# Medical Visual Question Answering on Med-R2 (test) Source: [Medical Visual Question Answering on Med-R2 (test)]() Collection index: [llms.txt]() Data updated: 2026-05-26T08:38:38.116Z Evaluation of model performance on the test split of the Med-R2 dataset for Medical Visual Question Answering, utilizing multi-faceted accuracy metrics. 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 | Medical Visual Question Answering | | Dataset (including any stated split) | Med-R2 (test) | | Metric | Image Quality ACC | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 14 of 14 | | 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 - [Image Quality ACC](): Higher is better; unit: PERCENT. - [Anatomy Level ACC](): Higher is better; unit: PERCENT. - [Lesion Level ACC](): Higher is better; unit: PERCENT. - [Joint QA ACC](): Higher is better; unit: PERCENT. - [Overall Average Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | Image Quality ACC | Result date | Report paper | | --- | --- | --- | --- | | Janus-Pro-7B | 74.6 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Linshu-7B | 67.3 | Not recorded | 6a15292fdbd7150506f7dfc9 | | GPT-5.2-thinking | 63.7 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Intern3-VL3-8B | 58.3 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Qwen3-VL-235b | 54.2 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Qwen3-VL-8B | 53.3 | Not recorded | 6a15292fdbd7150506f7dfc9 | | GPT-4o | 53 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Qwen2.5-VL-7B | 49.3 | Not recorded | 6a15292fdbd7150506f7dfc9 | | HuatuoGPT-Vision-34B | 48 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Hulu-Med-14B | 47.4 | Not recorded | 6a15292fdbd7150506f7dfc9 | | HuatuoGPT-Vision-7B | 41.9 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Hulu-Med-4B | 41.4 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Hulu-Med-7B | 36.5 | Not recorded | 6a15292fdbd7150506f7dfc9 | | Medgemma-4B | 24.2 | Not recorded | 6a15292fdbd7150506f7dfc9 | ### Janus-Pro-7B Image Quality ACC: 74.6. Result ID: 6a155c07fbc373592c0cce81. [Reporting paper]() Recorded evaluation configuration: {"Model Category":"Open Source","Zero-shot":"true"}. This record does not assert independent reproduction. ### Linshu-7B Image Quality ACC: 67.3. Result ID: 6a155c07fbc373592c0cce7d. [Reporting paper]() Recorded evaluation configuration: {"Model Category":"Medical Specific","Zero-shot":"true"}. This record does not assert independent reproduction. ### GPT-5.2-thinking Image Quality ACC: 63.7. Result ID: 6a155c07fbc373592c0cce86. [Reporting paper]() Recorded evaluation configuration: {"Model Category":"Proprietary","Zero-shot":"true"}. This record does not assert independent reproduction. ### Intern3-VL3-8B Image Quality ACC: 58.3. Result ID: 6a155c07fbc373592c0cce82. [Reporting paper]() Recorded evaluation configuration: {"Model Category":"Open Source","Zero-shot":"true"}. This record does not assert independent reproduction. ### Qwen3-VL-235b Image Quality ACC: 54.2. Result ID: 6a155c07fbc373592c0cce87. [Reporting paper]() Recorded evaluation configuration: {"Model Category":"Proprietary","Zero-shot":"true"}. This record does not assert independent reproduction. ### Qwen3-VL-8B Image Quality ACC: 53.3. Result ID: 6a155c07fbc373592c0cce83. [Reporting paper]() Recorded evaluation configuration: {"Model Category":"Open Source","Zero-shot":"true"}. This record does not assert independent reproduction. ### GPT-4o Image Quality ACC: 53. Result ID: 6a155c07fbc373592c0cce85. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Category":"Proprietary","Zero-shot":"true"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B Image Quality ACC: 49.3. Result ID: 6a155c07fbc373592c0cce84. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Category":"Open Source","Zero-shot":"true"}. This record does not assert independent reproduction. ### HuatuoGPT-Vision-34B Image Quality ACC: 48. Result ID: 6a155c07fbc373592c0cce80. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Category":"Medical Specific","Zero-shot":"true"}. This record does not assert independent reproduction. ### Hulu-Med-14B Image Quality ACC: 47.4. Result ID: 6a155c07fbc373592c0cce7c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Category":"Medical Specific","Zero-shot":"true"}. This record does not assert independent reproduction. ### HuatuoGPT-Vision-7B Image Quality ACC: 41.9. Result ID: 6a155c07fbc373592c0cce7f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Category":"Medical Specific","Zero-shot":"true"}. This record does not assert independent reproduction. ### Hulu-Med-4B Image Quality ACC: 41.4. Result ID: 6a155c07fbc373592c0cce7a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Category":"Medical Specific","Zero-shot":"true"}. This record does not assert independent reproduction. ### Hulu-Med-7B Image Quality ACC: 36.5. Result ID: 6a155c07fbc373592c0cce7b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Category":"Medical Specific","Zero-shot":"true"}. This record does not assert independent reproduction. ### Medgemma-4B Image Quality ACC: 24.2. Result ID: 6a155c07fbc373592c0cce7e. [Reporting paper]() Recorded evaluation configuration: {"Model Category":"Medical Specific","Zero-shot":"true"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.