# Multimodal Video Understanding on MMMU Video

Source: [Multimodal Video Understanding on MMMU Video](<https://www.sota2.com/research/sota/multimodal-video-understanding-on-mmmu-video>)

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

Data updated: 2026-03-10T07:26:46.184Z

Evaluation of model performance on the MMMU Video dataset specifically for Multimodal Video Understanding tasks, reporting Original and Filtered Scores.

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 | Multimodal Video Understanding |
| Dataset (including any stated split) | MMMU Video |
| Metric | Original Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [Original Score](<https://www.sota2.com/research/sota/multimodal-video-understanding-on-mmmu-video.md?metric=Original%20Score>): Higher is better; unit: PERCENT.

- [Filtered Score](<https://www.sota2.com/research/sota/multimodal-video-understanding-on-mmmu-video.md?metric=Filtered%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Original Score | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen2.5-Omni | 48.8 | Not recorded | 68fa7b891e209131cea88f71 |
| LLaVA-Video | 38.3 | Not recorded | 68fa7b891e209131cea88f71 |
| UniMambaMia | 36.7 | Not recorded | 68fa7b891e209131cea88f71 |
| LLaVA-OneVision | 33.1 | Not recorded | 68fa7b891e209131cea88f71 |
| Qwen2-VL | 32.9 | Not recorded | 68fa7b891e209131cea88f71 |
| LLaVA-NeXT-Vid. | 16.3 | Not recorded | 68fa7b891e209131cea88f71 |

### Qwen2.5-Omni

Original Score: 48.8. Result ID: 69afc79e6dbb7075b957b80b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7b891e209131cea88f71.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77a21e209131cea7482c.md>)

Recorded evaluation configuration: {"LLM":"Qwen2.5-7B","Use Audio":"✓","Latency (s)":"4.12"}. This record does not assert independent reproduction.

### LLaVA-Video

Original Score: 38.3. Result ID: 69afc79e6dbb7075b957b808.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7b891e209131cea88f71.md>)

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

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

Recorded evaluation configuration: {"LLM":"Qwen2-7B","Use Audio":"×","Latency (s)":"1.00"}. This record does not assert independent reproduction.

### UniMambaMia

Original Score: 36.7. Result ID: 69afc79e6dbb7075b957b809.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7b891e209131cea88f71.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b891e209131cea88f71.md>)

Recorded evaluation configuration: {"LLM":"Qwen2-7B","Use Audio":"✓","Latency (s)":"1.60"}. This record does not assert independent reproduction.

### LLaVA-OneVision

Original Score: 33.1. Result ID: 69afc79e6dbb7075b957b807.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7b891e209131cea88f71.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":"Qwen2-7B","Use Audio":"×","Latency (s)":"0.97"}. This record does not assert independent reproduction.

### Qwen2-VL

Original Score: 32.9. Result ID: 69afc79e6dbb7075b957b806.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7b891e209131cea88f71.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75151e209131cea65976.md>)

[Implementation](<https://github.com/QwenLM/Qwen2-VL>)

Recorded evaluation configuration: {"LLM":"Qwen2-7B","Use Audio":"×","Latency (s)":"1.63"}. This record does not assert independent reproduction.

### LLaVA-NeXT-Vid.

Original Score: 16.3. Result ID: 69afc79e6dbb7075b957b80a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7b891e209131cea88f71.md>)

Recorded evaluation configuration: {"LLM":"Vicuna-7B","Use Audio":"×","Latency (s)":"0.24"}. This record does not assert independent reproduction.

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