Medical Visual Question Answering on MedXpertQA
56AccuracyGEMINI-3-FLASH
Evaluation results
| Method | Links | |
|---|---|---|
| GEMINI-3-FLASHModel Type=Proprietary2026.01 | 56 | |
| GPT-5Model Type=Proprietary2026.01 | 54.8 | |
| Gemini 2.5 ProCategory=Close-Source SOTA2026.04 | 46.6 | |
| OpenAI-o3Category=Close-Source SOTA2026.04 | 44.1 | |
| GPT-4.1Category=Close-Source SOTA2026.04 | 40.8 | |
| GPT-5Category=Close-Source SOTA2026.04 | 40.4 | |
| MedGemma 27BMedical Samples=30M2026.06 | 33.7 | |
| OctoMed 7BMedical Samples=8M2026.06 | 33.32 | |
| QWEN3-VL-8B-INSTRUCT + MED-SCOUTParameters=8B, Enhancement=Med-Scout2026.01 | 30.8 | |
| QWEN3-VL-8B-INSTRUCTParameters=8B2026.01 | 30.4 | |
| LINGSHU-7B + MED-SCOUTParameters=7B, Enhancement=Med-Scout2026.01 | 28 | |
| MedAgent-ProCategory=Multimodal Medical Agents2026.04 | 27.8 | |
| MedGemma 1.5 4BMedical Samples=30M2026.06 | 27.79 | |
| QWEN3-VL-4B-INSTRUCT + MED-SCOUTParameters=4B, Enhancement=Med-Scout2026.01 | 27.7 | |
| LINGSHU-7BParameters=7B2026.01 | 27.4 | |
| QWEN3-VL-4B-INSTRUCTParameters=4B2026.01 | 27 | |
| Qwen2.5-VL-32BCategory=Open-Source SOTA2026.04 | 26.8 | |
| MedTutor-R1Backbone=Qwen2.5VL-7B-Instruct2025.12 | 25.1 | |
| Qwen2.5-VL-7B + OPENMEDREASONMedical Samples=450K2026.06 | 24.95 | |
| QWEN2.5-VL-3B-INSTRUCTParameters=3B2026.01 | 24.3 | |
| Mini-o3-7B-v1Category=MLLMs can Think with Images2026.04 | 24.3 | |
| MedLVRCategory=MLLMs can Think with Images2026.04 | 24.3 | |
| InternVL3-8BCategory=Open-Source SOTA2026.04 | 23.8 | |
| MedVL-Thinker 7BMedical Samples=200k2026.06 | 23.8 | |
| HuatuoGPT-Vision-34BCategory=Medical MLLMs2026.04 | 23.6 | |
| DeepEyes-7BCategory=MLLMs can Think with Images2026.04 | 23.6 | |
| Lingshu 7BMedical Samples=5M2026.06 | 23.51 | |
| AURACategory=Multimodal Medical Agents2026.04 | 23.5 | |
| PixelReasoner-RL-v1-7BCategory=MLLMs can Think with Images2026.04 | 23.5 | |
| VILA-M3-40BCategory=Multimodal Medical Agents2026.04 | 23 | |
| Med-R1Category=MLLMs can Think about Images2026.04 | 22.9 | |
| HUATUOGPT-VISION-7B + MED-SCOUTParameters=7B, Enhancement=Med-Scout2026.01 | 22.7 | |
| MedTutor-R1 w/ LLaVA-basedBackbone=LLaVA2025.12 | 22.67 | |
| Qwen2.5-VL-7BMedical Samples=N/A2026.06 | 22.62 | |
| MMedAgent-RL-7BCategory=Multimodal Medical Agents2026.04 | 22.6 | |
| Qwen2.5-VL-7BCategory=MLLMs can Think with Images2026.04 | 22.5 | |
| INTERNVL3-8BParameters=8B2026.01 | 22.4 | |
| HUATUOGPT-VISION-7BParameters=7B2026.01 | 22.4 | |
| MMedAgent-7BCategory=Multimodal Medical Agents2026.04 | 22.3 | |
| MEDGEMMA-4B-ITParameters=4B2026.01 | 22 | |
| QWEN2.5-VL-7B-INSTRUCTParameters=7B2026.01 | 21.9 | |
| MedVLM-R1Category=MLLMs can Think about Images2026.04 | 21.7 | |
| QoQ-Med-VL 7BMedical Samples=2.6M2026.06 | 21.4 | |
| MedTutor-R1 w/o RLReinforcement Learning=disabled2025.12 | 20.8 | |
| LLaVA-Next-7BCategory=Open-Source SOTA2026.04 | 20.7 | |
| LLAVA-MED-7BParameters=7B2026.01 | 19.9 | |
| LLaVA-Med-7BCategory=Medical MLLMs2026.04 | 19.9 | |
| RadFMCategory=Medical MLLMs2026.04 | 19.8 | |
| LLaVA-Next-13BCategory=Open-Source SOTA2026.04 | 19.6 | |
| SMR-AgentsCategory=Multimodal Medical Agents2026.04 | 19.6 | |
| Med-FlamingoCategory=Medical MLLMs2026.04 | 19.3 | |
| Qwen2.5VLBackbone=Qwen2.5VL-7B-Instruct2025.12 | 18.39 |