# Mathematical Reasoning on MATH 500 (Avg@4) Source: [Mathematical Reasoning on MATH 500 (Avg@4)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:36:32.298Z Evaluation of model performance on the MATH 500 dataset for mathematical reasoning, specifically reporting the Avg@4 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 | Mathematical Reasoning | | Dataset (including any stated split) | MATH 500 | | Metric | Avg@4 | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 30 of 30 | | 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 - [Avg@4](): Higher is better; unit: PERCENT. ## Recorded results | Method | Avg@4 | Result date | Report paper | | --- | --- | --- | --- | | GSPO + CBS | 83.1 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GSPO + CBS\* | 79.65 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GSPO | 79.2 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO + CBS | 78.25 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GSPO + CBS\* | 77.85 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO | 77.65 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GSPO | 77.55 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO + CBS | 77.45 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO + CBS\* | 77.45 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO + CBS\* | 76.85 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO + CBS\* | 76.3 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO | 76.15 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GSPO + CBS | 76.05 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO + CBS | 75.85 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO + CBS\* | 75.55 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO | 75.05 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO + CBS | 74.45 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO | 73.7 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GSPO + CBS\* | 67.5 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GSPO | 65.95 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GSPO + CBS | 65.7 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO + CBS\* | 64.8 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO + CBS | 63.9 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO + CBS | 63.65 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO + CBS\* | 62.45 | Not recorded | 698abc26c97a0b3e3b5de3ed | | GRPO | 60.95 | Not recorded | 698abc26c97a0b3e3b5de3ed | | DAPO | 58.7 | Not recorded | 698abc26c97a0b3e3b5de3ed | | Base Model | 53.25 | Not recorded | 698abc26c97a0b3e3b5de3ed | | Base Model | 31.9 | Not recorded | 698abc26c97a0b3e3b5de3ed | | Base Model | 29.1 | Not recorded | 698abc26c97a0b3e3b5de3ed | ### GSPO + CBS Avg@4: 83.1. Result ID: 6998d8605698665ec43c459b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GSPO + CBS\* Avg@4: 79.65. Result ID: 6998d8605698665ec43c459c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GSPO Avg@4: 79.2. Result ID: 6998d8605698665ec43c459a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### DAPO + CBS Avg@4: 78.25. Result ID: 6998d8605698665ec43c4598. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GSPO + CBS\* Avg@4: 77.85. Result ID: 6998d8605698665ec43c4592. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### DAPO Avg@4: 77.65. Result ID: 6998d8605698665ec43c4597. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GSPO Avg@4: 77.55. Result ID: 6998d8605698665ec43c4590. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### GRPO + CBS Avg@4: 77.45. Result ID: 6998d8605698665ec43c4595. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GRPO + CBS\* Avg@4: 77.45. Result ID: 6998d8605698665ec43c4596. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### DAPO + CBS\* Avg@4: 76.85. Result ID: 6998d8605698665ec43c4599. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### DAPO + CBS\* Avg@4: 76.3. Result ID: 6998d8605698665ec43c458f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### GRPO Avg@4: 76.15. Result ID: 6998d8605698665ec43c4594. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GSPO + CBS Avg@4: 76.05. Result ID: 6998d8605698665ec43c4591. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### DAPO + CBS Avg@4: 75.85. Result ID: 6998d8605698665ec43c458e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### GRPO + CBS\* Avg@4: 75.55. Result ID: 6998d8605698665ec43c458c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### DAPO Avg@4: 75.05. Result ID: 6998d8605698665ec43c458d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### GRPO + CBS Avg@4: 74.45. Result ID: 6998d8605698665ec43c458b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### GRPO Avg@4: 73.7. Result ID: 6998d8605698665ec43c458a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### GSPO + CBS\* Avg@4: 67.5. Result ID: 6998d8605698665ec43c4588. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### GSPO Avg@4: 65.95. Result ID: 6998d8605698665ec43c4586. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### GSPO + CBS Avg@4: 65.7. Result ID: 6998d8605698665ec43c4587. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### GRPO + CBS\* Avg@4: 64.8. Result ID: 6998d8605698665ec43c4582. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### GRPO + CBS Avg@4: 63.9. Result ID: 6998d8605698665ec43c4581. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### DAPO + CBS Avg@4: 63.65. Result ID: 6998d8605698665ec43c4584. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### DAPO + CBS\* Avg@4: 62.45. Result ID: 6998d8605698665ec43c4585. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### GRPO Avg@4: 60.95. Result ID: 6998d8605698665ec43c4580. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### DAPO Avg@4: 58.7. Result ID: 6998d8605698665ec43c4583. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### Base Model Avg@4: 53.25. Result ID: 6998d8605698665ec43c4593. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### Base Model Avg@4: 31.9. Result ID: 6998d8605698665ec43c4589. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-4B-Base"}. This record does not assert independent reproduction. ### Base Model Avg@4: 29.1. Result ID: 6998d8605698665ec43c457f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.