# Challenging Cases Understanding on Vibe-Eval

Source: [Challenging Cases Understanding on Vibe-Eval](<https://www.sota2.com/research/sota/challenging-cases-understanding-on-vibe-eval>)

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

Data updated: 2026-02-26T05:30:00.790Z

Evaluation of model performance on the Vibe-Eval benchmark, focusing on Challenging Cases Understanding 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 | Challenging Cases Understanding |
| Dataset (including any stated split) | Vibe-Eval |
| 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/challenging-cases-understanding-on-vibe-eval.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| GPT-4o | 63.1 | Not recorded | 68fa756f1e209131cea67e58 |
| GPT-4V | 57.9 | Not recorded | 68fa756f1e209131cea67e58 |
| LLaVA-OneVision-7B | 51.7 | Not recorded | 68fa756f1e209131cea67e58 |
| LLaVA-OneVision-72B | 50.7 | Not recorded | 68fa756f1e209131cea67e58 |
| LLaVA-OneVision-0.5B | 33.8 | Not recorded | 68fa756f1e209131cea67e58 |

### GPT-4o

Accuracy: 63.1. Result ID: 6993a4093a6c3148491779d8.

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

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### GPT-4V

Accuracy: 57.9. Result ID: 6993a4093a6c3148491779d7.

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

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

### LLaVA-OneVision-7B

Accuracy: 51.7. Result ID: 6993a4093a6c3148491779d5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.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 Scale":"7B"}. This record does not assert independent reproduction.

### LLaVA-OneVision-72B

Accuracy: 50.7. Result ID: 6993a4093a6c3148491779d6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.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 Scale":"72B"}. This record does not assert independent reproduction.

### LLaVA-OneVision-0.5B

Accuracy: 33.8. Result ID: 6993a4093a6c3148491779d4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa756f1e209131cea67e58.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 Scale":"0.5B"}. This record does not assert independent reproduction.

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