# Mathematical Reasoning on MathVista (Accuracy %) Source: [Mathematical Reasoning on MathVista (Accuracy %)]() Collection index: [llms.txt]() Data updated: 2026-06-01T08:31:53.179Z Evaluation of model performance on the MathVista dataset for Mathematical Reasoning, reporting results based on overall Accuracy percentage. 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 | | Metric | Accuracy (%) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 29 of 29 | | 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 | | --- | --- | --- | --- | | EASE-7B | 75.2 | Not recorded | 6a1d11e2dbd7150506f83dde | | VGPO-7B | 74.1 | Not recorded | 6a1d11e2dbd7150506f83dde | | ThinkLite-VL-7B | 73.6 | Not recorded | 6a1d11e2dbd7150506f83dde | | NoisyRollout-7B | 72.9 | Not recorded | 6a1d11e2dbd7150506f83dde | | PAPO\_D-7B | 72.3 | Not recorded | 6a1d11e2dbd7150506f83dde | | Vision-Zero(CLEVR) | 70.6 | Not recorded | 6a15293ddbd7150506f7e0f7 | | VPPO-RL-7B | 70.5 | Not recorded | 6a1d11e2dbd7150506f83dde | | DUEL | 69.8 | Not recorded | 6a15293ddbd7150506f7e0f7 | | EvoLMM | 69.7 | Not recorded | 6a15293ddbd7150506f7e0f7 | | IVR-R1 | 69.6 | Not recorded | 6a152902dbd7150506f7ddd9 | | MM-UPT | 69.4 | Not recorded | 6a15293ddbd7150506f7e0f7 | | OpenVLThinker-7B | 68.9 | Not recorded | 6a15293ddbd7150506f7e0f7 | | Qwen3-VL-32B-Instruct | 68.7 | Not recorded | 6a152902dbd7150506f7ddd9 | | DUEL | 68.6 | Not recorded | 6a15293ddbd7150506f7e0f7 | | Qwen2.5-VL-7B | 68.4 | Not recorded | 6a15293ddbd7150506f7e0f7 | | Vision-R1 | 68.3 | Not recorded | 6a152902dbd7150506f7ddd9 | | Vision-SR1 | 68.2 | Not recorded | 6a152902dbd7150506f7ddd9 | | VLAA-Thinker-7B | 68.2 | Not recorded | 6a15293ddbd7150506f7e0f7 | | IVR-R1 | 67.9 | Not recorded | 6a152902dbd7150506f7ddd9 | | Vision-R1 | 67.4 | Not recorded | 6a152902dbd7150506f7ddd9 | | MM-Eureka-7B | 66.3 | Not recorded | 6a1d11e2dbd7150506f83dde | | Supervised Fine-tuning (before RL) | 65.6 | Not recorded | 6a152902dbd7150506f7ddd9 | | VL-Rethinker-7B | 65.2 | Not recorded | 6a1d11e2dbd7150506f83dde | | Supervised Fine-tuning (before RL) | 63.6 | Not recorded | 6a152902dbd7150506f7ddd9 | | Vision-SR1 | 63.5 | Not recorded | 6a152902dbd7150506f7ddd9 | | Zero-shot Inference (before RL) | 54.5 | Not recorded | 6a152902dbd7150506f7ddd9 | | Qwen2.5-VL-72B-Instruct | 48.1 | Not recorded | 6a152902dbd7150506f7ddd9 | | Qwen2.5-VL-32B-Instruct | 37.2 | Not recorded | 6a152902dbd7150506f7ddd9 | | Zero-shot Inference (before RL) | 32.6 | Not recorded | 6a152902dbd7150506f7ddd9 | ### EASE-7B Accuracy (%): 75.2. Result ID: 6a1d4366fbc373592c1675ae. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### VGPO-7B Accuracy (%): 74.1. Result ID: 6a1d4366fbc373592c1675ad. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### ThinkLite-VL-7B Accuracy (%): 73.6. Result ID: 6a1d4366fbc373592c1675a7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### NoisyRollout-7B Accuracy (%): 72.9. Result ID: 6a1d4366fbc373592c1675aa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### PAPO\_D-7B Accuracy (%): 72.3. Result ID: 6a1d4366fbc373592c1675ab. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### Vision-Zero(CLEVR) Accuracy (%): 70.6. Result ID: 6a15648cfbc373592c0d1079. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### VPPO-RL-7B Accuracy (%): 70.5. Result ID: 6a1d4366fbc373592c1675ac. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### DUEL Accuracy (%): 69.8. Result ID: 6a15648cfbc373592c0d107e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Role":"Solver"}. This record does not assert independent reproduction. ### EvoLMM Accuracy (%): 69.7. Result ID: 6a15648cfbc373592c0d107a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### IVR-R1 Accuracy (%): 69.6. Result ID: 6a154936fbc373592c0c38c5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen3-VL-4B"}. This record does not assert independent reproduction. ### MM-UPT Accuracy (%): 69.4. Result ID: 6a15648cfbc373592c0d1078. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### OpenVLThinker-7B Accuracy (%): 68.9. Result ID: 6a15648cfbc373592c0d107c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Qwen3-VL-32B-Instruct Accuracy (%): 68.7. Result ID: 6a154936fbc373592c0c38bb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen3-VL-32B-Instruct"}. This record does not assert independent reproduction. ### DUEL Accuracy (%): 68.6. Result ID: 6a15648cfbc373592c0d107d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Role":"Challenger"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B Accuracy (%): 68.4. Result ID: 6a15648cfbc373592c0d1077. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Vision-R1 Accuracy (%): 68.3. Result ID: 6a154936fbc373592c0c38c3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen3-VL-4B"}. This record does not assert independent reproduction. ### Vision-SR1 Accuracy (%): 68.2. Result ID: 6a154936fbc373592c0c38c4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen3-VL-4B"}. This record does not assert independent reproduction. ### VLAA-Thinker-7B Accuracy (%): 68.2. Result ID: 6a15648cfbc373592c0d107b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### IVR-R1 Accuracy (%): 67.9. Result ID: 6a154936fbc373592c0c38c0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction. ### Vision-R1 Accuracy (%): 67.4. Result ID: 6a154936fbc373592c0c38be. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction. ### MM-Eureka-7B Accuracy (%): 66.3. Result ID: 6a1d4366fbc373592c1675a9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### Supervised Fine-tuning (before RL) Accuracy (%): 65.6. Result ID: 6a154936fbc373592c0c38c2. [Reporting paper]() Recorded evaluation configuration: {"Backbone model":"Qwen3-VL-4B"}. This record does not assert independent reproduction. ### VL-Rethinker-7B Accuracy (%): 65.2. Result ID: 6a1d4366fbc373592c1675a8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"7B"}. This record does not assert independent reproduction. ### Supervised Fine-tuning (before RL) Accuracy (%): 63.6. Result ID: 6a154936fbc373592c0c38bd. [Reporting paper]() Recorded evaluation configuration: {"Backbone model":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction. ### Vision-SR1 Accuracy (%): 63.5. Result ID: 6a154936fbc373592c0c38bf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction. ### Zero-shot Inference (before RL) Accuracy (%): 54.5. Result ID: 6a154936fbc373592c0c38c1. [Reporting paper]() Recorded evaluation configuration: {"Backbone model":"Qwen3-VL-4B"}. This record does not assert independent reproduction. ### Qwen2.5-VL-72B-Instruct Accuracy (%): 48.1. Result ID: 6a154936fbc373592c0c38b9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen2.5-VL-72B-Instruct"}. This record does not assert independent reproduction. ### Qwen2.5-VL-32B-Instruct Accuracy (%): 37.2. Result ID: 6a154936fbc373592c0c38ba. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Backbone model":"Qwen2.5-VL-32B-Instruct"}. This record does not assert independent reproduction. ### Zero-shot Inference (before RL) Accuracy (%): 32.6. Result ID: 6a154936fbc373592c0c38bc. [Reporting paper]() Recorded evaluation configuration: {"Backbone model":"Qwen2.5-VL-7B"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.