# Mathematical Reasoning on AIME (Accuracy, Peak Memory)

Source: [Mathematical Reasoning on AIME (Accuracy, Peak Memory)](<https://www.sota2.com/research/sota/mathematical-reasoning-on-aime-accuracy-peak-memory>)

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

Data updated: 2026-05-22T08:18:23.636Z

Evaluation of model performance on the American Invitational Mathematics Examination (AIME) dataset, reporting standard accuracy and peak memory consumption.

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 |
| Metric | Accuracy |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/mathematical-reasoning-on-aime-accuracy-peak-memory.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Peak Memory (GiB)](<https://www.sota2.com/research/sota/mathematical-reasoning-on-aime-accuracy-peak-memory.md?metric=Peak%20Memory%20(GiB)>): Lower is better; unit: SIZE.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| FullKV | 35.6 | Not recorded | 6a0fe32cf984739ed6739f06 |
| ArborKV | 33.8 | Not recorded | 6a0fe32cf984739ed6739f06 |
| ThinKV | 33.2 | Not recorded | 6a0fe32cf984739ed6739f06 |
| StreamingLLM | 31.9 | Not recorded | 6a0fe32cf984739ed6739f06 |
| H2O | 31.5 | Not recorded | 6a0fe32cf984739ed6739f06 |
| FullKV | 28.3 | Not recorded | 6a0fe32cf984739ed6739f06 |
| ArborKV | 27.2 | Not recorded | 6a0fe32cf984739ed6739f06 |
| ThinKV | 26.5 | Not recorded | 6a0fe32cf984739ed6739f06 |
| StreamingLLM | 26.1 | Not recorded | 6a0fe32cf984739ed6739f06 |
| ArborKV | 26.1 | Not recorded | 6a0fe32cf984739ed6739f06 |
| H2O | 25.6 | Not recorded | 6a0fe32cf984739ed6739f06 |
| ThinKV | 25 | Not recorded | 6a0fe32cf984739ed6739f06 |
| ArborKV | 24.3 | Not recorded | 6a0fe32cf984739ed6739f06 |
| ThinKV | 23.1 | Not recorded | 6a0fe32cf984739ed6739f06 |
| StreamingLLM | 22.4 | Not recorded | 6a0fe32cf984739ed6739f06 |
| StreamingLLM | 21.6 | Not recorded | 6a0fe32cf984739ed6739f06 |
| H2O | 20.5 | Not recorded | 6a0fe32cf984739ed6739f06 |
| H2O | 19.6 | Not recorded | 6a0fe32cf984739ed6739f06 |

### FullKV

Accuracy: 35.6. Result ID: 6a10114bfbc373592c08f656.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"1.00"}. This record does not assert independent reproduction.

### ArborKV

Accuracy: 33.8. Result ID: 6a10114bfbc373592c08f65a.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"0.50"}. This record does not assert independent reproduction.

### ThinKV

Accuracy: 33.2. Result ID: 6a10114bfbc373592c08f659.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bd21e209131cea8ab82.md>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"0.50"}. This record does not assert independent reproduction.

### StreamingLLM

Accuracy: 31.9. Result ID: 6a10114bfbc373592c08f658.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72d71e209131cea58d79.md>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"0.50"}. This record does not assert independent reproduction.

### H2O

Accuracy: 31.5. Result ID: 6a10114bfbc373592c08f657.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa718c1e209131cea51ead.md>)

[Implementation](<https://github.com/FMInference/H2O>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"0.50"}. This record does not assert independent reproduction.

### FullKV

Accuracy: 28.3. Result ID: 6a10114bfbc373592c08f64d.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"1.00"}. This record does not assert independent reproduction.

### ArborKV

Accuracy: 27.2. Result ID: 6a10114bfbc373592c08f651.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"0.50"}. This record does not assert independent reproduction.

### ThinKV

Accuracy: 26.5. Result ID: 6a10114bfbc373592c08f650.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bd21e209131cea8ab82.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"0.50"}. This record does not assert independent reproduction.

### StreamingLLM

Accuracy: 26.1. Result ID: 6a10114bfbc373592c08f64f.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72d71e209131cea58d79.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"0.50"}. This record does not assert independent reproduction.

### ArborKV

Accuracy: 26.1. Result ID: 6a10114bfbc373592c08f65e.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"0.25"}. This record does not assert independent reproduction.

### H2O

Accuracy: 25.6. Result ID: 6a10114bfbc373592c08f64e.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa718c1e209131cea51ead.md>)

[Implementation](<https://github.com/FMInference/H2O>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"0.50"}. This record does not assert independent reproduction.

### ThinKV

Accuracy: 25. Result ID: 6a10114bfbc373592c08f65d.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bd21e209131cea8ab82.md>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"0.25"}. This record does not assert independent reproduction.

### ArborKV

Accuracy: 24.3. Result ID: 6a10114bfbc373592c08f655.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"0.25"}. This record does not assert independent reproduction.

### ThinKV

Accuracy: 23.1. Result ID: 6a10114bfbc373592c08f654.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bd21e209131cea8ab82.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"0.25"}. This record does not assert independent reproduction.

### StreamingLLM

Accuracy: 22.4. Result ID: 6a10114bfbc373592c08f65c.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72d71e209131cea58d79.md>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"0.25"}. This record does not assert independent reproduction.

### StreamingLLM

Accuracy: 21.6. Result ID: 6a10114bfbc373592c08f653.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa72d71e209131cea58d79.md>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"0.25"}. This record does not assert independent reproduction.

### H2O

Accuracy: 20.5. Result ID: 6a10114bfbc373592c08f652.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa718c1e209131cea51ead.md>)

[Implementation](<https://github.com/FMInference/H2O>)

Recorded evaluation configuration: {"Model":"Llama-3.1-8B","rho":"0.25"}. This record does not assert independent reproduction.

### H2O

Accuracy: 19.6. Result ID: 6a10114bfbc373592c08f65b.

[Reporting paper](<https://www.sota2.com/research/paper/6a0fe32cf984739ed6739f06.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa718c1e209131cea51ead.md>)

[Implementation](<https://github.com/FMInference/H2O>)

Recorded evaluation configuration: {"Model":"Qwen-2.5-7B","rho":"0.25"}. This record does not assert independent reproduction.

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