# Mathematical Reasoning on PRM800K (test)

Source: [Mathematical Reasoning on PRM800K (test)](<https://www.sota2.com/research/sota/mathematical-reasoning-on-prm800k-test>)

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

Data updated: 2026-04-10T08:36:37.129Z

Evaluation of model performance on the test split of the PRM800K dataset for mathematical reasoning, measured by 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 | Mathematical Reasoning |
| Dataset (including any stated split) | PRM800K (test) |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 15 of 15 |
| 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-prm800k-test.md?metric=Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| HRM | 80 | Not recorded | 68fa78691e209131cea784ba |
| HRM | 80 | Not recorded | 68fa78691e209131cea784ba |
| HRM | 74.4 | Not recorded | 68fa78691e209131cea784ba |
| HRM | 72.2 | Not recorded | 68fa78691e209131cea784ba |
| HRM | 71.1 | Not recorded | 68fa78691e209131cea784ba |
| PRM | 70 | Not recorded | 68fa78691e209131cea784ba |
| ORM | 67.7 | Not recorded | 68fa78691e209131cea784ba |
| ORM | 65.5 | Not recorded | 68fa78691e209131cea784ba |
| ORM | 65.5 | Not recorded | 68fa78691e209131cea784ba |
| PRM | 64.4 | Not recorded | 68fa78691e209131cea784ba |
| ORM | 63.3 | Not recorded | 68fa78691e209131cea784ba |
| ORM | 62.2 | Not recorded | 68fa78691e209131cea784ba |
| PRM | 61.1 | Not recorded | 68fa78691e209131cea784ba |
| PRM | 58.8 | Not recorded | 68fa78691e209131cea784ba |
| PRM | 57.7 | Not recorded | 68fa78691e209131cea784ba |

### HRM

Accuracy: 80. Result ID: 69d8b69047f34ce924c1990c.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

Recorded evaluation configuration: {"N":"16","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### HRM

Accuracy: 80. Result ID: 69d8b69047f34ce924c1990d.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

Recorded evaluation configuration: {"N":"24","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### HRM

Accuracy: 74.4. Result ID: 69d8b69047f34ce924c1990b.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

Recorded evaluation configuration: {"N":"8","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### HRM

Accuracy: 72.2. Result ID: 69d8b69047f34ce924c19909.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

Recorded evaluation configuration: {"N":"2","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### HRM

Accuracy: 71.1. Result ID: 69d8b69047f34ce924c1990a.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

Recorded evaluation configuration: {"N":"4","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### PRM

Accuracy: 70. Result ID: 69d8b69047f34ce924c19904.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"2","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### ORM

Accuracy: 67.7. Result ID: 69d8b69047f34ce924c19900.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"4","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### ORM

Accuracy: 65.5. Result ID: 69d8b69047f34ce924c19901.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"8","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### ORM

Accuracy: 65.5. Result ID: 69d8b69047f34ce924c19902.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"16","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### PRM

Accuracy: 64.4. Result ID: 69d8b69047f34ce924c19905.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"4","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### ORM

Accuracy: 63.3. Result ID: 69d8b69047f34ce924c19903.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"24","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### ORM

Accuracy: 62.2. Result ID: 69d8b69047f34ce924c198ff.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"2","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### PRM

Accuracy: 61.1. Result ID: 69d8b69047f34ce924c19906.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"8","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### PRM

Accuracy: 58.8. Result ID: 69d8b69047f34ce924c19907.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"16","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

### PRM

Accuracy: 57.7. Result ID: 69d8b69047f34ce924c19908.

[Reporting paper](<https://www.sota2.com/research/paper/68fa78691e209131cea784ba.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f9a1e209131cea468c7.md>)

[Implementation](<https://github.com/openai/prm800k>)

Recorded evaluation configuration: {"N":"24","policy\_model":"Qwen2.5-72B-Math-Instruct","reward\_model\_training":"fine-tuned on manually labeled PRM800K dataset"}. This record does not assert independent reproduction.

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