# Video Question Answering on VideoMME

Source: [Video Question Answering on VideoMME](<https://www.sota2.com/research/sota/video-question-answering-on-videomme>)

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

Data updated: 2026-06-30T09:05:19.143Z

Evaluation of model performance on the VideoMME dataset for Video Question Answering tasks, reporting scores across Short, Medium, Long, and Overall question lengths.

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 Question Answering |
| Dataset (including any stated split) | VideoMME |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 341 of 341 |
| 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-question-answering-on-videomme.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Accuracy (Short)](<https://www.sota2.com/research/sota/video-question-answering-on-videomme.md?metric=Accuracy%20(Short)>): Higher is better; unit: PERCENT.

- [Accuracy (Medium)](<https://www.sota2.com/research/sota/video-question-answering-on-videomme.md?metric=Accuracy%20(Medium)>): Higher is better; unit: PERCENT.

- [Accuracy (Long)](<https://www.sota2.com/research/sota/video-question-answering-on-videomme.md?metric=Accuracy%20(Long)>): Higher is better; unit: PERCENT.

- [Accuracy (Overall)](<https://www.sota2.com/research/sota/video-question-answering-on-videomme.md?metric=Accuracy%20(Overall)>): Higher is better; unit: PERCENT.

- [Think Ratio](<https://www.sota2.com/research/sota/video-question-answering-on-videomme.md?metric=Think%20Ratio>): Higher is better; unit: SCALAR.

- [Response Length](<https://www.sota2.com/research/sota/video-question-answering-on-videomme.md?metric=Response%20Length>): Higher is better; unit: SCALAR.

- [Average Accuracy](<https://www.sota2.com/research/sota/video-question-answering-on-videomme.md?metric=Average%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Gemini 2.5 Pro | 85.1 | Not recorded | 6965d1856343fa78c8323529 |
| Gemini-2.5-Pro | 84.3 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| Gemini 1.5 Pro | 78.6 | Not recorded | 6965d1856343fa78c8323529 |
| Seed1.5-VL | 77.9 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| Gemini-1.5-Flash | 76.1 | Not recorded | 6965d1856343fa78c8323529 |
| Gemini 1.5 Pro | 75 | Not recorded | 68fa766c1e209131cea6cdce |
| Gemini-1.5-Pro | 75 | Not recorded | 68fa77401e209131cea71fc7 |
| Gemini-1.5-Pro | 75 | Not recorded | 698575d4c97a0b3e3b5d5e5a |
| Gemini-1.5-Pro | 75 | Not recorded | 6a02b460dbc836b87d824a46 |
| Qwen3-VL-8B | 72.5 | Not recorded | 69608b7853fdd5c36174444d |
| GPT 4o | 71.9 | Not recorded | 68fa766c1e209131cea6cdce |
| GPT-4o | 71.9 | Not recorded | 68fa77401e209131cea71fc7 |
| GPT4-o | 71.9 | Not recorded | 698575d4c97a0b3e3b5d5e5a |
| GPT-4o | 71.9 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| GPT-4o | 71.9 | Not recorded | 69cdf8470636e5bf4afccd52 |
| GPT-4o | 71.9 | Not recorded | 6a02b460dbc836b87d824a46 |
| VideoAuto-R1 | 71.7 | Not recorded | 69608b7853fdd5c36174444d |
| Qwen2-VL-7B-Instruct | 71.4 | Not recorded | 6965d1856343fa78c8323529 |
| Gemini-1.5-Flash | 70.3 | Not recorded | 68fa77401e209131cea71fc7 |
| Gemini 1.5 Flash | 70.3 | Not recorded | 69cdf8470636e5bf4afccd52 |
| Gemini-1.5-Flash | 70.3 | Not recorded | 6a02b460dbc836b87d824a46 |
| Vanilla | 69.4 | Not recorded | 69cb55850636e5bf4afca474 |
| LLaVA-Video-7B+TIR-Flow | 68.9 | Not recorded | 6965d1856343fa78c8323529 |
| Qwen3-VL8B + OutRo | 68.33 | Not recorded | 69b8e0140636e5bf4afbea8e |
| Qwen2.5-VL-7B+TIR-Flow | 68.3 | Not recorded | 6965d1856343fa78c8323529 |
| T\* | 68.3 | Not recorded | 6a02b460dbc836b87d824a46 |
| Qwen3-VL8B | 67.44 | Not recorded | 69b8e0140636e5bf4afbea8e |
| ResAdapt | 67.4 | Not recorded | 69cb55850636e5bf4afca474 |
| VideoAuto-R1 | 67.3 | Not recorded | 69608b7853fdd5c36174444d |
| Random Drop | 67.2 | Not recorded | 69cb55850636e5bf4afca474 |
| ToMe | 67.2 | Not recorded | 69cb55850636e5bf4afca474 |
| VisionZip | 67.1 | Not recorded | 69cb55850636e5bf4afca474 |
| ResAdapt | 66.8 | Not recorded | 69cb55850636e5bf4afca474 |
| FixedScale | 66.7 | Not recorded | 69cb55850636e5bf4afca474 |
| InternVL3-8B | 66.3 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| LOVE-R1 | 66.2 | Not recorded | 69608b7853fdd5c36174444d |
| VideoLLaMA3-7B | 66.2 | Not recorded | 6965d1856343fa78c8323529 |
| VideoLLaMA3-7B | 66.2 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| VideoAuto-R1 + ResAdapt | 66.2 | Not recorded | 69cb55850636e5bf4afca474 |
| MIRA | 66.2 | Not recorded | 6a02b460dbc836b87d824a46 |
| Qwen2.5-VL-7B | 66 | Not recorded | 69608b7853fdd5c36174444d |
| InternVL3.5-8B | 66 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| ResAdapt | 65.6 | Not recorded | 69cb55850636e5bf4afca474 |
| TPO | 65.6 | Not recorded | 6a02b460dbc836b87d824a46 |
| LongVU | 65.4 | Not recorded | 68fa77611e209131cea72d9a |
| InternVL3.5-4B | 65.4 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| LLaVA-Video w/ AKS | 65.3 | Not recorded | 68fa77401e209131cea71fc7 |
| Weaver | 65.3 | Not recorded | 698575d4c97a0b3e3b5d5e5a |
| R-MSD (4B) | 65.3 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| Vanilla | 65.3 | Not recorded | 69cb55850636e5bf4afca474 |
| VideoChat-R1.5 | 65.2 | Not recorded | 69608b7853fdd5c36174444d |
| TTA-Vid | 65.11 | Not recorded | 69cdf8470636e5bf4afccd52 |
| Long-VILA-R1 | 65.1 | Not recorded | 698575d4c97a0b3e3b5d5e5a |
| LongVILA-R1 | 65.1 | Not recorded | 69608b7853fdd5c36174444d |
| ToMe | 65.1 | Not recorded | 69cb55850636e5bf4afca474 |
| CATS (ours)† | 65.04 | Not recorded | 6a02b460dbc836b87d824a46 |
| Vanilla | 65 | Not recorded | 69cb55850636e5bf4afca474 |
| DToMA | 65 | Not recorded | 6a02b460dbc836b87d824a46 |
| Random Drop | 64.9 | Not recorded | 69cb55850636e5bf4afca474 |
| VisionZip | 64.8 | Not recorded | 69cb55850636e5bf4afca474 |
| VideoChat-R1.5 | 64.8 | Not recorded | 69cdf8470636e5bf4afccd52 |
| VideoAuto-R1 | 64.7 | Not recorded | 69cb55850636e5bf4afca474 |
| VideoAuto-R1 + ResAdapt | 64.7 | Not recorded | 69cb55850636e5bf4afca474 |
| ToMe | 64.7 | Not recorded | 69cb55850636e5bf4afca474 |
| BIMBA | 64.7 | Not recorded | 6a02b460dbc836b87d824a46 |
| BIMBA-LLaVA | 64.67 | Not recorded | 68fa77611e209131cea72d9a |
| AKS† | 64.48 | Not recorded | 6a02b460dbc836b87d824a46 |
| LLaVA-Video | 64.4 | Not recorded | 68fa77401e209131cea71fc7 |
| Video-R1 | 64.3 | Not recorded | 69cdf8470636e5bf4afccd52 |
| InternVL2.5 | 64.2 | Not recorded | 698575d4c97a0b3e3b5d5e5a |
| InternVL2.5-8B | 64.2 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| VisionZip | 64.2 | Not recorded | 69cb55850636e5bf4afca474 |
| VITAL | 64.1 | Not recorded | 69608b7853fdd5c36174444d |
| FixedScale | 64.1 | Not recorded | 69cb55850636e5bf4afca474 |
| Random Drop | 64.1 | Not recorded | 69cb55850636e5bf4afca474 |
| VideoChat-R1 | 64.1 | Not recorded | 69cdf8470636e5bf4afccd52 |
| Video-RFT | 64.1 | Not recorded | 69cdf8470636e5bf4afccd52 |
| Original SFT+RL (4B) | 64 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| Qwen3-VL-8B-Instruct | 64 | Not recorded | 69b399e90636e5bf4afbc8e4 |
| FlashVid | 63.9 | Not recorded | 69cb55850636e5bf4afca474 |

### Gemini 2.5 Pro

Accuracy: 85.1. Result ID: 6997dd1b5698665ec43597ee.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Frames":null}. This record does not assert independent reproduction.

### Gemini-2.5-Pro

Accuracy: 84.3. Result ID: 69b3c12f47f34ce9249d8082.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c311e209131cea8cef5.md>)

Recorded evaluation configuration: {"Frames":"-"}. This record does not assert independent reproduction.

### Gemini 1.5 Pro

Accuracy: 78.6. Result ID: 6997dd1b5698665ec43597ed.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

Recorded evaluation configuration: {"Frames":"128"}. This record does not assert independent reproduction.

### Seed1.5-VL

Accuracy: 77.9. Result ID: 69b3c12f47f34ce9249d8083.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78791e209131cea788a1.md>)

Recorded evaluation configuration: {"Frames":"-"}. This record does not assert independent reproduction.

### Gemini-1.5-Flash

Accuracy: 76.1. Result ID: 6997dd1b5698665ec43597ec.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

Recorded evaluation configuration: {"Frames":"128"}. This record does not assert independent reproduction.

### Gemini 1.5 Pro

Accuracy: 75. Result ID: 69941f0a3a6c3148491b32af.

[Reporting paper](<https://www.sota2.com/research/paper/68fa766c1e209131cea6cdce.md>)

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

Recorded evaluation configuration: {"# Frames":"-"}. This record does not assert independent reproduction.

### Gemini-1.5-Pro

Accuracy: 75. Result ID: 69948d413a6c3148491df5dd.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77401e209131cea71fc7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78731e209131cea78622.md>)

Recorded evaluation configuration: {"Frames":"256"}. This record does not assert independent reproduction.

### Gemini-1.5-Pro

Accuracy: 75. Result ID: 699739e85698665ec48a0fa2.

[Reporting paper](<https://www.sota2.com/research/paper/698575d4c97a0b3e3b5d5e5a.md>)

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

Recorded evaluation configuration: {"#Frames":"1fps"}. This record does not assert independent reproduction.

### Gemini-1.5-Pro

Accuracy: 75. Result ID: 6a030edb47f34ce924e6dfed.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

Recorded evaluation configuration: {"Type":"Proprietary","Model":"–","Frames":"256"}. This record does not assert independent reproduction.

### Qwen3-VL-8B

Accuracy: 72.5. Result ID: 6997d6985698665ec42e92e5.

[Reporting paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Reasoning Mode":"None","Reproduced":true}. This record does not assert independent reproduction.

### GPT 4o

Accuracy: 71.9. Result ID: 69941f0a3a6c3148491b32b0.

[Reporting paper](<https://www.sota2.com/research/paper/68fa766c1e209131cea6cdce.md>)

Recorded evaluation configuration: {"# Frames":"-"}. This record does not assert independent reproduction.

### GPT-4o

Accuracy: 71.9. Result ID: 69948d413a6c3148491df5db.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77401e209131cea71fc7.md>)

Recorded evaluation configuration: {"Frames":"256"}. This record does not assert independent reproduction.

### GPT4-o

Accuracy: 71.9. Result ID: 699739e85698665ec48a0fa4.

[Reporting paper](<https://www.sota2.com/research/paper/698575d4c97a0b3e3b5d5e5a.md>)

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

Recorded evaluation configuration: {"#Frames":"1fps"}. This record does not assert independent reproduction.

### GPT-4o

Accuracy: 71.9. Result ID: 69b3c12f47f34ce9249d8081.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

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

Recorded evaluation configuration: {"Frames":"-"}. This record does not assert independent reproduction.

### GPT-4o

Accuracy: 71.9. Result ID: 69ce3c8647f34ce924b7db55.

[Reporting paper](<https://www.sota2.com/research/paper/69cdf8470636e5bf4afccd52.md>)

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

Recorded evaluation configuration: {"#params":"−","#frames":"−"}. This record does not assert independent reproduction.

### GPT-4o

Accuracy: 71.9. Result ID: 6a030edb47f34ce924e6dfeb.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

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

Recorded evaluation configuration: {"Type":"Proprietary","Model":"–","Frames":"256"}. This record does not assert independent reproduction.

### VideoAuto-R1

Accuracy: 71.7. Result ID: 6997d6985698665ec42e92ef.

[Reporting paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Implementation](<https://github.com/IVUL-KAUST/VideoAuto-R1/>)

Recorded evaluation configuration: {"Reasoning Mode":"AutoThink","Backbone":"Qwen3-VL-8B"}. This record does not assert independent reproduction.

### Qwen2-VL-7B-Instruct

Accuracy: 71.4. Result ID: 6997dd1b5698665ec43597f5.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

Recorded evaluation configuration: {"Frames":"2fps"}. This record does not assert independent reproduction.

### Gemini-1.5-Flash

Accuracy: 70.3. Result ID: 69948d413a6c3148491df5dc.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77401e209131cea71fc7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78731e209131cea78622.md>)

Recorded evaluation configuration: {"Frames":"256"}. This record does not assert independent reproduction.

### Gemini 1.5 Flash

Accuracy: 70.3. Result ID: 69ce3c8647f34ce924b7db56.

[Reporting paper](<https://www.sota2.com/research/paper/69cdf8470636e5bf4afccd52.md>)

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

Recorded evaluation configuration: {"#params":"−","#frames":"−"}. This record does not assert independent reproduction.

### Gemini-1.5-Flash

Accuracy: 70.3. Result ID: 6a030edb47f34ce924e6dfec.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

Recorded evaluation configuration: {"Type":"Proprietary","Model":"–","Frames":"256"}. This record does not assert independent reproduction.

### Vanilla

Accuracy: 69.4. Result ID: 69cbbf5847f34ce924b561be.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"100%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### LLaVA-Video-7B+TIR-Flow

Accuracy: 68.9. Result ID: 6997dd1b5698665ec43597f7.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

Recorded evaluation configuration: {"Frames":"32"}. This record does not assert independent reproduction.

### Qwen3-VL8B + OutRo

Accuracy: 68.33. Result ID: 69b926cc47f34ce924a25676.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e0140636e5bf4afbea8e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b8e0140636e5bf4afbea8e.md>)

Recorded evaluation configuration: {"Model":"Qwen3-VL8B","Decoding":"OutRo"}. This record does not assert independent reproduction.

### Qwen2.5-VL-7B+TIR-Flow

Accuracy: 68.3. Result ID: 6997dd1b5698665ec43597f9.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

Recorded evaluation configuration: {"Frames":"32"}. This record does not assert independent reproduction.

### T\*

Accuracy: 68.3. Result ID: 6a030edb47f34ce924e6dff8.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7acd1e209131cea85308.md>)

Recorded evaluation configuration: {"Type":"Training-free","Model":"LLaVA-OneVision-72B","Frames":"32"}. This record does not assert independent reproduction.

### Qwen3-VL8B

Accuracy: 67.44. Result ID: 69b926cc47f34ce924a25675.

[Reporting paper](<https://www.sota2.com/research/paper/69b8e0140636e5bf4afbea8e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Model":"Qwen3-VL8B"}. This record does not assert independent reproduction.

### ResAdapt

Accuracy: 67.4. Result ID: 69cbbf5847f34ce924b561c2.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Implementation](<https://github.com/Xnhyacinth/ResAdapt>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"22.9%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### VideoAuto-R1

Accuracy: 67.3. Result ID: 6997d6985698665ec42e92ee.

[Reporting paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Implementation](<https://github.com/IVUL-KAUST/VideoAuto-R1/>)

Recorded evaluation configuration: {"Reasoning Mode":"AutoThink","Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction.

### Random Drop

Accuracy: 67.2. Result ID: 69cbbf5847f34ce924b561bf.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"25.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### ToMe

Accuracy: 67.2. Result ID: 69cbbf5847f34ce924b561c0.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea4130c.md>)

[Implementation](<http://github.com/facebookresearch/ToMe>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"25.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### VisionZip

Accuracy: 67.1. Result ID: 69cbbf5847f34ce924b561c1.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

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

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

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"25.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### ResAdapt

Accuracy: 66.8. Result ID: 69cbbf5847f34ce924b561c7.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Implementation](<https://github.com/Xnhyacinth/ResAdapt>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"11.1%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### FixedScale

Accuracy: 66.7. Result ID: 69cbbf5847f34ce924b561c6.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"12.3%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### InternVL3-8B

Accuracy: 66.3. Result ID: 69b3c12f47f34ce9249d808e.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78241e209131cea76fa5.md>)

Recorded evaluation configuration: {"Frames":"-"}. This record does not assert independent reproduction.

### LOVE-R1

Accuracy: 66.2. Result ID: 6997d6985698665ec42e92ec.

[Reporting paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bae1e209131cea89d48.md>)

Recorded evaluation configuration: {"Reasoning Mode":"Think-Only"}. This record does not assert independent reproduction.

### VideoLLaMA3-7B

Accuracy: 66.2. Result ID: 6997dd1b5698665ec43597f3.

[Reporting paper](<https://www.sota2.com/research/paper/6965d1856343fa78c8323529.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79111e209131cea7bf26.md>)

Recorded evaluation configuration: {"Frames":"180"}. This record does not assert independent reproduction.

### VideoLLaMA3-7B

Accuracy: 66.2. Result ID: 69b3c12f47f34ce9249d8090.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79111e209131cea7bf26.md>)

Recorded evaluation configuration: {"Frames":"-"}. This record does not assert independent reproduction.

### VideoAuto-R1 + ResAdapt

Accuracy: 66.2. Result ID: 69cbbf5847f34ce924b561b0.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Implementation](<https://github.com/Xnhyacinth/ResAdapt>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Retention Ratio R":"23.8%","Reasoning (CoT)":true,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### MIRA

Accuracy: 66.2. Result ID: 6a030edb47f34ce924e6dffa.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

Recorded evaluation configuration: {"Type":"Training-free","Model":"LLaVA-Video-7B","Frames":"64"}. This record does not assert independent reproduction.

### Qwen2.5-VL-7B

Accuracy: 66. Result ID: 6997d6985698665ec42e92e4.

[Reporting paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Reasoning Mode":"None","Reproduced":true}. This record does not assert independent reproduction.

### InternVL3.5-8B

Accuracy: 66. Result ID: 69b3c12f47f34ce9249d808f.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7af01e209131cea85dc9.md>)

Recorded evaluation configuration: {"Frames":"-"}. This record does not assert independent reproduction.

### ResAdapt

Accuracy: 65.6. Result ID: 69cbbf5847f34ce924b561a9.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Implementation](<https://github.com/Xnhyacinth/ResAdapt>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Retention Ratio R":"22.9%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### TPO

Accuracy: 65.6. Result ID: 6a030edb47f34ce924e6dff6.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7af61e209131cea85fcf.md>)

Recorded evaluation configuration: {"Type":"Training-based","Model":"LLaVA-Video-7B","Frames":"64"}. This record does not assert independent reproduction.

### LongVU

Accuracy: 65.4. Result ID: 69949ab33a6c3148491e56f2.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77611e209131cea72d9a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75511e209131cea671f3.md>)

[Implementation](<https://github.com/Vision-CAIR/LongVU>)

Recorded evaluation configuration: {"LLM":"Qwen2-7B","Frames":"1fps"}. This record does not assert independent reproduction.

### InternVL3.5-4B

Accuracy: 65.4. Result ID: 69b3c12f47f34ce9249d808a.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7af01e209131cea85dc9.md>)

Recorded evaluation configuration: {"Frames":"-"}. This record does not assert independent reproduction.

### LLaVA-Video w/ AKS

Accuracy: 65.3. Result ID: 69948d413a6c3148491df5e6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77401e209131cea71fc7.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77401e209131cea71fc7.md>)

[Implementation](<https://github.com/ncTimTang/AKS>)

Recorded evaluation configuration: {"Frames":"64","LLM":"7B","sampling":"AKS"}. This record does not assert independent reproduction.

### Weaver

Accuracy: 65.3. Result ID: 699739e85698665ec48a0fae.

[Reporting paper](<https://www.sota2.com/research/paper/698575d4c97a0b3e3b5d5e5a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698575d4c97a0b3e3b5d5e5a.md>)

Recorded evaluation configuration: {"#Frames":"128+1fps"}. This record does not assert independent reproduction.

### R-MSD (4B)

Accuracy: 65.3. Result ID: 69b3c12f47f34ce9249d8093.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

Recorded evaluation configuration: {"Frames":"64"}. This record does not assert independent reproduction.

### Vanilla

Accuracy: 65.3. Result ID: 69cbbf5847f34ce924b561a5.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Retention Ratio R":"100%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### VideoChat-R1.5

Accuracy: 65.2. Result ID: 6997d6985698665ec42e92ed.

[Reporting paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b8f1e209131cea891be.md>)

Recorded evaluation configuration: {"Reasoning Mode":"Think-Only"}. This record does not assert independent reproduction.

### TTA-Vid

Accuracy: 65.11. Result ID: 69ce3c8647f34ce924b7db62.

[Reporting paper](<https://www.sota2.com/research/paper/69cdf8470636e5bf4afccd52.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cdf8470636e5bf4afccd52.md>)

Recorded evaluation configuration: {"base\_model":"InternVL-3","#params":"8B","#frames":"32"}. This record does not assert independent reproduction.

### Long-VILA-R1

Accuracy: 65.1. Result ID: 699739e85698665ec48a0faa.

[Reporting paper](<https://www.sota2.com/research/paper/698575d4c97a0b3e3b5d5e5a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a634.md>)

Recorded evaluation configuration: {"#Frames":"512"}. This record does not assert independent reproduction.

### LongVILA-R1

Accuracy: 65.1. Result ID: 6997d6985698665ec42e92eb.

[Reporting paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bc71e209131cea8a634.md>)

Recorded evaluation configuration: {"Reasoning Mode":"Think-Only"}. This record does not assert independent reproduction.

### ToMe

Accuracy: 65.1. Result ID: 69cbbf5847f34ce924b561a7.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea4130c.md>)

[Implementation](<http://github.com/facebookresearch/ToMe>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Retention Ratio R":"25.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### CATS (ours)†

Accuracy: 65.04. Result ID: 6a030edb47f34ce924e6dffb.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

Recorded evaluation configuration: {"Type":"Training-free","Model":"LLaVA-Video-7B","Frames":"32"}. This record does not assert independent reproduction.

### Vanilla

Accuracy: 65. Result ID: 69cbbf5847f34ce924b561b2.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"100%","Reasoning (CoT)":false,"Frames":"32 Frames"}. This record does not assert independent reproduction.

### DToMA

Accuracy: 65. Result ID: 6a030edb47f34ce924e6dff7.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

Recorded evaluation configuration: {"Type":"Training-free","Model":"LLaVA-Video-7B","Frames":"64"}. This record does not assert independent reproduction.

### Random Drop

Accuracy: 64.9. Result ID: 69cbbf5847f34ce924b561a6.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Retention Ratio R":"25.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### VisionZip

Accuracy: 64.8. Result ID: 69cbbf5847f34ce924b561a8.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.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","Retention Ratio R":"25.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### VideoChat-R1.5

Accuracy: 64.8. Result ID: 69ce3c8647f34ce924b7db5d.

[Reporting paper](<https://www.sota2.com/research/paper/69cdf8470636e5bf4afccd52.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b8f1e209131cea891be.md>)

Recorded evaluation configuration: {"#params":"7B","#frames":"128"}. This record does not assert independent reproduction.

### VideoAuto-R1

Accuracy: 64.7. Result ID: 69cbbf5847f34ce924b561af.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Implementation](<https://github.com/IVUL-KAUST/VideoAuto-R1/>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Retention Ratio R":"100%","Reasoning (CoT)":true,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### VideoAuto-R1 + ResAdapt

Accuracy: 64.7. Result ID: 69cbbf5847f34ce924b561b1.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Implementation](<https://github.com/Xnhyacinth/ResAdapt>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Retention Ratio R":"11.4%","Reasoning (CoT)":true,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### ToMe

Accuracy: 64.7. Result ID: 69cbbf5847f34ce924b561c4.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6ea91e209131cea4130c.md>)

[Implementation](<http://github.com/facebookresearch/ToMe>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"10.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### BIMBA

Accuracy: 64.7. Result ID: 6a030edb47f34ce924e6dff5.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

Recorded evaluation configuration: {"Type":"Training-based","Model":"BIMBA-7B","Frames":"128"}. This record does not assert independent reproduction.

### BIMBA-LLaVA

Accuracy: 64.67. Result ID: 69949ab33a6c3148491e56f5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77611e209131cea72d9a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77611e209131cea72d9a.md>)

[Implementation](<https://sites.google.com/view/bimba-mllm>)

Recorded evaluation configuration: {"LLM":"Qwen2-7B","Frames":128}. This record does not assert independent reproduction.

### AKS†

Accuracy: 64.48. Result ID: 6a030edb47f34ce924e6dff9.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b460dbc836b87d824a46.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77401e209131cea71fc7.md>)

[Implementation](<https://github.com/ncTimTang/AKS>)

Recorded evaluation configuration: {"Type":"Training-free","Model":"LLaVA-Video-7B","Frames":"32"}. This record does not assert independent reproduction.

### LLaVA-Video

Accuracy: 64.4. Result ID: 69948d413a6c3148491df5e5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa77401e209131cea71fc7.md>)

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

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

Recorded evaluation configuration: {"Frames":"64","LLM":"7B"}. This record does not assert independent reproduction.

### Video-R1

Accuracy: 64.3. Result ID: 69ce3c8647f34ce924b7db5b.

[Reporting paper](<https://www.sota2.com/research/paper/69cdf8470636e5bf4afccd52.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c6f1e209131cea8e0ec.md>)

Recorded evaluation configuration: {"#params":"7B","#frames":"128"}. This record does not assert independent reproduction.

### InternVL2.5

Accuracy: 64.2. Result ID: 699739e85698665ec48a0fa6.

[Reporting paper](<https://www.sota2.com/research/paper/698575d4c97a0b3e3b5d5e5a.md>)

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

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

Recorded evaluation configuration: {"#Frames":"16-64"}. This record does not assert independent reproduction.

### InternVL2.5-8B

Accuracy: 64.2. Result ID: 69b3c12f47f34ce9249d808d.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

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

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

Recorded evaluation configuration: {"Frames":"-"}. This record does not assert independent reproduction.

### VisionZip

Accuracy: 64.2. Result ID: 69cbbf5847f34ce924b561c5.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

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

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

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"10.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### VITAL

Accuracy: 64.1. Result ID: 6997d6985698665ec42e92ea.

[Reporting paper](<https://www.sota2.com/research/paper/69608b7853fdd5c36174444d.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b071e209131cea864cb.md>)

Recorded evaluation configuration: {"Reasoning Mode":"Think-Only"}. This record does not assert independent reproduction.

### FixedScale

Accuracy: 64.1. Result ID: 69cbbf5847f34ce924b561ad.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Retention Ratio R":"12.3%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### Random Drop

Accuracy: 64.1. Result ID: 69cbbf5847f34ce924b561c3.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"10.0%","Reasoning (CoT)":false,"Frames":"128 Frames"}. This record does not assert independent reproduction.

### VideoChat-R1

Accuracy: 64.1. Result ID: 69ce3c8647f34ce924b7db5c.

[Reporting paper](<https://www.sota2.com/research/paper/69cdf8470636e5bf4afccd52.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ba11e209131cea89767.md>)

Recorded evaluation configuration: {"#params":"7B","#frames":"128"}. This record does not assert independent reproduction.

### Video-RFT

Accuracy: 64.1. Result ID: 69ce3c8647f34ce924b7db5e.

[Reporting paper](<https://www.sota2.com/research/paper/69cdf8470636e5bf4afccd52.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c241e209131cea8ca8e.md>)

Recorded evaluation configuration: {"#params":"7B","#frames":"128"}. This record does not assert independent reproduction.

### Original SFT+RL (4B)

Accuracy: 64. Result ID: 69b3c12f47f34ce9249d808c.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

Recorded evaluation configuration: {"Frames":"64"}. This record does not assert independent reproduction.

### Qwen3-VL-8B-Instruct

Accuracy: 64. Result ID: 69b3c12f47f34ce9249d8091.

[Reporting paper](<https://www.sota2.com/research/paper/69b399e90636e5bf4afbc8e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6927db54d6b0dd3342930f66.md>)

Recorded evaluation configuration: {"Frames":"64"}. This record does not assert independent reproduction.

### FlashVid

Accuracy: 63.9. Result ID: 69cbbf5847f34ce924b561b6.

[Reporting paper](<https://www.sota2.com/research/paper/69cb55850636e5bf4afca474.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698abbc2c97a0b3e3b5de0ff.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-VL-8B","Retention Ratio R":"30.0%","Reasoning (CoT)":false,"Frames":"32 Frames"}. This record does not assert independent reproduction.

Page 1 of 4. This is an index page, not the full collection.

- [Next index page](<https://www.sota2.com/research/sota/video-question-answering-on-videomme.md?metric=Accuracy&page=2>): Follow this link to discover more pages.
