MMGist - Diagram and OCR: leaderboard
Metric: Accuracy (%) on the Diagram and OCR dimension (items kept from AI2D and OCRBench); 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 January 2028. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) (Medium) | 82.3 |
| 2 | Gemini 3.1 Flash Lite (Medium) | 73.4 |
| 3 | Qwen 3.6 Plus | 68.7 |
| 4 | Seed 2.0 Pro | 65.8 |
| 5 | Qwen 3.5 9B | 65.5 |
| 6 | Qwen 3.6 35B A3B | 65.4 |
| 7 | Gemma 4 31B | 63.8 |
| 8 | Seed 2.0 Lite | 63.2 |
| 9 | Seed 2.0 Mini | 63.2 |
| 10 | Step3 VL 10B | 63.1 |
| 11 | Gemma 4 26B A4B | 61.3 |
| 12 | Claude Sonnet 4.6 (Medium) | 61.3 |
| 13 | GPT-5 (Medium) | 58.6 |
| 14 | Qwen 3.5 27B | 55.4 |
| 15 | GPT-5 Mini (Medium) | 54.4 |
Interactive version: theaggregate.ai/benchmark?slug=mmgist-diagram-and-ocr · How It Works · Data refreshed daily, snapshot 2026-09-29.