Benchmark Almanac

DrawingVQA

Gemini 3 Pro 77.2% leads since April 2026
Not in model cards

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.

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
Language
English
In and out
Text, Images → Text
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
19
Entries on its own leaderboard, 18 models we track

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:

  1. 1 Gemini 3 Pro 77.2%
  2. 2 Gemini 2.5 Pro Jun 2025 71.7%
  3. 3 Gemini 2.5 Flash Sep 2025 66.3%
  4. 4 o3 58.7%
  5. 5 Qwen3-VL-32B-Instruct 58.7%
  6. 6 Claude Sonnet 4.5 57.6%
  7. 7 Claude Haiku 4.5 54.3%
  8. 8 Llama 4 Scout 53.3%
  9. 9 Qwen3-VL-8B-Instruct 53.3%
  10. 10 InternVL3.5-38B 50%
  11. 11 InternVL3.5-8B 50%
  12. 12 GPT-4o Mar 2025 48.9%
  13. 13 Qwen3-VL-30B-A3B-Instruct 47.8%
  14. 14 LLaVA-v1.6-34B 42.4%
  15. 15 InternVL3.5-30B-A3B 41.3%
  16. 16 LLaVA-OneVision-1.5-8B-Instruct 40.2%
  17. 17 Phi-4-multimodal 40.2%
  18. 18 Llama 3.2 11B 39.1%
  19. 19 llava-v1.6-mistral-7b 28.3%

Source: Official DrawingVQA leaderboard

#EntryScoreSettingDateLinkSource
1 Gemini 3 Pro 77.2% Nov 2025 deepmind.google Official leaderboard
2 Gemini 2.5 Pro Jun 2025 71.7% Jun 2025 deepmind.google Official leaderboard
3 Gemini 2.5 Flash Sep 2025 66.3% Sep 2025 deepmind.google Official leaderboard
4 o3 58.7% Apr 2025 platform.openai.com Official leaderboard
5 Qwen3-VL-32B-Instruct 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 Llama 4 Scout 53.3% Apr 2025 huggingface.co Official leaderboard
9 Qwen3-VL-8B-Instruct 53.3% Oct 2025 huggingface.co Official leaderboard
10 InternVL3.5-38B 50% Aug 2025 huggingface.co Official leaderboard
11 InternVL3.5-8B 50% Aug 2025 huggingface.co Official leaderboard
12 GPT-4o Mar 2025 48.9% Mar 2025 platform.openai.com Official leaderboard
13 Qwen3-VL-30B-A3B-Instruct 47.8% Oct 2025 huggingface.co Official leaderboard
14 LLaVA-v1.6-34B 42.4% Jan 2024 huggingface.co Official leaderboard
15 InternVL3.5-30B-A3B 41.3% Aug 2025 huggingface.co Official leaderboard
16 LLaVA-OneVision-1.5-8B-Instruct open weights 40.2% huggingface.co Official leaderboard
17 Phi-4-multimodal 40.2% Feb 2025 huggingface.co Official leaderboard
18 Llama 3.2 11B 39.1% Sep 2024 huggingface.co Official leaderboard
19 llava-v1.6-mistral-7b 28.3% Jan 2024 huggingface.co Official leaderboard
#ModelScoreSettingDateLinkSource
1 Gemini 3 Pro 77.2% Nov 2025 deepmind.google Official leaderboard
2 Gemini 2.5 Pro Jun 2025 71.7% Jun 2025 deepmind.google Official leaderboard
3 Gemini 2.5 Flash Sep 2025 66.3% Sep 2025 deepmind.google Official leaderboard
4 o3 58.7% Apr 2025 platform.openai.com Official leaderboard
5 Qwen3-VL-32B-Instruct 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 Llama 4 Scout 53.3% Apr 2025 huggingface.co Official leaderboard
9 Qwen3-VL-8B-Instruct 53.3% Oct 2025 huggingface.co Official leaderboard
10 InternVL3.5-38B 50% Aug 2025 huggingface.co Official leaderboard
11 InternVL3.5-8B 50% Aug 2025 huggingface.co Official leaderboard
12 GPT-4o Mar 2025 48.9% Mar 2025 platform.openai.com Official leaderboard
13 Qwen3-VL-30B-A3B-Instruct 47.8% Oct 2025 huggingface.co Official leaderboard
14 LLaVA-v1.6-34B 42.4% Jan 2024 huggingface.co Official leaderboard
15 InternVL3.5-30B-A3B 41.3% Aug 2025 huggingface.co Official leaderboard
16 Phi-4-multimodal 40.2% Feb 2025 huggingface.co Official leaderboard
17 Llama 3.2 11B 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. 1 Gemini 2.5 Pro Jun 2025 ↑1 80.6%
  2. 2 Gemini 3 Pro ↓1 80.6%
  3. 3 Gemini 2.5 Flash Sep 2025 77.4%
  4. 4 o3 64.5%
  5. 5 Claude Sonnet 4.5 ↑1 64.5%
  6. 6 GPT-4o Mar 2025 ↑6 61.3%
  7. 7 Llama 4 Scout ↑1 58.1%
  8. 8 InternVL3.5-8B ↑3 58.1%
  9. 9 InternVL3.5-30B-A3B ↑6 51.6%
  10. 10 Qwen3-VL-30B-A3B-Instruct ↑3 51.6%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Google DeepMind
Own leaderboard Not tested since Apr 2026
Gemini 3 Pro 77.2%

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.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) Gemini 3 Pro Gemini 2.5 Pro o3 Llama 4 Scout GPT-4o LLaVA-v1.6-34B 0% 20% 40% 60% 80% 100% 2024 2025 2026 Near the ceiling (85%) Gemini 3 Pro o3

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

All 636 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.

How we check