# Mathematical Reasoning on MathVista Math (ALL, GEO, ALG, GPS, TQA) Source: [Mathematical Reasoning on MathVista Math (ALL, GEO, ALG, GPS, TQA)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:22:29.955Z Evaluation of model performance on the MathVista Math dataset for Mathematical Reasoning, reporting scores across overall (ALL) and specific sub-tasks including Geometry (GEO), Algebra (ALG), Geometry Problem Solving (GPS), and Textual Question Answering (TQA). 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) | MathVista Math | | Metric | ALL Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 19 of 19 | | 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 - [ALL Accuracy](): Higher is better; unit: PERCENT. - [GEO Accuracy](): Higher is better; unit: PERCENT. - [ALG Accuracy](): Higher is better; unit: PERCENT. - [GPS Accuracy](): Higher is better; unit: PERCENT. - [TQA Accuracy](): Higher is better; unit: PERCENT. ## Recorded results | Method | ALL Accuracy | Result date | Report paper | | --- | --- | --- | --- | | VIRC-7B | 81.11 | Not recorded | 694239304561469362850763 | | InternVL2.5-8B-MPO | 75.93 | Not recorded | 694239304561469362850763 | | MINT-CoT-7B | 73.7 | Not recorded | 694239304561469362850763 | | MM-Eureka | 72.59 | Not recorded | 694239304561469362850763 | | VIRC-3B | 72.59 | Not recorded | 694239304561469362850763 | | DeepEyes | 71.11 | Not recorded | 694239304561469362850763 | | R1-VL-7B | 69.63 | Not recorded | 694239304561469362850763 | | Mini-o3 | 67.78 | Not recorded | 694239304561469362850763 | | Claude 3.5 Sonnet | 67.41 | Not recorded | 694239304561469362850763 | | LLaVA-OV-Qwen2-7b-ov | 67.04 | Not recorded | 694239304561469362850763 | | GPT-4o | 66.67 | Not recorded | 694239304561469362850763 | | Pixel Reasoner | 66.67 | Not recorded | 694239304561469362850763 | | Qwen2.5-VL-7B-Instruct | 66.67 | Not recorded | 694239304561469362850763 | | DeepSeek-VL2 | 65.56 | Not recorded | 694239304561469362850763 | | InternVL2.5-8B | 62.59 | Not recorded | 694239304561469362850763 | | Qwen3-VL-8B-Instruct | 59.63 | Not recorded | 694239304561469362850763 | | Hint-GRPO-Qwen2.5-VL-3B | 59.26 | Not recorded | 694239304561469362850763 | | Qwen2.5-VL-3B-Instruct | 55.93 | Not recorded | 694239304561469362850763 | | Hint-GRPO-Qwen2-VL-7B | 53.7 | Not recorded | 694239304561469362850763 | ### VIRC-7B ALL Accuracy: 81.11. Result ID: 69983bbb5698665ec497c519. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### InternVL2.5-8B-MPO ALL Accuracy: 75.93. Result ID: 69983bbb5698665ec497c50c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"8B","Model Category":"Open-Source General Model"}. This record does not assert independent reproduction. ### MINT-CoT-7B ALL Accuracy: 73.7. Result ID: 69983bbb5698665ec497c515. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### MM-Eureka ALL Accuracy: 72.59. Result ID: 69983bbb5698665ec497c514. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### VIRC-3B ALL Accuracy: 72.59. Result ID: 69983bbb5698665ec497c517. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"3B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### DeepEyes ALL Accuracy: 71.11. Result ID: 69983bbb5698665ec497c50f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### R1-VL-7B ALL Accuracy: 69.63. Result ID: 69983bbb5698665ec497c513. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### Mini-o3 ALL Accuracy: 67.78. Result ID: 69983bbb5698665ec497c510. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### Claude 3.5 Sonnet ALL Accuracy: 67.41. Result ID: 69983bbb5698665ec497c508. [Reporting paper]() Recorded evaluation configuration: {"#Params":"-","Model Category":"Closed-Source Model"}. This record does not assert independent reproduction. ### LLaVA-OV-Qwen2-7b-ov ALL Accuracy: 67.04. Result ID: 69983bbb5698665ec497c50a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source General Model"}. This record does not assert independent reproduction. ### GPT-4o ALL Accuracy: 66.67. Result ID: 69983bbb5698665ec497c507. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"-","Model Category":"Closed-Source Model"}. This record does not assert independent reproduction. ### Pixel Reasoner ALL Accuracy: 66.67. Result ID: 69983bbb5698665ec497c50e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B-Instruct ALL Accuracy: 66.67. Result ID: 69983bbb5698665ec497c518. [Reporting paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### DeepSeek-VL2 ALL Accuracy: 65.56. Result ID: 69983bbb5698665ec497c50d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"4.5B","Model Category":"Open-Source General Model"}. This record does not assert independent reproduction. ### InternVL2.5-8B ALL Accuracy: 62.59. Result ID: 69983bbb5698665ec497c50b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"#Params":"8B","Model Category":"Open-Source General Model"}. This record does not assert independent reproduction. ### Qwen3-VL-8B-Instruct ALL Accuracy: 59.63. Result ID: 69983bbb5698665ec497c509. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"8B","Model Category":"Open-Source General Model","tool":true}. This record does not assert independent reproduction. ### Hint-GRPO-Qwen2.5-VL-3B ALL Accuracy: 59.26. Result ID: 69983bbb5698665ec497c511. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"3B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### Qwen2.5-VL-3B-Instruct ALL Accuracy: 55.93. Result ID: 69983bbb5698665ec497c516. [Reporting paper]() Recorded evaluation configuration: {"#Params":"3B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. ### Hint-GRPO-Qwen2-VL-7B ALL Accuracy: 53.7. Result ID: 69983bbb5698665ec497c512. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"#Params":"7B","Model Category":"Open-Source Reasoning Model"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.