# Korean Multimodal Question Answering on KMMMU

Source: [Korean Multimodal Question Answering on KMMMU](<https://www.sota2.com/research/sota/korean-multimodal-question-answering-on-kmmmu>)

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

Data updated: 2026-04-13T07:50:33.122Z

Evaluation of model performance on the KMMMU dataset for Korean Multimodal Question Answering tasks, typically measured by Accuracy.

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 | Korean Multimodal Question Answering |
| Dataset (including any stated split) | KMMMU |
| Metric | Accuracy |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/korean-multimodal-question-answering-on-kmmmu.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Qwen3.5-27B | 51.7 | Not recorded | 69dc788d7d0172140ec78ad9 |
| EXAONE 4.5 33B | 42.7 | Not recorded | 69dc788d7d0172140ec78ad9 |
| GPT-5 mini | 42.6 | Not recorded | 69dc788d7d0172140ec78ad9 |
| Qwen3-VL-235B-A22B | 42.1 | Not recorded | 69dc788d7d0172140ec78ad9 |
| Qwen3-VL-32B | 37.8 | Not recorded | 69dc788d7d0172140ec78ad9 |

### Qwen3.5-27B

Accuracy: 51.7. Result ID: 69dca03a47f34ce924c35cab.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad9.md>)

Recorded evaluation configuration: {"Mode":"REASONING","Architecture":"Dense","# Total Params":"27B","# Activated Params":"27B"}. This record does not assert independent reproduction.

### EXAONE 4.5 33B

Accuracy: 42.7. Result ID: 69dca03a47f34ce924c35ca7.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad9.md>)

[Implementation](<https://github.com/LG-AI-EXAONE/EXAONE-4.5>)

Recorded evaluation configuration: {"Mode":"REASONING","Architecture":"Dense","# Total Params":"33B","# Activated Params":"33B"}. This record does not assert independent reproduction.

### GPT-5 mini

Accuracy: 42.6. Result ID: 69dca03a47f34ce924c35ca8.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad9.md>)

[Method origin paper](<https://www.sota2.com/research/paper/695f39f053fdd5c361743ecb.md>)

Recorded evaluation configuration: {"Mode":"REASONING: HIGH","Architecture":"-","# Total Params":"-","# Activated Params":"-"}. This record does not assert independent reproduction.

### Qwen3-VL-235B-A22B

Accuracy: 42.1. Result ID: 69dca03a47f34ce924c35caa.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad9.md>)

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

Recorded evaluation configuration: {"Mode":"Thinking","Architecture":"MoE","# Total Params":"236B","# Activated Params":"23B"}. This record does not assert independent reproduction.

### Qwen3-VL-32B

Accuracy: 37.8. Result ID: 69dca03a47f34ce924c35ca9.

[Reporting paper](<https://www.sota2.com/research/paper/69dc788d7d0172140ec78ad9.md>)

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

Recorded evaluation configuration: {"Mode":"Thinking","Architecture":"Dense","# Total Params":"33B","# Activated Params":"33B"}. This record does not assert independent reproduction.

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