MMGist - Visual Logic: leaderboard
Metric: Accuracy (%) on the Visual Logic dimension (items kept from LogicVista and ZeroBench); 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 2030. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) (Medium) | 40.4 |
| 2 | Seed 2.0 Pro | 37.7 |
| 3 | Seed 2.0 Lite | 35.1 |
| 4 | Qwen 3.6 Plus | 34.9 |
| 5 | Qwen 3.6 35B A3B | 31.5 |
| 6 | Seed 2.0 Mini | 30.3 |
| 7 | Qwen 3.5 27B | 28.1 |
| 8 | Gemma 4 31B | 27.7 |
| 9 | Gemini 3.1 Flash Lite (Medium) | 27.2 |
| 10 | Qwen 3.5 9B | 26.4 |
| 11 | GPT-5 (Medium) | 25.8 |
| 12 | GPT-5 Mini (Medium) | 24.4 |
| 13 | Gemma 4 26B A4B | 24.3 |
| 14 | Claude Sonnet 4.6 (Medium) | 22.4 |
| 15 | Step3 VL 10B | 21.6 |
Interactive version: theaggregate.ai/benchmark?slug=mmgist-visual-logic · How It Works · Data refreshed daily, snapshot 2026-09-29.