# Mathematical Reasoning on AIME 25 (Accuracy, Tokens, Token Reduction %) Source: [Mathematical Reasoning on AIME 25 (Accuracy, Tokens, Token Reduction %)]() Collection index: [llms.txt]() Data updated: 2026-07-01T08:01:56.494Z Evaluation of model performance on the AIME 25 dataset for mathematical reasoning, reporting results based on Accuracy, Token count, and Token Reduction percentage. 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 25 | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | 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](): Lower is better; unit: SCALAR. - [Token Reduction (%)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Parallel Thinking | 83.3 | Not recorded | 6a449f12e0224bd4f49930df | | Fork-think | 82.4 | Not recorded | 6a449f12e0224bd4f49930df | | Greedy | 80 | Not recorded | 6a449f12e0224bd4f49930df | | Fork-think | 75.6 | Not recorded | 6a449f12e0224bd4f49930df | | Parallel Thinking | 75.4 | Not recorded | 6a449f12e0224bd4f49930df | | CoT | 75.3 | Not recorded | 6a449f12e0224bd4f49930df | | Fork-think | 66.6 | Not recorded | 6a449f12e0224bd4f49930df | | CoT | 64 | Not recorded | 6a449f12e0224bd4f49930df | | Parallel Thinking | 62.6 | Not recorded | 6a449f12e0224bd4f49930df | | Greedy | 60 | Not recorded | 6a449f12e0224bd4f49930df | | CoT | 58.7 | Not recorded | 6a449f12e0224bd4f49930df | | Greedy | 46.6 | Not recorded | 6a449f12e0224bd4f49930df | ### Parallel Thinking Accuracy: 83.3. Result ID: 6a44c97373d097d845bef7c5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"DeepSeek-8B","Decoding Strategy":"Parallel Thinking","Branching factor (n)":"32"}. This record does not assert independent reproduction. ### Fork-think Accuracy: 82.4. Result ID: 6a44c97373d097d845bef7c6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"DeepSeek-8B","Decoding Strategy":"Fork-think","Branching factor (n)":"32","Seed path length (l0)":"2,048"}. This record does not assert independent reproduction. ### Greedy Accuracy: 80. Result ID: 6a44c97373d097d845bef7c3. [Reporting paper]() Recorded evaluation configuration: {"Model":"DeepSeek-8B","Decoding Strategy":"Greedy"}. This record does not assert independent reproduction. ### Fork-think Accuracy: 75.6. Result ID: 6a44c97373d097d845bef7c2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-8B","Decoding Strategy":"Fork-think","Branching factor (n)":"32","Seed path length (l0)":"2,048"}. This record does not assert independent reproduction. ### Parallel Thinking Accuracy: 75.4. Result ID: 6a44c97373d097d845bef7c1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-8B","Decoding Strategy":"Parallel Thinking","Branching factor (n)":"32"}. This record does not assert independent reproduction. ### CoT Accuracy: 75.3. Result ID: 6a44c97373d097d845bef7c4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"DeepSeek-8B","Decoding Strategy":"CoT"}. This record does not assert independent reproduction. ### Fork-think Accuracy: 66.6. Result ID: 6a44c97373d097d845bef7ca. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Phi-4-RP-14B","Decoding Strategy":"Fork-think","Branching factor (n)":"32","Seed path length (l0)":"2,048"}. This record does not assert independent reproduction. ### CoT Accuracy: 64. Result ID: 6a44c97373d097d845bef7c0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-8B","Decoding Strategy":"CoT"}. This record does not assert independent reproduction. ### Parallel Thinking Accuracy: 62.6. Result ID: 6a44c97373d097d845bef7c9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Phi-4-RP-14B","Decoding Strategy":"Parallel Thinking","Branching factor (n)":"32"}. This record does not assert independent reproduction. ### Greedy Accuracy: 60. Result ID: 6a44c97373d097d845bef7bf. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-8B","Decoding Strategy":"Greedy"}. This record does not assert independent reproduction. ### CoT Accuracy: 58.7. Result ID: 6a44c97373d097d845bef7c8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Phi-4-RP-14B","Decoding Strategy":"CoT"}. This record does not assert independent reproduction. ### Greedy Accuracy: 46.6. Result ID: 6a44c97373d097d845bef7c7. [Reporting paper]() Recorded evaluation configuration: {"Model":"Phi-4-RP-14B","Decoding Strategy":"Greedy"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.