# Multi-image Medical Visual Question Answering on Med-MIM Held-in Source: [Multi-image Medical Visual Question Answering on Med-MIM Held-in]() Collection index: [llms.txt]() Data updated: 2026-02-26T06:06:09.494Z Evaluation of model performance on the Med-MIM Held-in dataset for Multi-image Medical Visual Question Answering, reporting specialized scores across temporal, reasoning, comparison, and co-reference dimensions. 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) | Med-MIM Held-in | | Metric | Temporal (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 - [Temporal (C)](): Higher is better; unit: PERCENT. - [Temporal (O)](): Higher is better; unit: PERCENT. - [Reasoning (C)](): Higher is better; unit: PERCENT. - [Reasoning (O)](): Higher is better; unit: PERCENT. - [Comparison (C)](): Higher is better; unit: PERCENT. - [Comparison (O)](): Higher is better; unit: PERCENT. - [Co-reference (C)](): Higher is better; unit: PERCENT. - [Co-reference (O)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Temporal (C) | Result date | Report paper | | --- | --- | --- | --- | | Med-Mantis | 74.86 | Not recorded | 68fa78c01e209131cea7a40c | | MIM-LLAVA-Med | 74.31 | Not recorded | 68fa78c01e209131cea7a40c | | Mantis-8B | 68.89 | Not recorded | 68fa78c01e209131cea7a40c | | GPT-4o | 47.95 | Not recorded | 68fa78c01e209131cea7a40c | | InternVL2-8B | 46.07 | Not recorded | 68fa78c01e209131cea7a40c | | Med-Flamingo-9B | 39.65 | Not recorded | 68fa78c01e209131cea7a40c | | LLaVA-Med-7B | 38.87 | Not recorded | 68fa78c01e209131cea7a40c | | LLaVA-Med† | 35.44 | Not recorded | 68fa78c01e209131cea7a40c | | deepseek-VL-7B | 29.9 | Not recorded | 68fa78c01e209131cea7a40c | | Mantis† | 5.97 | Not recorded | 68fa78c01e209131cea7a40c | ### Med-Mantis Temporal (C): 74.86. Result ID: 69959fae3badfaffe69c195f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_dataset":"Med-MIM instruction dataset"}. This record does not assert independent reproduction. ### MIM-LLAVA-Med Temporal (C): 74.31. Result ID: 69959fae3badfaffe69c195d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"training\_dataset":"Med-MIM instruction dataset"}. This record does not assert independent reproduction. ### Mantis-8B Temporal (C): 68.89. Result ID: 69959fae3badfaffe69c195c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GPT-4o Temporal (C): 47.95. Result ID: 69959fae3badfaffe69c1957. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### InternVL2-8B Temporal (C): 46.07. Result ID: 69959fae3badfaffe69c195a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Med-Flamingo-9B Temporal (C): 39.65. Result ID: 69959fae3badfaffe69c1958. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### LLaVA-Med-7B Temporal (C): 38.87. Result ID: 69959fae3badfaffe69c195b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### LLaVA-Med† Temporal (C): 35.44. Result ID: 69959fae3badfaffe69c195e. [Reporting paper]() Recorded evaluation configuration: {"setup":"without composed instruction dataset"}. This record does not assert independent reproduction. ### deepseek-VL-7B Temporal (C): 29.9. Result ID: 69959fae3badfaffe69c1959. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Mantis† Temporal (C): 5.97. Result ID: 69959fae3badfaffe69c1960. [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.