# Visual Mathematical Reasoning on MathVista 1.0 (testmini) Source: [Visual Mathematical Reasoning on MathVista 1.0 (testmini)]() Collection index: [llms.txt]() Data updated: 2026-03-04T19:30:43.903Z Evaluation of model performance on the testmini split of the MathVista 1.0 dataset, focusing on Visual Mathematical Reasoning capabilities, measured by Accuracy. 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 | Visual Mathematical Reasoning | | Dataset (including any stated split) | MathVista 1.0 (testmini) | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 18 of 18 | | 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. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Qwen2.5-VL-7B | 67.6 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-7B + SFT+GRPO | 66.5 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-7B + SFT | 66.3 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-7B + SFT+DPO | 65.5 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-7B + SFT+DPO+GRPO | 65.2 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-7B + SPA-VL | 64 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-7B + SaFeR-VLM | 63.9 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B + SFT | 55.3 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B | 54.3 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B + SFT+DPO | 54 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B + VLGuard | 53.8 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B + SPA-VL | 53.3 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-7B + TIS | 53.2 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-7B + VLGuard | 52.7 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B + SaFeR-VLM | 52.6 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B + SFT+GRPO | 52 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B + SFT+DPO+GRPO | 51.9 | Not recorded | 69a7bc992d603e6de3e3d689 | | Qwen2.5-VL-3B + TIS | 45.6 | Not recorded | 69a7bc992d603e6de3e3d689 | ### Qwen2.5-VL-7B Accuracy: 67.6. Result ID: 69a888496dbb7075b951045c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + SFT+GRPO Accuracy: 66.5. Result ID: 69a888496dbb7075b9510463. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Stage":"SFT+GRPO"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + SFT Accuracy: 66.3. Result ID: 69a888496dbb7075b9510461. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Stage":"SFT"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + SFT+DPO Accuracy: 65.5. Result ID: 69a888496dbb7075b9510462. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Stage":"SFT+DPO"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + SFT+DPO+GRPO Accuracy: 65.2. Result ID: 69a888496dbb7075b9510464. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Stage":"SFT+DPO+GRPO"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + SPA-VL Accuracy: 64. Result ID: 69a888496dbb7075b951045f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Alignment":"SPA-VL"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + SaFeR-VLM Accuracy: 63.9. Result ID: 69a888496dbb7075b9510460. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Alignment":"SaFeR-VLM"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + SFT Accuracy: 55.3. Result ID: 69a888496dbb7075b9510458. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B","Stage":"SFT"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B Accuracy: 54.3. Result ID: 69a888496dbb7075b9510453. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + SFT+DPO Accuracy: 54. Result ID: 69a888496dbb7075b9510459. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B","Stage":"SFT+DPO"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + VLGuard Accuracy: 53.8. Result ID: 69a888496dbb7075b9510455. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B","Alignment":"VLGuard"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + SPA-VL Accuracy: 53.3. Result ID: 69a888496dbb7075b9510456. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B","Alignment":"SPA-VL"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + TIS Accuracy: 53.2. Result ID: 69a888496dbb7075b951045d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Alignment":"TIS"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B + VLGuard Accuracy: 52.7. Result ID: 69a888496dbb7075b951045e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-7B","Alignment":"VLGuard"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + SaFeR-VLM Accuracy: 52.6. Result ID: 69a888496dbb7075b9510457. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B","Alignment":"SaFeR-VLM"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + SFT+GRPO Accuracy: 52. Result ID: 69a888496dbb7075b951045a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B","Stage":"SFT+GRPO"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + SFT+DPO+GRPO Accuracy: 51.9. Result ID: 69a888496dbb7075b951045b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B","Stage":"SFT+DPO+GRPO"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B + TIS Accuracy: 45.6. Result ID: 69a888496dbb7075b9510454. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone":"Qwen2.5-VL-3B","Alignment":"TIS"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.