# Mathematical Reasoning on AIME 24&25 (Acc, Maj. Vote)

Source: [Mathematical Reasoning on AIME 24&25 (Acc, Maj. Vote)](<https://www.sota2.com/research/sota/mathematical-reasoning-on-aime-24-and-25-acc-maj-vote>)

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

Data updated: 2026-06-12T08:22:02.887Z

Evaluation of model performance on the combined AIME 2024 and 2025 datasets for mathematical reasoning, reporting standard Accuracy and Majority Vote scores.

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 | Mathematical Reasoning |
| Dataset (including any stated split) | AIME 24&25 |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| 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/mathematical-reasoning-on-aime-24-and-25-acc-maj-vote.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Majority Vote](<https://www.sota2.com/research/sota/mathematical-reasoning-on-aime-24-and-25-acc-maj-vote.md?metric=Majority%20Vote>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| Orch-RM | 67.5 | Not recorded | 6a2b92d4d33dfb23e783b2dd |
| GRPO | 65 | Not recorded | 6a2b92d4d33dfb23e783b2dd |
| RFT | 64.83 | Not recorded | 6a2b92d4d33dfb23e783b2dd |
| DPO | 64.67 | Not recorded | 6a2b92d4d33dfb23e783b2dd |
| Orch-RM | 64.33 | Not recorded | 6a2b92d4d33dfb23e783b2dd |
| LLM-as-a-judge | 59.17 | Not recorded | 6a2b92d4d33dfb23e783b2dd |
| MAS-Orchestra | 58.96 | Not recorded | 6a2b92d4d33dfb23e783b2dd |
| Qwen2.5-7B-Instruct | 29.83 | Not recorded | 6a2b92d4d33dfb23e783b2dd |

### Orch-RM

Accuracy: 67.5. Result ID: 6a2bc196fbc373592c2a304a.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

Recorded evaluation configuration: {"Context":"Continued-training","Sampling Level":"Orchestration"}. This record does not assert independent reproduction.

### GRPO

Accuracy: 65. Result ID: 6a2bc196fbc373592c2a3048.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

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

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

Recorded evaluation configuration: {"Context":"Continued-training","Sampling Level":"Trajectory"}. This record does not assert independent reproduction.

### RFT

Accuracy: 64.83. Result ID: 6a2bc196fbc373592c2a3047.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa708d1e209131cea4bb20.md>)

[Implementation](<https://github.com/OFA-Sys/gsm8k-ScRel>)

Recorded evaluation configuration: {"Context":"Continued-training","Sampling Level":"Trajectory"}. This record does not assert independent reproduction.

### DPO

Accuracy: 64.67. Result ID: 6a2bc196fbc373592c2a3046.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74511e209131cea6105d.md>)

Recorded evaluation configuration: {"Context":"Continued-training"}. This record does not assert independent reproduction.

### Orch-RM

Accuracy: 64.33. Result ID: 6a2bc196fbc373592c2a3044.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

Recorded evaluation configuration: {"Context":"Training-from-scratch","Sampling Level":"Orchestration"}. This record does not assert independent reproduction.

### LLM-as-a-judge

Accuracy: 59.17. Result ID: 6a2bc196fbc373592c2a3049.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa71aa1e209131cea526df.md>)

[Implementation](<https://github.com/lm-sys/FastChat/tree/main/fastchat/llm_judge>)

Recorded evaluation configuration: {"Context":"Continued-training","Sampling Level":"Orchestration"}. This record does not assert independent reproduction.

### MAS-Orchestra

Accuracy: 58.96. Result ID: 6a2bc196fbc373592c2a3045.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6971af07e25ad2fa9d1ed973.md>)

Recorded evaluation configuration: {"Context":"Training-from-scratch","Sampling Level":"Trajectory"}. This record does not assert independent reproduction.

### Qwen2.5-7B-Instruct

Accuracy: 29.83. Result ID: 6a2bc196fbc373592c2a3043.

[Reporting paper](<https://www.sota2.com/research/paper/6a2b92d4d33dfb23e783b2dd.md>)

Recorded evaluation configuration: {"Context":"Base orchestrator"}. This record does not assert independent reproduction.

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