# Visual Question Answering on PanoVQA mini

Source: [Visual Question Answering on PanoVQA mini](<https://www.sota2.com/research/sota/visual-question-answering-on-panovqa-mini>)

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

Data updated: 2026-03-11T09:05:00.826Z

Evaluation of model performance on the PanoVQA mini dataset for Visual Question Answering, measured using average metrics (Avg(N), Avg(O), Avg(D), Avg).

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 | Visual Question Answering |
| Dataset (including any stated split) | PanoVQA mini |
| Metric | Avg (N) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 3 of 3 |
| 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

- [Avg (N)](<https://www.sota2.com/research/sota/visual-question-answering-on-panovqa-mini.md?metric=Avg%20(N)>): Higher is better; unit: PERCENT.

- [Avg (O)](<https://www.sota2.com/research/sota/visual-question-answering-on-panovqa-mini.md?metric=Avg%20(O)>): Higher is better; unit: PERCENT.

- [Avg (D)](<https://www.sota2.com/research/sota/visual-question-answering-on-panovqa-mini.md?metric=Avg%20(D)>): Higher is better; unit: PERCENT.

- [Avg Accuracy](<https://www.sota2.com/research/sota/visual-question-answering-on-panovqa-mini.md?metric=Avg%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Avg (N) | Result date | Report paper |
| --- | --- | --- | --- |
| GPT-5 | 22.02 | Not recorded | 69b0f72f2d603e6de3e45fca |
| Claude-Sonnet-4.5 | 19.29 | Not recorded | 69b0f72f2d603e6de3e45fca |
| Gemini-3-Pro | 18.04 | Not recorded | 69b0f72f2d603e6de3e45fca |

### GPT-5

Avg (N): 22.02. Result ID: 69b130366dbb7075b95bc8e9.

[Reporting paper](<https://www.sota2.com/research/paper/69b0f72f2d603e6de3e45fca.md>)

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

### Claude-Sonnet-4.5

Avg (N): 19.29. Result ID: 69b130366dbb7075b95bc8ea.

[Reporting paper](<https://www.sota2.com/research/paper/69b0f72f2d603e6de3e45fca.md>)

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

### Gemini-3-Pro

Avg (N): 18.04. Result ID: 69b130366dbb7075b95bc8e8.

[Reporting paper](<https://www.sota2.com/research/paper/69b0f72f2d603e6de3e45fca.md>)

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

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