SADU - Structural Diagrams: leaderboard

SADU (2026) tests vision-language models on counting and retrieval questions about behavioral, structural and entity-relationship software architecture diagrams. This board is exact-match accuracy in percent on questions about structural diagrams; higher is better.

Metric: Exact-match accuracy (%) on the 848 structural-diagram questions of SADU (154 software architecture diagrams (behavioral, structural and ER) with 2,431 counting and retrieval questions, temperature 0); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash (Preview)68.87#152
2Gemini 2.5 Flash65.21#463
3Gemini 3.1 Flash Lite (Preview)60.85#417
4GPT-5 Nano50.59#906
5Claude Sonnet 4.546.7#261
6Claude Haiku 4.538.56#583
7Gemini 2.5 Flash Lite36.44#964
8Qwen 2.5 VL 32B36.44#941
9Qwen 2.5 VL 7B31.01#1862
10Qwen 2.5 VL 3B26.77#2639
11GPT-4o Mini10.14#1667

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=sadu-structural-diagrams · How It Works · Data refreshed daily, snapshot 2026-11-21.