7ccd74e5e3998.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### DivPrune

Accuracy: 72.11. Result ID: 699a74d65698665ec4fddb5d.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77c41e209131cea7532a.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### VisionZip

Accuracy: 72.11. Result ID: 699a74d65698665ec4fddb5a.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa760f1e209131cea6b33b.md>)

[Implementation](<https://github.com/dvlab-research/VisionZip>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### HiPrune

Accuracy: 71.97. Result ID: 699a74d65698665ec4fddb5b.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a6e1e209131cea832ae.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### VisPruner

Accuracy: 71.89. Result ID: 699a74d65698665ec4fddb5f.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78741e209131cea786bf.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### InternVL-2.5-26B (Reproduction)

Accuracy: 71.41. Result ID: 69986c705698665ec4cc2840.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bab1e209131cea89c7f.md>)

[Implementation](<https://github.com/OpenGVLab/InternVL>)

Recorded evaluation configuration: {"Backbone":"InternVL-2.5-26B","Red. Ratio":"0.00%","Latency":"2.9142","Speedup":"-"}. This record does not assert independent reproduction.

### InternVL-2.5-26B + LUVC

Accuracy: 71.28. Result ID: 69986c705698665ec4cc2843.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

Recorded evaluation configuration: {"Backbone":"InternVL-2.5-26B","Red. Ratio":"61.32%","Latency":"1.2255","Speedup":"2.38x"}. This record does not assert independent reproduction.

### DART

Accuracy: 71.27. Result ID: 699a74d65698665ec4fddb5e.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79461e209131cea7cfd7.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### InternVL-2.5-26B + VTW

Accuracy: 70.96. Result ID: 69986c705698665ec4cc2841.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6eb9c.md>)

Recorded evaluation configuration: {"Backbone":"InternVL-2.5-26B","Red. Ratio":"66.67%","Latency":"1.2586","Speedup":"2.32x"}. This record does not assert independent reproduction.

### VisionSelector

Accuracy: 70.75. Result ID: 699a74d65698665ec4fddb5c.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c541e209131cea8d99d.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### InternVL-2.5-26B + PACT

Accuracy: 70.24. Result ID: 69986c705698665ec4cc2842.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77ec1e209131cea76055.md>)

Recorded evaluation configuration: {"Backbone":"InternVL-2.5-26B","Red. Ratio":"59.61%","Latency":"1.3548","Speedup":"2.15x"}. This record does not assert independent reproduction.

### FastV

Accuracy: 69.22. Result ID: 699a74d65698665ec4fddb59.

[Reporting paper](<https://www.sota2.com/research/paper/6993f62a9ebc499c97409b88.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74ae1e209131cea632b7.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B-Instruct","Token Retention":"25%"}. This record does not assert independent reproduction.

### InternVL-2.5-8B (Reproduction)

Accuracy: 66.41. Result ID: 69986c705698665ec4cc283c.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bab1e209131cea89c7f.md>)

[Implementation](<https://github.com/OpenGVLab/InternVL>)

Recorded evaluation configuration: {"Backbone":"InternVL-2.5-8B","Red. Ratio":"0.00%","Latency":"1.2480","Speedup":"-"}. This record does not assert independent reproduction.

### InternVL-2.5-8B + LUVC

Accuracy: 65.85. Result ID: 69986c705698665ec4cc283f.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

Recorded evaluation configuration: {"Backbone":"InternVL-2.5-8B","Red. Ratio":"62.12%","Latency":"0.6293","Speedup":"1.98x"}. This record does not assert independent reproduction.

### InternVL-2.5-8B + VTW

Accuracy: 65.45. Result ID: 69986c705698665ec4cc283d.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6eb9c.md>)

Recorded evaluation configuration: {"Backbone":"InternVL-2.5-8B","Red. Ratio":"56.25%","Latency":"0.6482","Speedup":"1.93x"}. This record does not assert independent reproduction.

### InternVL-2.5-8B + PACT

Accuracy: 65.18. Result ID: 69986c705698665ec4cc283e.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77ec1e209131cea76055.md>)

Recorded evaluation configuration: {"Backbone":"InternVL-2.5-8B","Red. Ratio":"55.41%","Latency":"0.6949","Speedup":"1.80x"}. This record does not assert independent reproduction.

### Dream-VL

Accuracy: 58.1. Result ID: 699806af5698665ec4603711.

[Reporting paper](<https://www.sota2.com/research/paper/69535c8653fdd5c36173d9b8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69535c8653fdd5c36173d9b8.md>)

[Implementation](<https://github.com/DreamLM/Dream-VLX>)

Recorded evaluation configuration: {"LLM Backbone":"Dream 7B","Open Data":true}. This record does not assert independent reproduction.

### LlavaOV-7B + LUVC

Accuracy: 57.36. Result ID: 69986c705698665ec4cc283b.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-7B","Red. Ratio":"62.32%","Latency":"0.2583","Speedup":"2.01x"}. This record does not assert independent reproduction.

### MAmmoTH-VL

Accuracy: 57.1. Result ID: 699806af5698665ec460370f.

[Reporting paper](<https://www.sota2.com/research/paper/69535c8653fdd5c36173d9b8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79201e209131cea7c422.md>)

[Implementation](<https://mammoth-vl.github.io/>)

Recorded evaluation configuration: {"LLM Backbone":"Qwen2.5-7B","Open Data":true}. This record does not assert independent reproduction.

### LLaVA-OV

Accuracy: 56.9. Result ID: 699806af5698665ec460370e.

[Reporting paper](<https://www.sota2.com/research/paper/69535c8653fdd5c36173d9b8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.md>)

[Implementation](<https://llava-vl.github.io/blog/llava-onevision>)

Recorded evaluation configuration: {"LLM Backbone":"Qwen2-7B","Open Data":true}. This record does not assert independent reproduction.

### LlavaOV-7B

Accuracy: 56.9. Result ID: 69986c705698665ec4cc2837.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.md>)

[Implementation](<https://llava-vl.github.io/blog/llava-onevision>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-7B","Red. Ratio":"0.00%","Latency":"0.5194","Speedup":"-"}. This record does not assert independent reproduction.

### LlavaOV-7B + PACT

Accuracy: 56.8. Result ID: 69986c705698665ec4cc283a.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77ec1e209131cea76055.md>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-7B","Red. Ratio":"55.88%","Latency":"0.2711","Speedup":"1.92x"}. This record does not assert independent reproduction.

### LlavaOV-7B + FastV

Accuracy: 56.72. Result ID: 69986c705698665ec4cc2838.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74ae1e209131cea632b7.md>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-7B","Red. Ratio":"42.86%","Latency":"0.2693","Speedup":"1.93x"}. This record does not assert independent reproduction.

### LLaDA-V

Accuracy: 53.7. Result ID: 699806af5698665ec4603710.

[Reporting paper](<https://www.sota2.com/research/paper/69535c8653fdd5c36173d9b8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79101e209131cea7bebe.md>)

Recorded evaluation configuration: {"LLM Backbone":"LLaDA 8B","Open Data":true}. This record does not assert independent reproduction.

### LlavaOV-7B + VTW

Accuracy: 48.44. Result ID: 69986c705698665ec4cc2839.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6eb9c.md>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-7B","Red. Ratio":"57.14%","Latency":"0.2524","Speedup":"2.06x"}. This record does not assert independent reproduction.

### LlavaOV-0.5B + LUVC

Accuracy: 44.96. Result ID: 69986c705698665ec4cc2836.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-0.5B","Red. Ratio":"58.46%","Latency":"0.0610","Speedup":"1.27x"}. This record does not assert independent reproduction.

### LlavaOV-0.5B + PACT

Accuracy: 44.8. Result ID: 69986c705698665ec4cc2835.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77ec1e209131cea76055.md>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-0.5B","Red. Ratio":"48.05%","Latency":"0.0642","Speedup":"1.20x"}. This record does not assert independent reproduction.

### LlavaOV-0.5B

Accuracy: 44.2. Result ID: 69986c705698665ec4cc2831.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.md>)

[Implementation](<https://llava-vl.github.io/blog/llava-onevision>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-0.5B","Red. Ratio":"0.00%","Latency":"0.0772","Speedup":"-"}. This record does not assert independent reproduction.

### LlavaOV-0.5B (Reproduction)

Accuracy: 44.16. Result ID: 69986c705698665ec4cc2832.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.md>)

[Implementation](<https://llava-vl.github.io/blog/llava-onevision>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-0.5B","Red. Ratio":"0.00%","Latency":"0.0772","Speedup":"-"}. This record does not assert independent reproduction.

### LlavaOV-0.5B + VTW

Accuracy: 43.86. Result ID: 69986c705698665ec4cc2834.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76b91e209131cea6eb9c.md>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-0.5B","Red. Ratio":"54.17%","Latency":"0.0596","Speedup":"1.30x"}. This record does not assert independent reproduction.

### LlavaOV-0.5B + FastV

Accuracy: 43.55. Result ID: 69986c705698665ec4cc2833.

[Reporting paper](<https://www.sota2.com/research/paper/693a50186cfd488761a54ac3.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74ae1e209131cea632b7.md>)

Recorded evaluation configuration: {"Backbone":"LlavaOV-0.5B","Red. Ratio":"41.67%","Latency":"0.0711","Speedup":"1.00x"}. This record does not assert independent reproduction.

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