# Open-ended Question Answering on HLE

Source: [Open-ended Question Answering on HLE](<https://www.sota2.com/research/sota/open-ended-question-answering-on-hle>)

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

Data updated: 2026-05-12T08:40:39.622Z

Evaluation of model performance on the HLE dataset for Open-ended Question Answering, measured by Accuracy (Acc) and specific calibration metrics (Ver-ECE, Sem1-ECE, Sem2-ECE).

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 | Open-ended Question Answering |
| Dataset (including any stated split) | HLE |
| 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/open-ended-question-answering-on-hle.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Ver-ECE](<https://www.sota2.com/research/sota/open-ended-question-answering-on-hle.md?metric=Ver-ECE>): Lower is better; unit: SCALAR.

- [Sem1-ECE](<https://www.sota2.com/research/sota/open-ended-question-answering-on-hle.md?metric=Sem1-ECE>): Lower is better; unit: SCALAR.

- [Sem2-ECE](<https://www.sota2.com/research/sota/open-ended-question-answering-on-hle.md?metric=Sem2-ECE>): Lower is better; unit: SCALAR.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Anthropic | 20.5 | Not recorded | 6a02b428dbc836b87d824731 |
| Gemini | 15.1 | Not recorded | 6a02b428dbc836b87d824731 |
| OpenAI | 10.1 | Not recorded | 6a02b428dbc836b87d824731 |
| xAI | 6.9 | Not recorded | 6a02b428dbc836b87d824731 |
| Mistral | 5.3 | Not recorded | 6a02b428dbc836b87d824731 |

### Anthropic

Accuracy: 20.5. Result ID: 6a02e5a247f34ce924e5ae6f.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b428dbc836b87d824731.md>)

Recorded evaluation configuration: {"Provider":"Anthropic","Sample Size (N)":"1486"}. This record does not assert independent reproduction.

### Gemini

Accuracy: 15.1. Result ID: 6a02e5a247f34ce924e5ae70.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b428dbc836b87d824731.md>)

Recorded evaluation configuration: {"Provider":"Gemini","Sample Size (N)":"2111"}. This record does not assert independent reproduction.

### OpenAI

Accuracy: 10.1. Result ID: 6a02e5a247f34ce924e5ae6e.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b428dbc836b87d824731.md>)

Recorded evaluation configuration: {"Provider":"OpenAI","Sample Size (N)":"2128"}. This record does not assert independent reproduction.

### xAI

Accuracy: 6.9. Result ID: 6a02e5a247f34ce924e5ae71.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b428dbc836b87d824731.md>)

Recorded evaluation configuration: {"Provider":"xAI","Sample Size (N)":"2005"}. This record does not assert independent reproduction.

### Mistral

Accuracy: 5.3. Result ID: 6a02e5a247f34ce924e5ae72.

[Reporting paper](<https://www.sota2.com/research/paper/6a02b428dbc836b87d824731.md>)

Recorded evaluation configuration: {"Provider":"Mistral","Sample Size (N)":"2128"}. This record does not assert independent reproduction.

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