DrawingVQA - Domain-Expert Reasoning: leaderboard

Metric: Accuracy (%; reasoning depth R3, 27 domain-expert reasoning questions of the 92 expert-curated questions on 33 Issued for Construction drawings, multiple choice and open-ended). Source: arxiv.org. Saturation forecast: Around September 2027. 19 models tracked.

Top models

#ModelScore
1Gemini 3 Pro (Preview)70.4
2Gemini 2.5 Pro55.6
3O3 (2025-04-16)48.2
4Gemini 2.5 Flash44.4
5Qwen 3 VL 8B Instruct44.4
6Qwen 3 VL 32B Instruct44.4
7Claude Sonnet 4.540.7
8Llama 4 Scout Instruct40.7
9GPT-4o (2024-08-06)40.7
10Claude Haiku 4.533.3
11InternVL3.5-8B29.6
12Phi-4 Multimodal Instruct29.6
13Qwen 3 VL 30B A3B Instruct29.6
14Llama 3.2 11B Instruct22.2

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