# Multi-image Medical Visual Question Answering on MIM-ODIR (Held-out) Source: [Multi-image Medical Visual Question Answering on MIM-ODIR (Held-out)]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:08:03.918Z Evaluation of model performance on the held-out portion of the MIM-ODIR dataset for multi-image medical visual question answering, reporting Close (C) and Open (O) scores. 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 | Multi-image Medical Visual Question Answering | | Dataset (including any stated split) | MIM-ODIR (Held-out) | | Metric | VQA Close Accuracy (C) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 10 of 10 | | 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 - [VQA Close Accuracy (C)](): Higher is better; unit: PERCENT. - [VQA Open Accuracy (O)](): Higher is better; unit: PERCENT. ## Recorded results | Method | VQA Close Accuracy (C) | Result date | Report paper | | --- | --- | --- | --- | | GPT-4o | 45.67 | Not recorded | 68fa78c01e209131cea7a40c | | MIM-LLAVA-Med | 32.33 | Not recorded | 68fa78c01e209131cea7a40c | | InternVL2-8B | 31.67 | Not recorded | 68fa78c01e209131cea7a40c | | Med-Mantis | 30 | Not recorded | 68fa78c01e209131cea7a40c | | LLaVA-Med-7B | 29.67 | Not recorded | 68fa78c01e209131cea7a40c | | Mantis-8B | 23 | Not recorded | 68fa78c01e209131cea7a40c | | deepseek-VL-7B | 21.67 | Not recorded | 68fa78c01e209131cea7a40c | | Med-Flamingo-9B | 16 | Not recorded | 68fa78c01e209131cea7a40c | | Mantis† | 7 | Not recorded | 68fa78c01e209131cea7a40c | | LLaVA-Med† | 2.66 | Not recorded | 68fa78c01e209131cea7a40c | ### GPT-4o VQA Close Accuracy (C): 45.67. Result ID: 69959fae3badfaffe69c196b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### MIM-LLAVA-Med VQA Close Accuracy (C): 32.33. Result ID: 69959fae3badfaffe69c1971. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_dataset":"Med-MIM instruction dataset"}. This record does not assert independent reproduction. ### InternVL2-8B VQA Close Accuracy (C): 31.67. Result ID: 69959fae3badfaffe69c196e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Med-Mantis VQA Close Accuracy (C): 30. Result ID: 69959fae3badfaffe69c1973. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_dataset":"Med-MIM instruction dataset"}. This record does not assert independent reproduction. ### LLaVA-Med-7B VQA Close Accuracy (C): 29.67. Result ID: 69959fae3badfaffe69c196f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Mantis-8B VQA Close Accuracy (C): 23. Result ID: 69959fae3badfaffe69c1970. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### deepseek-VL-7B VQA Close Accuracy (C): 21.67. Result ID: 69959fae3badfaffe69c196d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Med-Flamingo-9B VQA Close Accuracy (C): 16. Result ID: 69959fae3badfaffe69c196c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Mantis† VQA Close Accuracy (C): 7. Result ID: 69959fae3badfaffe69c1974. [Reporting paper]() Recorded evaluation configuration: {"setup":"without composed instruction dataset"}. This record does not assert independent reproduction. ### LLaVA-Med† VQA Close Accuracy (C): 2.66. Result ID: 69959fae3badfaffe69c1972. [Reporting paper]() Recorded evaluation configuration: {"setup":"without composed instruction dataset"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.