# Medical Multi-task Visual Reasoning on MedAD-38K (test)

Source: [Medical Multi-task Visual Reasoning on MedAD-38K (test)](<https://www.sota2.com/research/sota/medical-multi-task-visual-reasoning-on-medad-38k-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T05:34:54.338Z

Evaluation of model performance on the test split of the MedAD-38K dataset, focusing on multi-task visual reasoning across anatomy identification, anomaly detection, lesion localization, modality classification, and pathology characterization.

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 Multi-task Visual Reasoning |
| Dataset (including any stated split) | MedAD-38K (test) |
| Metric | Anatomy ID Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 17 of 17 |
| 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

- [Anatomy ID Accuracy](<https://www.sota2.com/research/sota/medical-multi-task-visual-reasoning-on-medad-38k-test.md?metric=Anatomy%20ID%20Accuracy>): Higher is better; unit: PERCENT.

- [Anomaly Detection Accuracy](<https://www.sota2.com/research/sota/medical-multi-task-visual-reasoning-on-medad-38k-test.md?metric=Anomaly%20Detection%20Accuracy>): Higher is better; unit: PERCENT.

- [Lesion Localization Accuracy](<https://www.sota2.com/research/sota/medical-multi-task-visual-reasoning-on-medad-38k-test.md?metric=Lesion%20Localization%20Accuracy>): Higher is better; unit: PERCENT.

- [Modality Classification Accuracy](<https://www.sota2.com/research/sota/medical-multi-task-visual-reasoning-on-medad-38k-test.md?metric=Modality%20Classification%20Accuracy>): Higher is better; unit: PERCENT.

- [Pathology Characterization Accuracy](<https://www.sota2.com/research/sota/medical-multi-task-visual-reasoning-on-medad-38k-test.md?metric=Pathology%20Characterization%20Accuracy>): Higher is better; unit: PERCENT.

- [Overall Accuracy](<https://www.sota2.com/research/sota/medical-multi-task-visual-reasoning-on-medad-38k-test.md?metric=Overall%20Accuracy>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Anatomy ID Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| MedAD-R1 | 98.87 | Not recorded | 69818134a8b01e76cc8e2369 |
| Lingshu | 96.18 | Not recorded | 69818134a8b01e76cc8e2369 |
| HuatuoGPT-Vision | 95.65 | Not recorded | 69818134a8b01e76cc8e2369 |
| GLM-4.1V-Thinking | 94.87 | Not recorded | 69818134a8b01e76cc8e2369 |
| Qwen3VL | 94.68 | Not recorded | 69818134a8b01e76cc8e2369 |
| InternVL3.5 | 94.02 | Not recorded | 69818134a8b01e76cc8e2369 |
| Qwen2.5VL-72B | 93.75 | Not recorded | 69818134a8b01e76cc8e2369 |
| Grok4-Fast | 93.47 | Not recorded | 69818134a8b01e76cc8e2369 |
| InternVL3.5 | 93.39 | Not recorded | 69818134a8b01e76cc8e2369 |
| Qwen2.5-VL-3B | 93.2 | Not recorded | 69818134a8b01e76cc8e2369 |
| Llama-3.2-Vision | 92.38 | Not recorded | 69818134a8b01e76cc8e2369 |
| MedVLM | 91.17 | Not recorded | 69818134a8b01e76cc8e2369 |
| MiniCPM-V-4.5 | 91.16 | Not recorded | 69818134a8b01e76cc8e2369 |
| ERNIE-4.5-VL | 90.49 | Not recorded | 69818134a8b01e76cc8e2369 |
| MiMo-VL-RL | 88.43 | Not recorded | 69818134a8b01e76cc8e2369 |
| LLaVA-Med | 86.45 | Not recorded | 69818134a8b01e76cc8e2369 |
| Qwen2.5-VL-7B | 83.98 | Not recorded | 69818134a8b01e76cc8e2369 |

### MedAD-R1

Anatomy ID Accuracy: 98.87. Result ID: 699762c65698665ec4b5c0c1.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

Recorded evaluation configuration: {"Params":"3B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### Lingshu

Anatomy ID Accuracy: 96.18. Result ID: 699762c65698665ec4b5c0be.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa795a1e209131cea7d387.md>)

Recorded evaluation configuration: {"Params":"7B","Domain-specific medical LMM":true}. This record does not assert independent reproduction.

### HuatuoGPT-Vision

Anatomy ID Accuracy: 95.65. Result ID: 699762c65698665ec4b5c0bd.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa74f71e209131cea64ea0.md>)

Recorded evaluation configuration: {"Params":"7B","Domain-specific medical LMM":true}. This record does not assert independent reproduction.

### GLM-4.1V-Thinking

Anatomy ID Accuracy: 94.87. Result ID: 699762c65698665ec4b5c0b6.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7aaf1e209131cea849ba.md>)

Recorded evaluation configuration: {"Params":"9B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### Qwen3VL

Anatomy ID Accuracy: 94.68. Result ID: 699762c65698665ec4b5c0b9.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

Recorded evaluation configuration: {"Params":"32B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### InternVL3.5

Anatomy ID Accuracy: 94.02. Result ID: 699762c65698665ec4b5c0b4.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78241e209131cea76fa5.md>)

Recorded evaluation configuration: {"Params":"8B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### Qwen2.5VL-72B

Anatomy ID Accuracy: 93.75. Result ID: 699762c65698665ec4b5c0bb.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Params":"72B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### Grok4-Fast

Anatomy ID Accuracy: 93.47. Result ID: 699762c65698665ec4b5c0bc.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

Recorded evaluation configuration: {"Params":"null","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### InternVL3.5

Anatomy ID Accuracy: 93.39. Result ID: 699762c65698665ec4b5c0ba.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78241e209131cea76fa5.md>)

Recorded evaluation configuration: {"Params":"38B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### Qwen2.5-VL-3B

Anatomy ID Accuracy: 93.2. Result ID: 699762c65698665ec4b5c0b1.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Params":"3B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### Llama-3.2-Vision

Anatomy ID Accuracy: 92.38. Result ID: 699762c65698665ec4b5c0b7.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"Params":"11B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### MedVLM

Anatomy ID Accuracy: 91.17. Result ID: 699762c65698665ec4b5c0c0.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77831e209131cea73bcd.md>)

Recorded evaluation configuration: {"Params":"2B","Domain-specific medical LMM":true}. This record does not assert independent reproduction.

### MiniCPM-V-4.5

Anatomy ID Accuracy: 91.16. Result ID: 699762c65698665ec4b5c0b5.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7b8d1e209131cea8913f.md>)

Recorded evaluation configuration: {"Params":"9B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### ERNIE-4.5-VL

Anatomy ID Accuracy: 90.49. Result ID: 699762c65698665ec4b5c0b8.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

Recorded evaluation configuration: {"Params":"28B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### MiMo-VL-RL

Anatomy ID Accuracy: 88.43. Result ID: 699762c65698665ec4b5c0b3.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79321e209131cea7c916.md>)

Recorded evaluation configuration: {"Params":"7B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

### LLaVA-Med

Anatomy ID Accuracy: 86.45. Result ID: 699762c65698665ec4b5c0bf.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6f8e1e209131cea46356.md>)

[Implementation](<https://aka.ms/llava-med>)

Recorded evaluation configuration: {"Params":"7B","Domain-specific medical LMM":true}. This record does not assert independent reproduction.

### Qwen2.5-VL-7B

Anatomy ID Accuracy: 83.98. Result ID: 699762c65698665ec4b5c0b2.

[Reporting paper](<https://www.sota2.com/research/paper/69818134a8b01e76cc8e2369.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Params":"7B","Domain-specific medical LMM":false}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.
