# MMAU-Pro Audio Instruction Following

Source: [MMAU-Pro Audio Instruction Following](<https://www.sota2.com/research/sota/mmau-pro-audio-instruction-following>)

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

Data updated: 2026-05-20T09:04:10.959Z

Evaluation of model performance on the MMAU-Pro dataset specifically for Audio Instruction Following tasks, measured by the Audio Instruction Following (AIF) score.

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 Instruction Following |
| Dataset (including any stated split) | MMAU-Pro Audio Instruction Following |
| Metric | AIF Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 5 of 5 |
| 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

- [AIF Score](<https://www.sota2.com/research/sota/mmau-pro-audio-instruction-following.md?metric=AIF%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | AIF Score | Result date | Report paper |
| --- | --- | --- | --- |
| OScaR | 88.5 | Not recorded | 6a0d4051f984739ed6737b03 |
| 16bit baseline | 87.4 | Not recorded | 6a0d4051f984739ed6737b03 |
| OTT | 83.9 | Not recorded | 6a0d4051f984739ed6737b03 |
| TurboQuant+ | 79.3 | Not recorded | 6a0d4051f984739ed6737b03 |
| KIVI | 78.2 | Not recorded | 6a0d4051f984739ed6737b03 |

### OScaR

AIF Score: 88.5. Result ID: 6a0d78d4fbc373592c05afd2.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d4051f984739ed6737b03.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0d4051f984739ed6737b03.md>)

[Implementation](<https://github.com/ZunhaiSu/OScaR-KV-Quant>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Omni-30B-A3B","Quantization Method":"INT2","Group Size":"128","Bit-width":"INT2"}. This record does not assert independent reproduction.

### 16bit baseline

AIF Score: 87.4. Result ID: 6a0d78d4fbc373592c05afce.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d4051f984739ed6737b03.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Omni-30B-A3B","Bit-width":"16-bit"}. This record does not assert independent reproduction.

### OTT

AIF Score: 83.9. Result ID: 6a0d78d4fbc373592c05afd0.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d4051f984739ed6737b03.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa789e1e209131cea7961c.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Omni-30B-A3B","Quantization Method":"INT2","Group Size":"128","Bit-width":"INT2"}. This record does not assert independent reproduction.

### TurboQuant+

AIF Score: 79.3. Result ID: 6a0d78d4fbc373592c05afd1.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d4051f984739ed6737b03.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Omni-30B-A3B","Bit-width":"2.5-bit"}. This record does not assert independent reproduction.

### KIVI

AIF Score: 78.2. Result ID: 6a0d78d4fbc373592c05afcf.

[Reporting paper](<https://www.sota2.com/research/paper/6a0d4051f984739ed6737b03.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa744b1e209131cea60dab.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-Omni-30B-A3B","Quantization Method":"INT2","Group Size":"128","Bit-width":"INT2"}. This record does not assert independent reproduction.

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