# Audio-video understanding on OmniBench

Source: [Audio-video understanding on OmniBench](<https://www.sota2.com/research/sota/audio-video-understanding-on-omnibench>)

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

Data updated: 2026-03-31T09:10:41.542Z

Evaluation of model performance on the OmniBench dataset focusing specifically on Audio-video understanding tasks.

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 | Audio-video understanding |
| Dataset (including any stated split) | OmniBench |
| Metric | Score |
| Unit | SCALAR |
| Direction | Higher is better |
| Results loaded | 10 of 10 |
| 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

- [Score](<https://www.sota2.com/research/sota/audio-video-understanding-on-omnibench.md?metric=Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Score | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen2.5-Omni + AVATAR | 49.1 | Not recorded | 68fa7a631e209131cea82cee |
| AV-Reasoner | 48.3 | Not recorded | 68fa7a631e209131cea82cee |
| Ola-7B + AVATAR | 47.2 | Not recorded | 68fa7a631e209131cea82cee |
| Omni-R1 | 46.9 | Not recorded | 68fa7a631e209131cea82cee |
| Ola-7B + GRPO | 46.8 | Not recorded | 68fa7a631e209131cea82cee |
| Echolnk | 46.5 | Not recorded | 68fa7a631e209131cea82cee |
| Qwen2.5-Omni + GRPO | 45.4 | Not recorded | 68fa7a631e209131cea82cee |
| Ola-7B | 45.3 | Not recorded | 68fa7a631e209131cea82cee |
| HumanOmni | 44.9 | Not recorded | 68fa7a631e209131cea82cee |
| Qwen2.5-Omni | 44.2 | Not recorded | 68fa7a631e209131cea82cee |

### Qwen2.5-Omni + AVATAR

Score: 49.1. Result ID: 69cb8eb147f34ce924b4119e.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"AVATAR"}. This record does not assert independent reproduction.

### AV-Reasoner

Score: 48.3. Result ID: 69cb8eb147f34ce924b41198.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a2d1e209131cea81903.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"N/A"}. This record does not assert independent reproduction.

### Ola-7B + AVATAR

Score: 47.2. Result ID: 69cb8eb147f34ce924b4119b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"AVATAR"}. This record does not assert independent reproduction.

### Omni-R1

Score: 46.9. Result ID: 69cb8eb147f34ce924b41196.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78da1e209131cea7acde.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"N/A"}. This record does not assert independent reproduction.

### Ola-7B + GRPO

Score: 46.8. Result ID: 69cb8eb147f34ce924b4119a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79131e209131cea7c004.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"GRPO"}. This record does not assert independent reproduction.

### Echolnk

Score: 46.5. Result ID: 69cb8eb147f34ce924b41195.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79811e209131cea7ddb0.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"N/A"}. This record does not assert independent reproduction.

### Qwen2.5-Omni + GRPO

Score: 45.4. Result ID: 69cb8eb147f34ce924b4119d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

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

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"GRPO"}. This record does not assert independent reproduction.

### Ola-7B

Score: 45.3. Result ID: 69cb8eb147f34ce924b41199.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"Baseline"}. This record does not assert independent reproduction.

### HumanOmni

Score: 44.9. Result ID: 69cb8eb147f34ce924b41197.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76bb1e209131cea6ecef.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"N/A"}. This record does not assert independent reproduction.

### Qwen2.5-Omni

Score: 44.2. Result ID: 69cb8eb147f34ce924b4119c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a631e209131cea82cee.md>)

Recorded evaluation configuration: {"Reinforcement Learning Strategy":"Baseline"}. This record does not assert independent reproduction.

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