# Mathematical Reasoning on Math500 (Pass@1, TPS, Latency)

Source: [Mathematical Reasoning on Math500 (Pass@1, TPS, Latency)](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math500-pass-at-1-tps-latency>)

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

Data updated: 2026-03-19T09:09:37.044Z

Evaluation of model performance and inference efficiency on the Math500 dataset for Mathematical Reasoning, specifically reporting Pass@1 accuracy, Tokens Per Second (TPS), and average Latency.

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) | Math-500 |
| Metric | Pass@1 Acc |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 18 of 18 |
| 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

- [Pass@1 Acc](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math500-pass-at-1-tps-latency.md?metric=Pass%401%20Acc>): Higher is better; unit: PERCENT.

- [TPS](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math500-pass-at-1-tps-latency.md?metric=TPS>): Higher is better; unit: SCALAR.

- [Latency](<https://www.sota2.com/research/sota/mathematical-reasoning-on-math500-pass-at-1-tps-latency.md?metric=Latency>): Lower is better; unit: TIME.

## Recorded results

| Method | Pass@1 Acc | Result date | Report paper |
| --- | --- | --- | --- |
| Target Model | 93.5 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| Target Model | 92.8 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| SemanticSpec | 92.17 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| SemanticSpec | 91 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| PreSafe | 90.7 | Not recorded | 69bb831d0636e5bf4afc1aed |
| SafePath | 90.4 | Not recorded | 69bb831d0636e5bf4afc1aed |
| SafeChain | 90.2 | Not recorded | 69bb831d0636e5bf4afc1aed |
| DS-7B | 89.8 | Not recorded | 69bb831d0636e5bf4afc1aed |
| SpecSampling | 89.5 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| UnSafeChain | 89.1 | Not recorded | 69bb831d0636e5bf4afc1aed |
| SpecSampling | 88 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| R2D | 84.4 | Not recorded | 69bb831d0636e5bf4afc1aed |
| SpecReason | 84 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| SpecThinking | 82.3 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| Draft Model | 81.42 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| Draft Model | 81.42 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| SpecReason | 77 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |
| SpecThinking | 71 | Not recorded | 6982d2f5a8b01e76cc8e99a0 |

### Target Model

Pass@1 Acc: 93.5. Result ID: 69964ebb5698665ec4935f32.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"QwQ-32B"}. This record does not assert independent reproduction.

### Target Model

Pass@1 Acc: 92.8. Result ID: 69964ebb5698665ec4935f2c.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"DeepseekR1-32B"}. This record does not assert independent reproduction.

### SemanticSpec

Pass@1 Acc: 92.17. Result ID: 69964ebb5698665ec4935f36.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"QwQ-32B"}. This record does not assert independent reproduction.

### SemanticSpec

Pass@1 Acc: 91. Result ID: 69964ebb5698665ec4935f30.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"DeepseekR1-32B"}. This record does not assert independent reproduction.

### PreSafe

Pass@1 Acc: 90.7. Result ID: 69bbbd4c47f34ce924a5cf48.

[Reporting paper](<https://www.sota2.com/research/paper/69bb831d0636e5bf4afc1aed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69bb831d0636e5bf4afc1aed.md>)

Recorded evaluation configuration: {"base\_model":"DeepSeek-R1-Distill-Qwen-7B"}. This record does not assert independent reproduction.

### SafePath

Pass@1 Acc: 90.4. Result ID: 69bbbd4c47f34ce924a5cf46.

[Reporting paper](<https://www.sota2.com/research/paper/69bb831d0636e5bf4afc1aed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa790c1e209131cea7bde6.md>)

Recorded evaluation configuration: {"base\_model":"DeepSeek-R1-Distill-Qwen-7B"}. This record does not assert independent reproduction.

### SafeChain

Pass@1 Acc: 90.2. Result ID: 69bbbd4c47f34ce924a5cf44.

[Reporting paper](<https://www.sota2.com/research/paper/69bb831d0636e5bf4afc1aed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76ff1e209131cea7058f.md>)

Recorded evaluation configuration: {"base\_model":"DeepSeek-R1-Distill-Qwen-7B"}. This record does not assert independent reproduction.

### DS-7B

Pass@1 Acc: 89.8. Result ID: 69bbbd4c47f34ce924a5cf43.

[Reporting paper](<https://www.sota2.com/research/paper/69bb831d0636e5bf4afc1aed.md>)

Recorded evaluation configuration: {"base\_model":"DeepSeek-R1-Distill-Qwen-7B"}. This record does not assert independent reproduction.

### SpecSampling

Pass@1 Acc: 89.5. Result ID: 69964ebb5698665ec4935f35.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6f1e209131cea458d8.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"QwQ-32B"}. This record does not assert independent reproduction.

### UnSafeChain

Pass@1 Acc: 89.1. Result ID: 69bbbd4c47f34ce924a5cf45.

[Reporting paper](<https://www.sota2.com/research/paper/69bb831d0636e5bf4afc1aed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7a421e209131cea81fe3.md>)

Recorded evaluation configuration: {"base\_model":"DeepSeek-R1-Distill-Qwen-7B"}. This record does not assert independent reproduction.

### SpecSampling

Pass@1 Acc: 88. Result ID: 69964ebb5698665ec4935f2f.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f6f1e209131cea458d8.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"DeepseekR1-32B"}. This record does not assert independent reproduction.

### R2D

Pass@1 Acc: 84.4. Result ID: 69bbbd4c47f34ce924a5cf47.

[Reporting paper](<https://www.sota2.com/research/paper/69bb831d0636e5bf4afc1aed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b741e209131cea88992.md>)

Recorded evaluation configuration: {"base\_model":"DeepSeek-R1-Distill-Qwen-7B"}. This record does not assert independent reproduction.

### SpecReason

Pass@1 Acc: 84. Result ID: 69964ebb5698665ec4935f2d.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78961e209131cea791f9.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"DeepseekR1-32B"}. This record does not assert independent reproduction.

### SpecThinking

Pass@1 Acc: 82.3. Result ID: 69964ebb5698665ec4935f2e.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa780a1e209131cea767b6.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"DeepseekR1-32B"}. This record does not assert independent reproduction.

### Draft Model

Pass@1 Acc: 81.42. Result ID: 69964ebb5698665ec4935f2b.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"DeepseekR1-32B"}. This record does not assert independent reproduction.

### Draft Model

Pass@1 Acc: 81.42. Result ID: 69964ebb5698665ec4935f31.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"QwQ-32B"}. This record does not assert independent reproduction.

### SpecReason

Pass@1 Acc: 77. Result ID: 69964ebb5698665ec4935f33.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78961e209131cea791f9.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"QwQ-32B"}. This record does not assert independent reproduction.

### SpecThinking

Pass@1 Acc: 71. Result ID: 69964ebb5698665ec4935f34.

[Reporting paper](<https://www.sota2.com/research/paper/6982d2f5a8b01e76cc8e99a0.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa780a1e209131cea767b6.md>)

Recorded evaluation configuration: {"Draft Model":"DeepseekR1-1.5B","Target Model":"QwQ-32B"}. This record does not assert independent reproduction.

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