# Accuracy on LongVideoBench (Long-video understanding)

Source: [Accuracy on LongVideoBench (Long-video understanding)](<https://www.sota2.com/research/sota/accuracy-on-longvideobench-long-video-understanding>)

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

Data updated: 2026-06-23T09:34:29.801Z

Evaluation of long-form video understanding capabilities on the LongVideoBench dataset, reporting overall performance 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 | Long-video understanding |
| Dataset (including any stated split) | LongVideoBench average |
| Metric | Accuracy |
| 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

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

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| GPT-4o | 66.7 | Not recorded | 6a3a1354206542251d7c7323 |
| HPP | 66.4 | Not recorded | 6a3a1354206542251d7c7323 |
| VideoRAG | 65.4 | Not recorded | 6a3a1354206542251d7c7323 |
| LLaVA-Video | 64.9 | Not recorded | 6a3a1354206542251d7c7323 |
| Aria | 64.2 | Not recorded | 6a3a1354206542251d7c7323 |
| Gemini 1.5 Pro | 64 | Not recorded | 6a3a1354206542251d7c7323 |
| InternVL-2.5 | 63.6 | Not recorded | 6a3a1354206542251d7c7323 |
| LLaVA-OneVision | 63.2 | Not recorded | 6a3a1354206542251d7c7323 |
| LLaVA-Video | 58.2 | Not recorded | 6a3a1354206542251d7c7323 |
| Claude 3.5 Opus | 57.3 | Not recorded | 6a3a1354206542251d7c7323 |
| LongVA | 51.3 | Not recorded | 6a3a1354206542251d7c7323 |
| VideoTree | 41.3 | Not recorded | 6a3a1354206542251d7c7323 |
| ShareGPT4Video | 39.7 | Not recorded | 6a3a1354206542251d7c7323 |
| VideoChat2 | 39.3 | Not recorded | 6a3a1354206542251d7c7323 |

### GPT-4o

Accuracy: 66.7. Result ID: 6a3a5315fbc373592c35e4a2.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75681e209131cea67a98.md>)

Recorded evaluation configuration: {"Model":"GPT-4o"}. This record does not assert independent reproduction.

### HPP

Accuracy: 66.4. Result ID: 6a3a5315fbc373592c35e4ae.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

Recorded evaluation configuration: {"Model":"Ours HPP"}. This record does not assert independent reproduction.

### VideoRAG

Accuracy: 65.4. Result ID: 6a3a5315fbc373592c35e4ad.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa764b1e209131cea6c370.md>)

[Implementation](<https://github.com/Leon1207/Video-RAG-master>)

Recorded evaluation configuration: {"Parameters":"72B"}. This record does not assert independent reproduction.

### LLaVA-Video

Accuracy: 64.9. Result ID: 6a3a5315fbc373592c35e4a9.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a561e209131cea82934.md>)

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

Recorded evaluation configuration: {"Parameters":"72B"}. This record does not assert independent reproduction.

### Aria

Accuracy: 64.2. Result ID: 6a3a5315fbc373592c35e4a8.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76921e209131cea6dd17.md>)

[Implementation](<https://github.com/rhymes-ai/Aria>)

Recorded evaluation configuration: {"Parameters":"28B"}. This record does not assert independent reproduction.

### Gemini 1.5 Pro

Accuracy: 64. Result ID: 6a3a5315fbc373592c35e4a3.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76361e209131cea6be11.md>)

Recorded evaluation configuration: {"Model":"Gemini 1.5 Pro"}. This record does not assert independent reproduction.

### InternVL-2.5

Accuracy: 63.6. Result ID: 6a3a5315fbc373592c35e4aa.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

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

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

Recorded evaluation configuration: {"Parameters":"78B"}. This record does not assert independent reproduction.

### LLaVA-OneVision

Accuracy: 63.2. Result ID: 6a3a5315fbc373592c35e4ab.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

Recorded evaluation configuration: {"Parameters":"72B"}. This record does not assert independent reproduction.

### LLaVA-Video

Accuracy: 58.2. Result ID: 6a3a5315fbc373592c35e4a7.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a561e209131cea82934.md>)

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

Recorded evaluation configuration: {"Parameters":"7B"}. This record does not assert independent reproduction.

### Claude 3.5 Opus

Accuracy: 57.3. Result ID: 6a3a5315fbc373592c35e4a1.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

Recorded evaluation configuration: {"Model":"Claude 3.5 Opus"}. This record does not assert independent reproduction.

### LongVA

Accuracy: 51.3. Result ID: 6a3a5315fbc373592c35e4a6.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73ea1e209131cea5edfe.md>)

[Implementation](<https://github.com/EvolvingLMMs-Lab/LongVA>)

Recorded evaluation configuration: {"Parameters":"7B"}. This record does not assert independent reproduction.

### VideoTree

Accuracy: 41.3. Result ID: 6a3a5315fbc373592c35e4ac.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa777a1e209131cea73736.md>)

[Implementation](<https://videotree2024.github.io/>)

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

### ShareGPT4Video

Accuracy: 39.7. Result ID: 6a3a5315fbc373592c35e4a4.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa738e1e209131cea5d310.md>)

[Implementation](<https://sharegpt4video.github.io/>)

Recorded evaluation configuration: {"Parameters":"8B"}. This record does not assert independent reproduction.

### VideoChat2

Accuracy: 39.3. Result ID: 6a3a5315fbc373592c35e4a5.

[Reporting paper](<https://www.sota2.com/research/paper/6a3a1354206542251d7c7323.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73411e209131cea5b4b7.md>)

Recorded evaluation configuration: {"Parameters":"7B"}. This record does not assert independent reproduction.

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