# Mathematical Reasoning on AIME 2025 (Acc, Tokens, EFF) Source: [Mathematical Reasoning on AIME 2025 (Acc, Tokens, EFF)]() Collection index: [llms.txt]() Data updated: 2026-03-04T20:57:11.351Z Evaluation of model performance on the American Invitational Mathematics Examination (AIME) 2025 dataset, focusing on Accuracy (Acc), Token usage, and Efficiency Factor (EFF). 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 2025 | | 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](): Higher is better; unit: PERCENT. - [Tokens Used](): Lower is better; unit: SCALAR. - [Efficiency Factor (EFF)](): Higher is better; unit: SCALAR. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | D-SFT-\>D-RL3k-\>D-RL6k | 64.79 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | REO-RL(Q-Spec)\* | 64.4 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | Original | 64.38 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | D-SFT-\>D-RL6k | 63.54 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | D-SFT-\>D-RL6k | 61.88 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | R1-distill+DR.SAF\* | 57.9 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | D-SFT-\>D-RL3k-\>D-RL6k | 56.67 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | R1-distill+Length-Penalty\* | 54.7 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | ThinkPrune(6k) | 53.96 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | D-SFT-\>D-RL6k | 48.75 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | ThinkPrune(6k-\>3k) | 48.13 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | R1-distill+FEDH\* | 46.7 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | D-SFT-\>D-RL3k-\>D-RL6k | 45.42 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | D-SFT-\>D-RL3k | 43.96 | Not recorded | 69a66bcb2d603e6de3e3cb9e | | D-SFT | 9.17 | Not recorded | 69a66bcb2d603e6de3e3cb9e | ### D-SFT-\>D-RL3k-\>D-RL6k Accuracy: 64.79. Result ID: 69a89bbd6dbb7075b9517f44. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### REO-RL(Q-Spec)\* Accuracy: 64.4. Result ID: 69a89bbd6dbb7075b9517f4b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### Original Accuracy: 64.38. Result ID: 69a89bbd6dbb7075b9517f40. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### D-SFT-\>D-RL6k Accuracy: 63.54. Result ID: 69a89bbd6dbb7075b9517f47. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### D-SFT-\>D-RL6k Accuracy: 61.88. Result ID: 69a89bbd6dbb7075b9517f48. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"adaptive","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### R1-distill+DR.SAF\* Accuracy: 57.9. Result ID: 69a89bbd6dbb7075b9517f4e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### D-SFT-\>D-RL3k-\>D-RL6k Accuracy: 56.67. Result ID: 69a89bbd6dbb7075b9517f45. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"adaptive","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### R1-distill+Length-Penalty\* Accuracy: 54.7. Result ID: 69a89bbd6dbb7075b9517f4d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### ThinkPrune(6k) Accuracy: 53.96. Result ID: 69a89bbd6dbb7075b9517f49. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### D-SFT-\>D-RL6k Accuracy: 48.75. Result ID: 69a89bbd6dbb7075b9517f46. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"draft","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### ThinkPrune(6k-\>3k) Accuracy: 48.13. Result ID: 69a89bbd6dbb7075b9517f4a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### R1-distill+FEDH\* Accuracy: 46.7. Result ID: 69a89bbd6dbb7075b9517f4c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"step","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### D-SFT-\>D-RL3k-\>D-RL6k Accuracy: 45.42. Result ID: 69a89bbd6dbb7075b9517f43. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"draft","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### D-SFT-\>D-RL3k Accuracy: 43.96. Result ID: 69a89bbd6dbb7075b9517f42. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"draft","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. ### D-SFT Accuracy: 9.17. Result ID: 69a89bbd6dbb7075b9517f41. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Prompt":"draft","Backbone":"Qwen3-8B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.