MMGist - Medical: leaderboard

Metric: Accuracy (%) on the Medical dimension (items kept from SLAKE and MedXpertQA); MMGist: 7,262 curated items from 18 vision-language benchmarks after removing items answerable without the image, items nearly every model solves and items with faulty labels; eight samples per item at temperature 1.0, step-by-step answer in a boxed block, 16,384 output tokens, medium effort where configurable; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 27 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview) (Medium)58
2Qwen 3.6 Plus50
3GPT-5 (Medium)46.6
4Qwen 3.6 35B A3B43
5Seed 2.0 Pro43
6Gemini 3.1 Flash Lite (Medium)41.8
7Seed 2.0 Lite39.6
8Gemma 4 31B39
9Claude Sonnet 4.6 (Medium)37.5
10GPT-5 Mini (Medium)37.2
11Gemma 4 26B A4B35.5
12Seed 2.0 Mini35.2
13Qwen 3.5 9B35
14Qwen 3.5 27B33.2
15Qwen 3.5 4B32.1

Interactive version: theaggregate.ai/benchmark?slug=mmgist-medical · How It Works · Data refreshed daily, snapshot 2026-09-29.