# Long Video Question Answering on Video-MME Long without Subtitles

Source: [Long Video Question Answering on Video-MME Long without Subtitles](<https://www.sota2.com/research/sota/long-video-question-answering-on-video-mme-long-without-subtitles>)

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

Data updated: 2026-04-08T08:19:05.363Z

Evaluation of model performance on the Long subset of the Video-MME dataset, specifically excluding samples containing subtitles, focusing on Long-video Question Answering accuracy and gain 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-video Question Answering |
| Dataset (including any stated split) | Video-MME Long without Subtitles |
| Metric | Overall Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 16 of 16 |
| 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

- [Overall Accuracy](<https://www.sota2.com/research/sota/long-video-question-answering-on-video-mme-long-without-subtitles.md?metric=Overall%20Accuracy>): Higher is better; unit: PERCENT.

- [Gain](<https://www.sota2.com/research/sota/long-video-question-answering-on-video-mme-long-without-subtitles.md?metric=Gain>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Overall Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Video-RAG | 62.3 | Not recorded | 69d5e0ef7d0172140ec75925 |
| VideoStir | 62.1 | Not recorded | 69d5e0ef7d0172140ec75925 |
| TV-RAG | 61.8 | Not recorded | 69d5e0ef7d0172140ec75925 |
| LLaVA-Video | 61.5 | Not recorded | 69d5e0ef7d0172140ec75925 |
| Video-RAG | 59.6 | Not recorded | 69d5e0ef7d0172140ec75925 |
| VideoStir | 59.4 | Not recorded | 69d5e0ef7d0172140ec75925 |
| TV-RAG | 59 | Not recorded | 69d5e0ef7d0172140ec75925 |
| Aria | 58.8 | Not recorded | 69d5e0ef7d0172140ec75925 |
| VideoStir | 51 | Not recorded | 69d5e0ef7d0172140ec75925 |
| TV-RAG | 50.9 | Not recorded | 69d5e0ef7d0172140ec75925 |
| Video-RAG | 50.6 | Not recorded | 69d5e0ef7d0172140ec75925 |
| mPLUG-Owl3 | 50.1 | Not recorded | 69d5e0ef7d0172140ec75925 |
| VideoStir | 46.9 | Not recorded | 69d5e0ef7d0172140ec75925 |
| Video-RAG | 46.7 | Not recorded | 69d5e0ef7d0172140ec75925 |
| TV-RAG | 46.3 | Not recorded | 69d5e0ef7d0172140ec75925 |
| InternVL-1.5 | 45.6 | Not recorded | 69d5e0ef7d0172140ec75925 |

### Video-RAG

Overall Accuracy: 62.3. Result ID: 69d60f7147f34ce924be8a7e.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

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

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

Recorded evaluation configuration: {"Backbone":"72B","Native Input":"false"}. This record does not assert independent reproduction.

### VideoStir

Overall Accuracy: 62.1. Result ID: 69d60f7147f34ce924be8a80.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

Recorded evaluation configuration: {"Backbone":"72B","Native Input":"true"}. This record does not assert independent reproduction.

### TV-RAG

Overall Accuracy: 61.8. Result ID: 69d60f7147f34ce924be8a7f.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69535c9953fdd5c36173db5c.md>)

Recorded evaluation configuration: {"Backbone":"72B","Native Input":"false"}. This record does not assert independent reproduction.

### LLaVA-Video

Overall Accuracy: 61.5. Result ID: 69d60f7147f34ce924be8a7d.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

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

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

Recorded evaluation configuration: {"Backbone":"72B","Native Input":"true"}. This record does not assert independent reproduction.

### Video-RAG

Overall Accuracy: 59.6. Result ID: 69d60f7147f34ce924be8a76.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

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

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

Recorded evaluation configuration: {"Backbone":"25B","Native Input":"false"}. This record does not assert independent reproduction.

### VideoStir

Overall Accuracy: 59.4. Result ID: 69d60f7147f34ce924be8a78.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

Recorded evaluation configuration: {"Backbone":"25B","Native Input":"true"}. This record does not assert independent reproduction.

### TV-RAG

Overall Accuracy: 59. Result ID: 69d60f7147f34ce924be8a77.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69535c9953fdd5c36173db5c.md>)

Recorded evaluation configuration: {"Backbone":"25B","Native Input":"false"}. This record does not assert independent reproduction.

### Aria

Overall Accuracy: 58.8. Result ID: 69d60f7147f34ce924be8a75.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

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

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

Recorded evaluation configuration: {"Backbone":"25B","Native Input":"true"}. This record does not assert independent reproduction.

### VideoStir

Overall Accuracy: 51. Result ID: 69d60f7147f34ce924be8a74.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

Recorded evaluation configuration: {"Backbone":"8B","Native Input":"true"}. This record does not assert independent reproduction.

### TV-RAG

Overall Accuracy: 50.9. Result ID: 69d60f7147f34ce924be8a73.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69535c9953fdd5c36173db5c.md>)

Recorded evaluation configuration: {"Backbone":"8B","Native Input":"false"}. This record does not assert independent reproduction.

### Video-RAG

Overall Accuracy: 50.6. Result ID: 69d60f7147f34ce924be8a72.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

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

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

Recorded evaluation configuration: {"Backbone":"8B","Native Input":"false"}. This record does not assert independent reproduction.

### mPLUG-Owl3

Overall Accuracy: 50.1. Result ID: 69d60f7147f34ce924be8a71.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa746c1e209131cea61a16.md>)

[Implementation](<https://github.com/X-PLUG/mPLUG-Owl>)

Recorded evaluation configuration: {"Backbone":"8B","Native Input":"true"}. This record does not assert independent reproduction.

### VideoStir

Overall Accuracy: 46.9. Result ID: 69d60f7147f34ce924be8a7c.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

Recorded evaluation configuration: {"Backbone":"26B","Native Input":"true"}. This record does not assert independent reproduction.

### Video-RAG

Overall Accuracy: 46.7. Result ID: 69d60f7147f34ce924be8a7a.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

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

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

Recorded evaluation configuration: {"Backbone":"26B","Native Input":"false"}. This record does not assert independent reproduction.

### TV-RAG

Overall Accuracy: 46.3. Result ID: 69d60f7147f34ce924be8a7b.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69535c9953fdd5c36173db5c.md>)

Recorded evaluation configuration: {"Backbone":"26B","Native Input":"false"}. This record does not assert independent reproduction.

### InternVL-1.5

Overall Accuracy: 45.6. Result ID: 69d60f7147f34ce924be8a79.

[Reporting paper](<https://www.sota2.com/research/paper/69d5e0ef7d0172140ec75925.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73081e209131cea59f8b.md>)

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

Recorded evaluation configuration: {"Backbone":"26B","Native Input":"true"}. This record does not assert independent reproduction.

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