DrawingVQA
Models answer 92 multiple-choice questions about 33 construction drawings. They read symbols, cross-reference details, and calculate quantities.
The authors list 19 models. Energent reports 89.1% accuracy by majority vote across three runs. The authors withhold original drawing images.
- Made by
- Louisiana State University, University of Illinois Urbana-Champaign, and an independent researcher
- Released
- April 2026
- Kind
- Documents, charts and text in images
- Level
- University
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- Public, plus a private test set
- Measures
- Capability
- Subject
- Civil engineering
- Language
- English
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- A host
- In views
- Vision
- Strongest model tested
-
Gemini 3 Pro #56 on the capability index
- Last model added
- November 2025
- Scores refreshed
- November 20th, 2026
Gemini 3 Pro leads with 77.2% among models released by November 2025 Gemini 3 Pro leads with 77.2% among models released by November 2025
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in November 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 77.2
Gemini 3 Pro
- 71.7
Gemini 2.5 Pro
- 66.3
Gemini 2.5 Flash
- 58.7 o3
- 58.7
Qwen3-VL-32B-Instruct
- 57.6 Claude Sonnet 4.5
- 54.3 Claude Haiku 4.5
- 53.3
Llama 4 Scout
- 53.3
Qwen3-VL-8B-Instruct
- 50.0
InternVL3.5-38B
- 50.0
InternVL3.5-8B
- 48.9 GPT-4o
- 47.8
Qwen3-VL-30B-A3B-Instruct
- 42.4
LLaVA-v1.6-34B
- 41.3
InternVL3.5-30B-A3B
- 40.2 LLaVA-OneVision-1.5-8B-Instruct
- 40.2
Phi-4-multimodal
- 39.1
Llama 3.2 11B
- 28.3 llava-v1.6-mistral-7b
- 1
Gemini 3 Pro 77.2%
- 2
Gemini 2.5 Pro Jun 2025 71.7%
- 3
Gemini 2.5 Flash Sep 2025 66.3%
- 4 o3 58.7%
- 5
Qwen3-VL-32B-Instruct 58.7%
- 6 Claude Sonnet 4.5 57.6%
- 7 Claude Haiku 4.5 54.3%
- 8
Llama 4 Scout 53.3%
- 9
Qwen3-VL-8B-Instruct 53.3%
- 10
InternVL3.5-38B 50%
- 11
InternVL3.5-8B 50%
- 12 GPT-4o Mar 2025 48.9%
- 13
Qwen3-VL-30B-A3B-Instruct 47.8%
- 14
LLaVA-v1.6-34B 42.4%
- 15
InternVL3.5-30B-A3B 41.3%
- 16 LLaVA-OneVision-1.5-8B-Instruct 40.2%
- 17
Phi-4-multimodal 40.2%
- 18
Llama 3.2 11B 39.1%
- 19 llava-v1.6-mistral-7b 28.3%
Source: Official DrawingVQA leaderboard
- 77.2
Gemini 3 Pro
- 71.7
Gemini 2.5 Pro
- 66.3
Gemini 2.5 Flash
- 58.7 o3
- 58.7
Qwen3-VL-32B-Instruct
- 57.6 Claude Sonnet 4.5
- 54.3 Claude Haiku 4.5
- 53.3
Llama 4 Scout
- 53.3
Qwen3-VL-8B-Instruct
- 50.0
InternVL3.5-38B
- 50.0
InternVL3.5-8B
- 48.9 GPT-4o
- 47.8
Qwen3-VL-30B-A3B-Instruct
- 42.4
LLaVA-v1.6-34B
- 41.3
InternVL3.5-30B-A3B
- 40.2
Phi-4-multimodal
- 39.1
Llama 3.2 11B
- 28.3 llava-v1.6-mistral-7b
- 1
Gemini 3 Pro 77.2%
- 2
Gemini 2.5 Pro Jun 2025 71.7%
- 3
Gemini 2.5 Flash Sep 2025 66.3%
- 4 o3 58.7%
- 5
Qwen3-VL-32B-Instruct 58.7%
- 6 Claude Sonnet 4.5 57.6%
- 7 Claude Haiku 4.5 54.3%
- 8
Llama 4 Scout 53.3%
- 9
Qwen3-VL-8B-Instruct 53.3%
- 10
InternVL3.5-38B 50%
- 11
InternVL3.5-8B 50%
- 12 GPT-4o Mar 2025 48.9%
- 13
Qwen3-VL-30B-A3B-Instruct 47.8%
- 14
LLaVA-v1.6-34B 42.4%
- 15
InternVL3.5-30B-A3B 41.3%
- 16
Phi-4-multimodal 40.2%
- 17
Llama 3.2 11B 39.1%
- 18 llava-v1.6-mistral-7b 28.3%
Source: Official DrawingVQA leaderboard
| # | Entry | Score | Setting | Date | Link | Source |
|---|---|---|---|---|---|---|
| 1 | | 77.2% | Nov 2025 | deepmind.google | Official leaderboard | |
| 2 | | 71.7% | Jun 2025 | deepmind.google | Official leaderboard | |
| 3 | | 66.3% | Sep 2025 | deepmind.google | Official leaderboard | |
| 4 | o3 | 58.7% | Apr 2025 | platform.openai.com | Official leaderboard | |
| 5 | | 58.7% | Oct 2025 | huggingface.co | Official leaderboard | |
| 6 | Claude Sonnet 4.5 | 57.6% | Sep 2025 | anthropic.com | Official leaderboard | |
| 7 | Claude Haiku 4.5 | 54.3% | Oct 2025 | anthropic.com | Official leaderboard | |
| 8 | | 53.3% | Apr 2025 | huggingface.co | Official leaderboard | |
| 9 | | 53.3% | Oct 2025 | huggingface.co | Official leaderboard | |
| 10 | | 50% | Aug 2025 | huggingface.co | Official leaderboard | |
| 11 | | 50% | Aug 2025 | huggingface.co | Official leaderboard | |
| 12 | GPT-4o Mar 2025 | 48.9% | Mar 2025 | platform.openai.com | Official leaderboard | |
| 13 | | 47.8% | Oct 2025 | huggingface.co | Official leaderboard | |
| 14 | | 42.4% | Jan 2024 | huggingface.co | Official leaderboard | |
| 15 | | 41.3% | Aug 2025 | huggingface.co | Official leaderboard | |
| 16 | LLaVA-OneVision-1.5-8B-Instruct open weights | 40.2% | huggingface.co | Official leaderboard | ||
| 17 | | 40.2% | Feb 2025 | huggingface.co | Official leaderboard | |
| 18 | | 39.1% | Sep 2024 | huggingface.co | Official leaderboard | |
| 19 | llava-v1.6-mistral-7b | 28.3% | Jan 2024 | huggingface.co | Official leaderboard |
| # | Model | Score | Setting | Date | Link | Source |
|---|---|---|---|---|---|---|
| 1 | | 77.2% | Nov 2025 | deepmind.google | Official leaderboard | |
| 2 | | 71.7% | Jun 2025 | deepmind.google | Official leaderboard | |
| 3 | | 66.3% | Sep 2025 | deepmind.google | Official leaderboard | |
| 4 | o3 | 58.7% | Apr 2025 | platform.openai.com | Official leaderboard | |
| 5 | | 58.7% | Oct 2025 | huggingface.co | Official leaderboard | |
| 6 | Claude Sonnet 4.5 | 57.6% | Sep 2025 | anthropic.com | Official leaderboard | |
| 7 | Claude Haiku 4.5 | 54.3% | Oct 2025 | anthropic.com | Official leaderboard | |
| 8 | | 53.3% | Apr 2025 | huggingface.co | Official leaderboard | |
| 9 | | 53.3% | Oct 2025 | huggingface.co | Official leaderboard | |
| 10 | | 50% | Aug 2025 | huggingface.co | Official leaderboard | |
| 11 | | 50% | Aug 2025 | huggingface.co | Official leaderboard | |
| 12 | GPT-4o Mar 2025 | 48.9% | Mar 2025 | platform.openai.com | Official leaderboard | |
| 13 | | 47.8% | Oct 2025 | huggingface.co | Official leaderboard | |
| 14 | | 42.4% | Jan 2024 | huggingface.co | Official leaderboard | |
| 15 | | 41.3% | Aug 2025 | huggingface.co | Official leaderboard | |
| 16 | | 40.2% | Feb 2025 | huggingface.co | Official leaderboard | |
| 17 | | 39.1% | Sep 2024 | huggingface.co | Official leaderboard | |
| 18 | llava-v1.6-mistral-7b | 28.3% | Jan 2024 | huggingface.co | Official leaderboard |
Rankings by part
We read 14 parts from the benchmark’s own leaderboard in 3 groups: reasoning depth, mllm dimensions and construction eng. domain aspects. Each ranks models on one part of the test. Gemini 3 Pro leads 11 of them, and 3 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
R1
18 models we track have a score here.
- 1
Gemini 2.5 Pro Jun 2025 ↑1 80.6%
- 2
Gemini 3 Pro ↓1 80.6%
- 3
Gemini 2.5 Flash Sep 2025 77.4%
- 4 o3 64.5%
- 5 Claude Sonnet 4.5 ↑1 64.5%
- 6 GPT-4o Mar 2025 ↑6 61.3%
- 7
Llama 4 Scout ↑1 58.1%
- 8
InternVL3.5-8B ↑3 58.1%
- 9
InternVL3.5-30B-A3B ↑6 51.6%
- 10
Qwen3-VL-30B-A3B-Instruct ↑3 51.6%
R2
18 models we track have a score here.
- 1
Qwen3-VL-32B-Instruct ↑4 79.4%
- 2
Gemini 3 Pro ↓1 79.4%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 76.5%
- 4
Gemini 2.5 Flash Sep 2025 ↓1 73.5%
- 5 Claude Haiku 4.5 ↑2 73.5%
- 6
InternVL3.5-38B ↑4 64.7%
- 7 Claude Sonnet 4.5 ↓1 64.7%
- 8 o3 ↓4 61.8%
- 9
Qwen3-VL-8B-Instruct 61.8%
- 10
Llama 4 Scout ↓2 58.8%
R3
18 models we track have a score here.
- 1
Gemini 3 Pro 70.4%
- 2
Gemini 2.5 Pro Jun 2025 55.6%
- 3 o3 ↑1 48.2%
- 4
LLaVA-v1.6-34B ↑10 44.4%
- 5
Gemini 2.5 Flash Sep 2025 ↓2 44.4%
- 6
Qwen3-VL-8B-Instruct ↑3 44.4%
- 7
Qwen3-VL-32B-Instruct ↓2 44.4%
- 8 GPT-4o Mar 2025 ↑4 40.7%
- 9
Llama 4 Scout ↓1 40.7%
- 10 Claude Sonnet 4.5 ↓4 40.7%
VP
18 models we track have a score here.
- 1
Gemini 3 Pro 76.5%
- 2
Gemini 2.5 Pro Jun 2025 67.6%
- 3
Gemini 2.5 Flash Sep 2025 64.7%
- 4 o3 58.8%
- 5 Claude Sonnet 4.5 ↑1 58.8%
- 6
Qwen3-VL-32B-Instruct ↓1 55.9%
- 7
Llama 4 Scout ↑1 52.9%
- 8 Claude Haiku 4.5 ↓1 52.9%
- 9
LLaVA-v1.6-34B ↑5 50%
- 10
Qwen3-VL-8B-Instruct ↓1 50%
K
18 models we track have a score here.
- 1
Gemini 3 Pro 81.3%
- 2
Gemini 2.5 Flash Sep 2025 ↑1 81.2%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 78.1%
- 4 o3 65.6%
- 5 Claude Haiku 4.5 ↑2 65.6%
- 6
Llama 4 Scout ↑2 62.5%
- 7 GPT-4o Mar 2025 ↑5 59.4%
- 8
InternVL3.5-38B ↑2 59.4%
- 9 Claude Sonnet 4.5 ↓3 59.4%
- 10
Qwen3-VL-32B-Instruct ↓5 56.3%
R
18 models we track have a score here.
- 1
Gemini 3 Pro 72.1%
- 2
Gemini 2.5 Pro Jun 2025 67.2%
- 3 Claude Sonnet 4.5 ↑3 57.9%
- 4
Gemini 2.5 Flash Sep 2025 ↓1 57.4%
- 5
Qwen3-VL-32B-Instruct 55.7%
- 6 o3 ↓2 54.1%
- 7
Llama 4 Scout ↑1 52.5%
- 8 Claude Haiku 4.5 ↓1 52.5%
- 9
InternVL3.5-8B ↑2 49.2%
- 10
InternVL3.5-38B 47.5%
OCR
18 models we track have a score here.
- 1
Gemini 3 Pro 80.3%
- 2
Gemini 2.5 Pro Jun 2025 72.4%
- 3
Gemini 2.5 Flash Sep 2025 63.2%
- 4
Qwen3-VL-32B-Instruct ↑1 63.2%
- 5 o3 ↓1 59.2%
- 6
Llama 4 Scout ↑2 55.3%
- 7
Qwen3-VL-8B-Instruct ↑2 55.3%
- 8 Claude Sonnet 4.5 ↓2 54.1%
- 9
InternVL3.5-38B ↑1 54%
- 10 Claude Haiku 4.5 ↓3 54%
Adm.
18 models we track have a score here.
- 1
Gemini 3 Pro 93.8%
- 2
Gemini 2.5 Pro Jun 2025 87.5%
- 3
Gemini 2.5 Flash Sep 2025 87.5%
- 4
Qwen3-VL-8B-Instruct ↑5 81.2%
- 5
Qwen3-VL-32B-Instruct 75%
- 6 Claude Sonnet 4.5 68.8%
- 7
Llama 4 Scout ↑1 62.5%
- 8 o3 ↓4 62.5%
- 9
InternVL3.5-38B ↑1 62.5%
- 10
Qwen3-VL-30B-A3B-Instruct ↑3 62.5%
EI
18 models we track have a score here.
- 1
Gemini 3 Pro 75.7%
- 2 o3 ↑2 56.8%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 56.8%
- 4
Gemini 2.5 Flash Sep 2025 ↓1 51.4%
- 5 Claude Sonnet 4.5 ↑1 48.7%
- 6
Qwen3-VL-32B-Instruct ↓1 48.7%
- 7 Claude Haiku 4.5 43.2%
- 8
Phi-4-multimodal ↑8 40.5%
- 9
InternVL3.5-30B-A3B ↑6 40.5%
- 10
Qwen3-VL-8B-Instruct ↓1 40.5%
LU
18 models we track have a score here.
- 1
Gemini 3 Pro 83.3%
- 2
Gemini 2.5 Pro Jun 2025 79.2%
- 3
Gemini 2.5 Flash Sep 2025 79.2%
- 4 GPT-4o Mar 2025 ↑8 70.8%
- 5 Claude Sonnet 4.5 ↑1 66.7%
- 6
Qwen3-VL-8B-Instruct ↑3 66.7%
- 7
Qwen3-VL-32B-Instruct ↓2 66.7%
- 8
InternVL3.5-38B ↑2 62.5%
- 9
Qwen3-VL-30B-A3B-Instruct ↑4 62.5%
- 10
Llama 4 Scout ↓2 58.3%
DR
18 models we track have a score here.
- 1
Gemini 3 Pro 77.4%
- 2
Gemini 2.5 Pro Jun 2025 67.7%
- 3
Qwen3-VL-32B-Instruct ↑2 67.7%
- 4
InternVL3.5-38B ↑6 54.8%
- 5
Gemini 2.5 Flash Sep 2025 ↓2 51.6%
- 6 o3 ↓2 48.4%
- 7 Claude Haiku 4.5 48.4%
- 8
Qwen3-VL-30B-A3B-Instruct ↑5 45.2%
- 9
Qwen3-VL-8B-Instruct 45.2%
- 10
InternVL3.5-30B-A3B ↑5 45.1%
SU
18 models we track have a score here.
- 1
Gemini 3 Pro 78.8%
- 2
Gemini 2.5 Pro Jun 2025 75%
- 3
Gemini 2.5 Flash Sep 2025 67.3%
- 4 o3 61.5%
- 5 Claude Sonnet 4.5 ↑1 57.7%
- 6
Qwen3-VL-32B-Instruct ↓1 57.7%
- 7 Claude Haiku 4.5 51.9%
- 8 GPT-4o Mar 2025 ↑4 50%
- 9
Llama 4 Scout ↓1 48.1%
- 10
Qwen3-VL-30B-A3B-Instruct ↑3 48.1%
QTO
18 models we track have a score here.
- 1
InternVL3.5-8B ↑10 58.3%
- 2
Llama 4 Scout ↑6 50%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 41.7%
- 4 Claude Sonnet 4.5 ↑2 41.7%
- 5
LLaVA-v1.6-34B ↑9 33.3%
- 6
Gemini 2.5 Flash Sep 2025 ↓3 33.3%
- 7 Claude Haiku 4.5 33.3%
- 8
Qwen3-VL-8B-Instruct ↑1 33.3%
- 9
Qwen3-VL-32B-Instruct ↓4 33.3%
- 10
Gemini 3 Pro ↓9 33.3%
Comp.
18 models we track have a score here.
- 1
Gemini 3 Pro 85.7%
- 2
Gemini 2.5 Flash Sep 2025 ↑1 71.4%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 57.1%
- 4 Claude Sonnet 4.5 ↑2 57.1%
- 5
LLaVA-v1.6-34B ↑9 42.9%
- 6 GPT-4o Mar 2025 ↑6 42.9%
- 7
Llama 4 Scout ↑1 42.9%
- 8 o3 ↓4 42.9%
- 9 Claude Haiku 4.5 ↓2 42.9%
- 10
Qwen3-VL-8B-Instruct ↓1 42.9%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official DrawingVQA leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official DrawingVQA leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Vision
- MMMU In use 11.5K college-level multimodal questions across 30 subjects. Multiple-choice and open accuracy.
Qwen 3.6 Plus 86%
- MMMU-Pro In use Harder MMMU: college-level multimodal questions with more options and vision-only inputs to remove text shortcuts. GPT-5.6 Sol 84.6%
- CharXiv In use Chart understanding on real arXiv figures. Reasoning questions require synthesizing information from complex charts. Claude Fable 5 93.5%
- MathVista In use Math reasoning in visual contexts (charts, geometry, puzzles) from 31 datasets. Answer accuracy.
Seed2.1 Pro 90.7%
- MATH-Vision (MathVision) In use 3,040 competition math problems with visual context. Answer accuracy.
Qwen 3.8 Max 97.7%
- MMStar In use Models answer 1,500 human-curated image questions that cannot be solved without the image, across 6 core capabilities. Scored by accuracy.
Qwen 3.8 Max 85.9%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 12 months ago, in November 2025, for Gemini 3 Pro. Every benchmark
Catalog record 2 pass
- Pass Task access Public tasks, plus a private test set.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the official leaderboard (19 entries).
- Pass Models scored 18 models we track, plus 1 other entry on its leaderboard.
- Warning Newest score The newest came 12 months ago, in November 2025, for Gemini 3 Pro.
- Note Frontier models None of the top 10 has a score. The strongest tested is Gemini 3 Pro (#56).
- Pass Headroom The best is 77.2%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on joonv2.github.io: 18 rows for models we track and 1 other entry at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.