# Mathematical Reasoning on AIME 25 (Peak avg@32 score)

Source: [Mathematical Reasoning on AIME 25 (Peak avg@32 score)](<https://www.sota2.com/research/sota/mathematical-reasoning-on-aime-25-peak-avg-at-32-score>)

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

Data updated: 2026-02-26T05:57:39.213Z

Evaluation of model performance on the American Invitational Mathematics Examination (AIME) 25 dataset for single-turn mathematical reasoning, measured using the Peak avg@32 score metric.

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 | Single-Turn Mathematical Reasoning |
| Dataset (including any stated split) | AIME 25 |
| Metric | Peak avg@32 Score |
| Unit | SCALAR |
| 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

- [Peak avg@32 Score](<https://www.sota2.com/research/sota/mathematical-reasoning-on-aime-25-peak-avg-at-32-score.md?metric=Peak%20avg%4032%20Score>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Peak avg@32 Score | Result date | Report paper |
| --- | --- | --- | --- |
| OTB | 30.31 | Not recorded | 698abbadc97a0b3e3b5de029 |
| OGB | 30.1 | Not recorded | 698abbadc97a0b3e3b5de029 |
| OPO | 27.29 | Not recorded | 698abbadc97a0b3e3b5de029 |
| GRPO | 25.31 | Not recorded | 698abbadc97a0b3e3b5de029 |
| RLOO | 24.38 | Not recorded | 698abbadc97a0b3e3b5de029 |

### OTB

Peak avg@32 Score: 30.31. Result ID: 6998e8f25698665ec44d5dfa.

[Reporting paper](<https://www.sota2.com/research/paper/698abbadc97a0b3e3b5de029.md>)

[Method origin paper](<https://www.sota2.com/research/paper/698abbadc97a0b3e3b5de029.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8b-Base","Reasoning Paradigm":"Single-Turn"}. This record does not assert independent reproduction.

### OGB

Peak avg@32 Score: 30.1. Result ID: 6998e8f25698665ec44d5df9.

[Reporting paper](<https://www.sota2.com/research/paper/698abbadc97a0b3e3b5de029.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa60c41e209131ce9f12b5.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8b-Base","Reasoning Paradigm":"Single-Turn"}. This record does not assert independent reproduction.

### OPO

Peak avg@32 Score: 27.29. Result ID: 6998e8f25698665ec44d5df8.

[Reporting paper](<https://www.sota2.com/research/paper/698abbadc97a0b3e3b5de029.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79101e209131cea7bed6.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8b-Base","Reasoning Paradigm":"Single-Turn"}. This record does not assert independent reproduction.

### GRPO

Peak avg@32 Score: 25.31. Result ID: 6998e8f25698665ec44d5df6.

[Reporting paper](<https://www.sota2.com/research/paper/698abbadc97a0b3e3b5de029.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa73101e209131cea5a395.md>)

[Implementation](<https://github.com/deepseek-ai/DeepSeek-Math>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8b-Base","Reasoning Paradigm":"Single-Turn"}. This record does not assert independent reproduction.

### RLOO

Peak avg@32 Score: 24.38. Result ID: 6998e8f25698665ec44d5df7.

[Reporting paper](<https://www.sota2.com/research/paper/698abbadc97a0b3e3b5de029.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa725a1e209131cea55df7.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-8b-Base","Reasoning Paradigm":"Single-Turn"}. This record does not assert independent reproduction.

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