# Question Answering on OracleProto

Source: [Question Answering on OracleProto](<https://www.sota2.com/research/sota/question-answering-on-oracleproto>)

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

Data updated: 2026-05-06T08:35:33.372Z

Evaluation of model performance on the OracleProto dataset for Question Answering tasks, measured using Composite 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 | Question Answering |
| Dataset (including any stated split) | OracleProto |
| Metric | Composite Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

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

## Recorded results

| Method | Composite Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| DeepSeek-V3.2-Exp | 60.16 | Not recorded | 69facb14c0f56ed84308a3da |
| GLM 5 | 60.02 | Not recorded | 69facb14c0f56ed84308a3da |
| Qwen3.5-Flash | 58.96 | Not recorded | 69facb14c0f56ed84308a3da |
| Doubao Seed 2.0 Lite | 58.58 | Not recorded | 69facb14c0f56ed84308a3da |
| Kimi K2.5 | 58 | Not recorded | 69facb14c0f56ed84308a3da |
| MiniMax M2.5 | 54.94 | Not recorded | 69facb14c0f56ed84308a3da |

### DeepSeek-V3.2-Exp

Composite Accuracy: 60.16. Result ID: 69fafc5b47f34ce924dde75c.

[Reporting paper](<https://www.sota2.com/research/paper/69facb14c0f56ed84308a3da.md>)

Recorded evaluation configuration: {"Total Cost (USD)":"3.60","Per-Correct Cost (USD)":"0.025"}. This record does not assert independent reproduction.

### GLM 5

Composite Accuracy: 60.02. Result ID: 69fafc5b47f34ce924dde75d.

[Reporting paper](<https://www.sota2.com/research/paper/69facb14c0f56ed84308a3da.md>)

Recorded evaluation configuration: {"Total Cost (USD)":"7.06","Per-Correct Cost (USD)":"0.048"}. This record does not assert independent reproduction.

### Qwen3.5-Flash

Composite Accuracy: 58.96. Result ID: 69fafc5b47f34ce924dde75e.

[Reporting paper](<https://www.sota2.com/research/paper/69facb14c0f56ed84308a3da.md>)

Recorded evaluation configuration: {"Total Cost (USD)":"0.45","Per-Correct Cost (USD)":"0.003"}. This record does not assert independent reproduction.

### Doubao Seed 2.0 Lite

Composite Accuracy: 58.58. Result ID: 69fafc5b47f34ce924dde761.

[Reporting paper](<https://www.sota2.com/research/paper/69facb14c0f56ed84308a3da.md>)

Recorded evaluation configuration: {"Total Cost (USD)":"0.89","Per-Correct Cost (USD)":"0.006"}. This record does not assert independent reproduction.

### Kimi K2.5

Composite Accuracy: 58. Result ID: 69fafc5b47f34ce924dde760.

[Reporting paper](<https://www.sota2.com/research/paper/69facb14c0f56ed84308a3da.md>)

Recorded evaluation configuration: {"Total Cost (USD)":"6.79","Per-Correct Cost (USD)":"0.049"}. This record does not assert independent reproduction.

### MiniMax M2.5

Composite Accuracy: 54.94. Result ID: 69fafc5b47f34ce924dde75f.

[Reporting paper](<https://www.sota2.com/research/paper/69facb14c0f56ed84308a3da.md>)

Recorded evaluation configuration: {"Total Cost (USD)":"3.21","Per-Correct Cost (USD)":"0.024"}. This record does not assert independent reproduction.

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