MMGist - STEM Reasoning: leaderboard
MMGist (2026) is a curated multimodal benchmark: 7,262 items kept from 18 vision-language benchmarks after dropping items that can be answered without the image, items nearly all models solve, and items with faulty labels. This board is accuracy in percent on the STEM Reasoning items, drawn from MathVista, MathVision, DynaMath and MME-SCI; higher is better.
Metric: Accuracy (%) on the STEM Reasoning dimension (items kept from MathVista, MathVision, DynaMath and MME-SCI); 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 February 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.6 Plus | 62.6 | #222 |
| 2 | Gemini 3.1 Pro (Preview) (Medium) | 61.2 | #94 (Gemini 3.1 Pro (Preview)) |
| 3 | Seed 2.0 Pro | 57.6 | #211 |
| 4 | Qwen 3.6 35B A3B | 54.3 | #462 |
| 5 | Seed 2.0 Lite | 53.1 | #289 |
| 6 | Seed 2.0 Mini | 48.3 | #508 |
| 7 | Qwen 3.5 27B | 46.6 | #359 |
| 8 | GPT-5 (Medium) | 46.1 | #167 (GPT-5) |
| 9 | Claude Sonnet 4.6 (Medium) | 44.6 | #161 (Claude Sonnet 4.6) |
| 10 | Gemini 3.1 Flash Lite (Medium) | 41.5 | #364 (Gemini 3.1 Flash Lite) |
| 11 | Gemma 4 31B | 40.4 | #394 |
| 12 | Qwen 3.5 9B | 37.1 | #803 |
| 13 | Gemma 4 26B A4B | 36.8 | #615 |
| 14 | GPT-5 Mini (Medium) | 36.8 | #348 (GPT-5 Mini) |
| 15 | Qwen 3.5 4B | 31.1 | #1175 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=mmgist-stem-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-13.