# Video Question Answering on VideoMME (Short, Median, Long, Avg. Metrics)

Source: [Video Question Answering on VideoMME (Short, Median, Long, Avg. Metrics)](<https://www.sota2.com/research/sota/video-question-answering-on-videomme-short-median-long-avg-metrics>)

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

Data updated: 2026-07-07T11:35:12.049Z

Evaluation of Video Question Answering performance on the VideoMME dataset, reporting scores segmented by question length categories: Short, Median, Long, and the overall Average.

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 | Short VQA Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 32 of 32 |
| 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

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

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

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

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

## Recorded results

| Method | Short VQA Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Full Tokens | 76 | Not recorded | 6a4c87e085dba69fef8b237c |
| Gemini 3.0 Pro | 75.11 | Not recorded | 6a4c883585dba69fef8b2bcd |
| OmniFocus | 74.78 | Not recorded | 6a4c87e085dba69fef8b237c |
| Seed 2.0 Pro | 74.67 | Not recorded | 6a4c883585dba69fef8b2bcd |
| Full Tokens | 74.11 | Not recorded | 6a4c87e085dba69fef8b237c |
| OmniZip | 73.67 | Not recorded | 6a4c87e085dba69fef8b237c |
| CuRe | 73.67 | Not recorded | 6a4c883585dba69fef8b2bcd |
| OmniFocus | 73.22 | Not recorded | 6a4c87e085dba69fef8b237c |
| Qwen3.6-35B-A3B | 73.11 | Not recorded | 6a4c883585dba69fef8b2bcd |
| DyCoke | 73.01 | Not recorded | 6a4c87e085dba69fef8b237c |
| V-LynX-7B | 73 | Not recorded | 6a1e63bfdbd7150506f85346 |
| OmniZip | 72.89 | Not recorded | 6a4c87e085dba69fef8b237c |
| Random | 72.11 | Not recorded | 6a4c87e085dba69fef8b237c |
| DyCoke | 72.01 | Not recorded | 6a4c87e085dba69fef8b237c |
| OmniFocus | 71.89 | Not recorded | 6a4c87e085dba69fef8b237c |
| Random | 71.44 | Not recorded | 6a4c87e085dba69fef8b237c |
| OmniFocus | 71.22 | Not recorded | 6a4c87e085dba69fef8b237c |
| PAVE-7B | 71.1 | Not recorded | 6a1e63bfdbd7150506f85346 |
| OmniZip | 70.56 | Not recorded | 6a4c87e085dba69fef8b237c |
| OmniZip | 70.56 | Not recorded | 6a4c87e085dba69fef8b237c |
| LLaVA-OV-7B | 70.1 | Not recorded | 6a1e63bfdbd7150506f85346 |
| DyCoke | 70.01 | Not recorded | 6a4c87e085dba69fef8b237c |
| Random | 67.89 | Not recorded | 6a4c87e085dba69fef8b237c |
| DyCoke | 67.89 | Not recorded | 6a4c87e085dba69fef8b237c |
| Random | 67.67 | Not recorded | 6a4c87e085dba69fef8b237c |
| V-LynX-0.5B | 63.1 | Not recorded | 6a1e63bfdbd7150506f85346 |
| Qwen3-VL-235B-A22B-Instruct | 61.33 | Not recorded | 6a4c883585dba69fef8b2bcd |
| Qwen3-VL-30B-A3B-Instruct | 61 | Not recorded | 6a4c883585dba69fef8b2bcd |
| PAVE-0.5B | 57.8 | Not recorded | 6a1e63bfdbd7150506f85346 |
| OwlCap-7B | 53.78 | Not recorded | 6a4c883585dba69fef8b2bcd |
| LLaVA-OV-0.5B | 53.4 | Not recorded | 6a1e63bfdbd7150506f85346 |
| Tarsier2-7B | 40.44 | Not recorded | 6a4c883585dba69fef8b2bcd |

### Full Tokens

Short VQA Accuracy: 76. Result ID: 6a4cb47b60b88bb339c837f9.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"100","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### Gemini 3.0 Pro

Short VQA Accuracy: 75.11. Result ID: 6a4ce45f60b88bb339cbb761.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

Recorded evaluation configuration: {"Access":"Proprietary","Protocol":"Prism"}. This record does not assert independent reproduction.

### OmniFocus

Short VQA Accuracy: 74.78. Result ID: 6a4cb47b60b88bb339c837fd.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"35","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### Seed 2.0 Pro

Short VQA Accuracy: 74.67. Result ID: 6a4ce45f60b88bb339cbb760.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a45f082dc71372d7795a1ae.md>)

Recorded evaluation configuration: {"Access":"Proprietary","Protocol":"Prism"}. This record does not assert independent reproduction.

### Full Tokens

Short VQA Accuracy: 74.11. Result ID: 6a4cb47b60b88bb339c83802.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"100","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### OmniZip

Short VQA Accuracy: 73.67. Result ID: 6a4cb47b60b88bb339c837fc.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/691d4f52d6b0dd334292d93f.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"35","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### CuRe

Short VQA Accuracy: 73.67. Result ID: 6a4ce45f60b88bb339cbb767.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

Recorded evaluation configuration: {"Access":"Open-source","Protocol":"Prism"}. This record does not assert independent reproduction.

### OmniFocus

Short VQA Accuracy: 73.22. Result ID: 6a4cb47b60b88bb339c83801.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"25","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### Qwen3.6-35B-A3B

Short VQA Accuracy: 73.11. Result ID: 6a4ce45f60b88bb339cbb765.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

Recorded evaluation configuration: {"Access":"Open-source","Protocol":"Prism"}. This record does not assert independent reproduction.

### DyCoke

Short VQA Accuracy: 73.01. Result ID: 6a4cb47b60b88bb339c837fb.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77bb1e209131cea75041.md>)

[Implementation](<https://github.com/KD-TAO/DyCoke>)

Recorded evaluation configuration: {"Retained Ratio (%)":"35","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### V-LynX-7B

Short VQA Accuracy: 73. Result ID: 6a1e91f0fbc373592c18a485.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e63bfdbd7150506f85346.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e63bfdbd7150506f85346.md>)

Recorded evaluation configuration: {"Model Category":"Task-specific models ≥ 7B","ΔParams.":"195.0M"}. This record does not assert independent reproduction.

### OmniZip

Short VQA Accuracy: 72.89. Result ID: 6a4cb47b60b88bb339c83800.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/691d4f52d6b0dd334292d93f.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"25","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### Random

Short VQA Accuracy: 72.11. Result ID: 6a4cb47b60b88bb339c837fa.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"35","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### DyCoke

Short VQA Accuracy: 72.01. Result ID: 6a4cb47b60b88bb339c837ff.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77bb1e209131cea75041.md>)

[Implementation](<https://github.com/KD-TAO/DyCoke>)

Recorded evaluation configuration: {"Retained Ratio (%)":"25","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### OmniFocus

Short VQA Accuracy: 71.89. Result ID: 6a4cb47b60b88bb339c83806.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"35","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### Random

Short VQA Accuracy: 71.44. Result ID: 6a4cb47b60b88bb339c837fe.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"25","Model Backbone":"Qwen2.5-Omni-7B"}. This record does not assert independent reproduction.

### OmniFocus

Short VQA Accuracy: 71.22. Result ID: 6a4cb47b60b88bb339c8380a.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"25","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### PAVE-7B

Short VQA Accuracy: 71.1. Result ID: 6a1e91f0fbc373592c18a484.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e63bfdbd7150506f85346.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa779f1e209131cea746cf.md>)

[Implementation](<https://github.com/dragonlzm/PAVE>)

Recorded evaluation configuration: {"Model Category":"Task-specific models ≥ 7B","ΔParams.":"500.5M"}. This record does not assert independent reproduction.

### OmniZip

Short VQA Accuracy: 70.56. Result ID: 6a4cb47b60b88bb339c83805.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/691d4f52d6b0dd334292d93f.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"35","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### OmniZip

Short VQA Accuracy: 70.56. Result ID: 6a4cb47b60b88bb339c83809.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/691d4f52d6b0dd334292d93f.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"25","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### LLaVA-OV-7B

Short VQA Accuracy: 70.1. Result ID: 6a1e91f0fbc373592c18a483.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e63bfdbd7150506f85346.md>)

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

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

Recorded evaluation configuration: {"Model Category":"Task-specific models ≥ 7B"}. This record does not assert independent reproduction.

### DyCoke

Short VQA Accuracy: 70.01. Result ID: 6a4cb47b60b88bb339c83808.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77bb1e209131cea75041.md>)

[Implementation](<https://github.com/KD-TAO/DyCoke>)

Recorded evaluation configuration: {"Retained Ratio (%)":"25","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### Random

Short VQA Accuracy: 67.89. Result ID: 6a4cb47b60b88bb339c83803.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"35","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### DyCoke

Short VQA Accuracy: 67.89. Result ID: 6a4cb47b60b88bb339c83804.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77bb1e209131cea75041.md>)

[Implementation](<https://github.com/KD-TAO/DyCoke>)

Recorded evaluation configuration: {"Retained Ratio (%)":"35","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### Random

Short VQA Accuracy: 67.67. Result ID: 6a4cb47b60b88bb339c83807.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c87e085dba69fef8b237c.md>)

Recorded evaluation configuration: {"Retained Ratio (%)":"25","Model Backbone":"Qwen2.5-Omni-3B"}. This record does not assert independent reproduction.

### V-LynX-0.5B

Short VQA Accuracy: 63.1. Result ID: 6a1e91f0fbc373592c18a482.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e63bfdbd7150506f85346.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a1e63bfdbd7150506f85346.md>)

Recorded evaluation configuration: {"Model Category":"Task-specific models \< 7B","ΔParams.":"68.7M"}. This record does not assert independent reproduction.

### Qwen3-VL-235B-A22B-Instruct

Short VQA Accuracy: 61.33. Result ID: 6a4ce45f60b88bb339cbb766.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

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

Recorded evaluation configuration: {"Access":"Open-source","Protocol":"Prism"}. This record does not assert independent reproduction.

### Qwen3-VL-30B-A3B-Instruct

Short VQA Accuracy: 61. Result ID: 6a4ce45f60b88bb339cbb764.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

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

Recorded evaluation configuration: {"Access":"Open-source","Protocol":"Prism"}. This record does not assert independent reproduction.

### PAVE-0.5B

Short VQA Accuracy: 57.8. Result ID: 6a1e91f0fbc373592c18a481.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e63bfdbd7150506f85346.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa779f1e209131cea746cf.md>)

[Implementation](<https://github.com/dragonlzm/PAVE>)

Recorded evaluation configuration: {"Model Category":"Task-specific models \< 7B","ΔParams.":"371.4M"}. This record does not assert independent reproduction.

### OwlCap-7B

Short VQA Accuracy: 53.78. Result ID: 6a4ce45f60b88bb339cbb763.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7add1e209131cea85953.md>)

Recorded evaluation configuration: {"Access":"Open-source","Protocol":"Prism"}. This record does not assert independent reproduction.

### LLaVA-OV-0.5B

Short VQA Accuracy: 53.4. Result ID: 6a1e91f0fbc373592c18a480.

[Reporting paper](<https://www.sota2.com/research/paper/6a1e63bfdbd7150506f85346.md>)

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

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

Recorded evaluation configuration: {"Model Category":"Task-specific models \< 7B"}. This record does not assert independent reproduction.

### Tarsier2-7B

Short VQA Accuracy: 40.44. Result ID: 6a4ce45f60b88bb339cbb762.

[Reporting paper](<https://www.sota2.com/research/paper/6a4c883585dba69fef8b2bcd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76bb1e209131cea6ecbe.md>)

[Implementation](<https://github.com/bytedance/tarsier>)

Recorded evaluation configuration: {"Access":"Open-source","Protocol":"Prism"}. This record does not assert independent reproduction.

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