DrawingVQA - Perceptual Understanding: leaderboard
Simple, direct questions about a construction drawing that need visual search or basic OCR, such as the sheet title; the first reasoning depth of DrawingVQA (2026). Accuracy in percent, higher is better.
Metric: Accuracy (%; reasoning depth R1, 31 perceptual-understanding questions of the 92 expert-curated questions on 33 Issued for Construction drawings, multiple choice and open-ended). Source: arxiv.org. Saturation forecast: Around February 2027. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 80.6 | #266 |
| 2 | Gemini 3 Pro (Preview) | 80.6 | #113 |
| 3 | Gemini 2.5 Flash | 77.4 | #453 |
| 4 | Claude Sonnet 4.5 | 64.5 | #250 |
| 5 | O3 (2025-04-16) | 64.5 | #216 |
| 6 | GPT-4o (2024-08-06) | 61.3 | #740 |
| 7 | Llama 4 Scout Instruct | 58.1 | #1543 |
| 8 | Claude Haiku 4.5 | 51.6 | #568 |
| 9 | Qwen 3 VL 8B Instruct | 51.6 | #981 |
| 10 | Qwen 3 VL 30B A3B Instruct | 51.6 | #871 |
| 11 | Qwen 3 VL 32B Instruct | 48.4 | #585 |
| 12 | Llama 3.2 11B Instruct | 45.2 | #4991 |
| 13 | Phi-4 Multimodal Instruct | 41.9 | #3882 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=drawingvqa-perceptual-understanding · How It Works · Data refreshed daily, snapshot 2026-11-13.