# Mathematical Reasoning on Math 500 (Flex Match, Exact Match, Avg. Gen. Length) Source: [Mathematical Reasoning on Math 500 (Flex Match, Exact Match, Avg. Gen. Length)]() Collection index: [llms.txt]() Data updated: 2026-02-27T11:58:19.266Z Evaluation of model performance on mathematical reasoning problems using the Math 500 dataset, reporting Flex Match, Exact Match accuracy, and average generation length. 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 | Flex Match | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 27 of 27 | | 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 - [Flex Match](): Higher is better; unit: PERCENT. - [Exact Match](): Higher is better; unit: PERCENT. - [Average Generation Length](): Higher is better; unit: SCALAR. ## Recorded results | Method | Flex Match | Result date | Report paper | | --- | --- | --- | --- | | QUOKA | 91.3 | Not recorded | 698abc35c97a0b3e3b5de4cc | | QUOKA | 91.1 | Not recorded | 698abc35c97a0b3e3b5de4cc | | Loki | 90.5 | Not recorded | 698abc35c97a0b3e3b5de4cc | | Loki | 90.3 | Not recorded | 698abc35c97a0b3e3b5de4cc | | GPT-OSS-20b | 89.3 | Not recorded | 698abc35c97a0b3e3b5de4cc | | LessIsMore | 85 | Not recorded | 698abc35c97a0b3e3b5de4cc | | SparQ | 84.8 | Not recorded | 698abc35c97a0b3e3b5de4cc | | QUOKA | 82.8 | Not recorded | 698abc35c97a0b3e3b5de4cc | | QUOKA | 82.3 | Not recorded | 698abc35c97a0b3e3b5de4cc | | Loki | 82.1 | Not recorded | 698abc35c97a0b3e3b5de4cc | | Qwen3-4B | 82 | Not recorded | 698abc35c97a0b3e3b5de4cc | | LessIsMore | 81.4 | Not recorded | 698abc35c97a0b3e3b5de4cc | | Loki | 78.5 | Not recorded | 698abc35c97a0b3e3b5de4cc | | LessIsMore | 78.2 | Not recorded | 698abc35c97a0b3e3b5de4cc | | SparQ | 77.7 | Not recorded | 698abc35c97a0b3e3b5de4cc | | SparQ | 74.4 | Not recorded | 698abc35c97a0b3e3b5de4cc | | LessIsMore | 72.9 | Not recorded | 698abc35c97a0b3e3b5de4cc | | SparQ | 72.7 | Not recorded | 698abc35c97a0b3e3b5de4cc | | SparQ | 66.5 | Not recorded | 698abc35c97a0b3e3b5de4cc | | QUOKA | 66.1 | Not recorded | 698abc35c97a0b3e3b5de4cc | | QUOKA | 66 | Not recorded | 698abc35c97a0b3e3b5de4cc | | Smollm3 | 65.9 | Not recorded | 698abc35c97a0b3e3b5de4cc | | LessIsMore | 65.7 | Not recorded | 698abc35c97a0b3e3b5de4cc | | Loki | 65.4 | Not recorded | 698abc35c97a0b3e3b5de4cc | | SparQ | 64.7 | Not recorded | 698abc35c97a0b3e3b5de4cc | | LessIsMore | 62.3 | Not recorded | 698abc35c97a0b3e3b5de4cc | | Loki | 61.5 | Not recorded | 698abc35c97a0b3e3b5de4cc | ### QUOKA Flex Match: 91.3. Result ID: 69a186c935dfcabb00e7a092. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"GPT-OSS-20b","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### QUOKA Flex Match: 91.1. Result ID: 69a186c935dfcabb00e7a091. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"GPT-OSS-20b","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### Loki Flex Match: 90.5. Result ID: 69a186c935dfcabb00e7a08d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"GPT-OSS-20b","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### Loki Flex Match: 90.3. Result ID: 69a186c935dfcabb00e7a08e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"GPT-OSS-20b","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### GPT-OSS-20b Flex Match: 89.3. Result ID: 69a186c935dfcabb00e7a08a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Budget":"Dense","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### LessIsMore Flex Match: 85. Result ID: 69a186c935dfcabb00e7a090. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"GPT-OSS-20b","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### SparQ Flex Match: 84.8. Result ID: 69a186c935dfcabb00e7a08c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"GPT-OSS-20b","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### QUOKA Flex Match: 82.8. Result ID: 69a186c935dfcabb00e7a09a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### QUOKA Flex Match: 82.3. Result ID: 69a186c935dfcabb00e7a09b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### Loki Flex Match: 82.1. Result ID: 69a186c935dfcabb00e7a097. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### Qwen3-4B Flex Match: 82. Result ID: 69a186c935dfcabb00e7a093. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Budget":"Dense","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### LessIsMore Flex Match: 81.4. Result ID: 69a186c935dfcabb00e7a099. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### Loki Flex Match: 78.5. Result ID: 69a186c935dfcabb00e7a096. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### LessIsMore Flex Match: 78.2. Result ID: 69a186c935dfcabb00e7a08f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"GPT-OSS-20b","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### SparQ Flex Match: 77.7. Result ID: 69a186c935dfcabb00e7a095. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### SparQ Flex Match: 74.4. Result ID: 69a186c935dfcabb00e7a08b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"GPT-OSS-20b","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### LessIsMore Flex Match: 72.9. Result ID: 69a186c935dfcabb00e7a098. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### SparQ Flex Match: 72.7. Result ID: 69a186c935dfcabb00e7a094. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-4B","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### SparQ Flex Match: 66.5. Result ID: 69a186c935dfcabb00e7a09e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Smollm3","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### QUOKA Flex Match: 66.1. Result ID: 69a186c935dfcabb00e7a0a4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Smollm3","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### QUOKA Flex Match: 66. Result ID: 69a186c935dfcabb00e7a0a3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Smollm3","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### Smollm3 Flex Match: 65.9. Result ID: 69a186c935dfcabb00e7a09c. [Reporting paper]() Recorded evaluation configuration: {"Budget":"Dense","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### LessIsMore Flex Match: 65.7. Result ID: 69a186c935dfcabb00e7a0a2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Smollm3","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### Loki Flex Match: 65.4. Result ID: 69a186c935dfcabb00e7a0a0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Smollm3","Budget":"256","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### SparQ Flex Match: 64.7. Result ID: 69a186c935dfcabb00e7a09d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Smollm3","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### LessIsMore Flex Match: 62.3. Result ID: 69a186c935dfcabb00e7a0a1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Smollm3","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. ### Loki Flex Match: 61.5. Result ID: 69a186c935dfcabb00e7a09f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Smollm3","Budget":"128","Reasoning generation length limit":"8192 tokens"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.