# Clinical Visual Question Answering on GMAI-MMBench standard MCQA Source: [Clinical Visual Question Answering on GMAI-MMBench standard MCQA]() Collection index: [llms.txt]() Data updated: 2026-06-16T10:22:17.836Z Evaluation of model performance on the GMAI-MMBench standard MCQA dataset for Clinical Visual Question Answering, reporting Accuracy, AUROC variants, and ECE variants. 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 | Clinical Visual Question Answering | | Dataset (including any stated split) | GMAI-MMBench standard MCQA | | Metric | Accuracy | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 12 of 12 | | 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. - [AUROC (Label NLL)](): Higher is better; unit: PERCENT. - [AUROC (ANLL)](): Higher is better; unit: PERCENT. - [AUROC (Max NLL)](): Higher is better; unit: PERCENT. - [AUROC (SC)](): Higher is better; unit: PERCENT. - [AUROC (SE)](): Higher is better; unit: PERCENT. - [AUROC (PRO)](): Higher is better; unit: PERCENT. - [AUROC (EE)](): Higher is better; unit: PERCENT. - [AUROC (RDS)](): Higher is better; unit: PERCENT. - [ECE (Label NLL)](): Lower is better; unit: PERCENT. - [ECE (ANLL)](): Lower is better; unit: PERCENT. - [ECE (Max NLL)](): Lower is better; unit: PERCENT. - [ECE (SC)](): Lower is better; unit: PERCENT. - [ECE (SE)](): Lower is better; unit: PERCENT. - [ECE (PRO)](): Lower is better; unit: PERCENT. - [ECE (EE)](): Lower is better; unit: PERCENT. - [ECE (RDS)](): Lower is better; unit: PERCENT. ## Recorded results | Method | Accuracy | Result date | Report paper | | --- | --- | --- | --- | | Qwen2.5-VL-72B | 47 | Not recorded | 6a30d8e3d33dfb23e783d011 | | Qwen2.5-VL-32B | 44.1 | Not recorded | 6a30d8e3d33dfb23e783d011 | | GPT-4.1-Mini | 43 | Not recorded | 6a30d8e3d33dfb23e783d011 | | LLaVA-OV-8B | 42.8 | Not recorded | 6a30d8e3d33dfb23e783d011 | | Qwen2-VL-7B | 42.4 | Not recorded | 6a30d8e3d33dfb23e783d011 | | Qwen2.5-VL-7B | 42.2 | Not recorded | 6a30d8e3d33dfb23e783d011 | | MedGemma-27B | 39.3 | Not recorded | 6a30d8e3d33dfb23e783d011 | | LLaVA-v1.6-34B | 39 | Not recorded | 6a30d8e3d33dfb23e783d011 | | MedGemma-4B | 37.2 | Not recorded | 6a30d8e3d33dfb23e783d011 | | LLaVA-v1.6-7B | 35.5 | Not recorded | 6a30d8e3d33dfb23e783d011 | | Molmo-72B | 35.2 | Not recorded | 6a30d8e3d33dfb23e783d011 | | Molmo-7B | 33.9 | Not recorded | 6a30d8e3d33dfb23e783d011 | ### Qwen2.5-VL-72B Accuracy: 47. Result ID: 6a3123c5fbc373592c2e2f19. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"Large (\>35B)"}. This record does not assert independent reproduction. ### Qwen2.5-VL-32B Accuracy: 44.1. Result ID: 6a3123c5fbc373592c2e2f16. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"Medium (8B–35B)"}. This record does not assert independent reproduction. ### GPT-4.1-Mini Accuracy: 43. Result ID: 6a3123c5fbc373592c2e2f1a. [Reporting paper]() Recorded evaluation configuration: {"Model Scale":"Large (\>35B)"}. This record does not assert independent reproduction. ### LLaVA-OV-8B Accuracy: 42.8. Result ID: 6a3123c5fbc373592c2e2f14. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"Small (≤8B)"}. This record does not assert independent reproduction. ### Qwen2-VL-7B Accuracy: 42.4. Result ID: 6a3123c5fbc373592c2e2f10. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"Small (≤8B)"}. This record does not assert independent reproduction. ### Qwen2.5-VL-7B Accuracy: 42.2. Result ID: 6a3123c5fbc373592c2e2f11. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"Small (≤8B)"}. This record does not assert independent reproduction. ### MedGemma-27B Accuracy: 39.3. Result ID: 6a3123c5fbc373592c2e2f15. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"Medium (8B–35B)"}. This record does not assert independent reproduction. ### LLaVA-v1.6-34B Accuracy: 39. Result ID: 6a3123c5fbc373592c2e2f17. [Reporting paper]() Recorded evaluation configuration: {"Model Scale":"Medium (8B–35B)"}. This record does not assert independent reproduction. ### MedGemma-4B Accuracy: 37.2. Result ID: 6a3123c5fbc373592c2e2f0f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"Small (≤8B)"}. This record does not assert independent reproduction. ### LLaVA-v1.6-7B Accuracy: 35.5. Result ID: 6a3123c5fbc373592c2e2f13. [Reporting paper]() Recorded evaluation configuration: {"Model Scale":"Small (≤8B)"}. This record does not assert independent reproduction. ### Molmo-72B Accuracy: 35.2. Result ID: 6a3123c5fbc373592c2e2f18. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"Large (\>35B)"}. This record does not assert independent reproduction. ### Molmo-7B Accuracy: 33.9. Result ID: 6a3123c5fbc373592c2e2f12. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"Small (≤8B)"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.