# Long-generation reasoning on MATH500 (pass@1) Source: [Long-generation reasoning on MATH500 (pass@1)]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:11:15.254Z Evaluation of model performance on the MATH500 dataset specifically for long-form generation reasoning tasks, measured using the pass@1 metric. 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 | Long-generation reasoning | | Dataset (including any stated split) | MATH500 | | Metric | pass@1 | | 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 - [pass@1](): Higher is better; unit: PERCENT. ## Recorded results | Method | pass@1 | Result date | Report paper | | --- | --- | --- | --- | | ParisKV | 0.93 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | ParisKV | 0.928 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | Qwen-3-4B | 0.886 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | Qwen-3-8B | 0.874 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | ParisKV | 0.844 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | DS-R1-Llam-8B | 0.804 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | PQCache | 0.6921 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | PQCache | 0.6675 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | PQCache | 0.588 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | MagicPIG | 0.49 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | MagicPIG | 0.464 | Not recorded | 698abbfbc97a0b3e3b5de2ab | | MagicPIG | 0.458 | Not recorded | 698abbfbc97a0b3e3b5de2ab | ### ParisKV pass@1: 0.93. Result ID: 6998dff35698665ec444202a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen-3-8B"}. This record does not assert independent reproduction. ### ParisKV pass@1: 0.928. Result ID: 6998dff35698665ec4442022. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen-3-4B"}. This record does not assert independent reproduction. ### Qwen-3-4B pass@1: 0.886. Result ID: 6998dff35698665ec444201f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Attention":"Full attention"}. This record does not assert independent reproduction. ### Qwen-3-8B pass@1: 0.874. Result ID: 6998dff35698665ec4442027. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Attention":"Full attention"}. This record does not assert independent reproduction. ### ParisKV pass@1: 0.844. Result ID: 6998dff35698665ec4442026. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"DS-R1-Llam-8B"}. This record does not assert independent reproduction. ### DS-R1-Llam-8B pass@1: 0.804. Result ID: 6998dff35698665ec4442023. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Attention":"Full attention"}. This record does not assert independent reproduction. ### PQCache pass@1: 0.6921. Result ID: 6998dff35698665ec4442028. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen-3-8B"}. This record does not assert independent reproduction. ### PQCache pass@1: 0.6675. Result ID: 6998dff35698665ec4442024. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"DS-R1-Llam-8B"}. This record does not assert independent reproduction. ### PQCache pass@1: 0.588. Result ID: 6998dff35698665ec4442020. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen-3-4B"}. This record does not assert independent reproduction. ### MagicPIG pass@1: 0.49. Result ID: 6998dff35698665ec4442025. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"DS-R1-Llam-8B"}. This record does not assert independent reproduction. ### MagicPIG pass@1: 0.464. Result ID: 6998dff35698665ec4442021. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen-3-4B"}. This record does not assert independent reproduction. ### MagicPIG pass@1: 0.458. Result ID: 6998dff35698665ec4442029. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen-3-8B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.