MechVQA - Projection and Multi-view: leaderboard

MechVQA (2026) tests multimodal LLMs on understanding mechanical engineering drawings, with 20,778 questions on 3,281 drawings scored as accuracy by three LLM judges against reference answers. This board covers the Projection and Multi-view subtask of the reasoning capability, as accuracy in percent on the test split; higher is better.

Metric: Accuracy (%) on the Projection and Multi-view subtask (Reasoning capability), MechVQA test split (drawing-level 8:1:1 split of 20,778 questions on 3,281 mechanical drawings), answers judged against the reference by three LLM judges (GPT-OSS-120B, DeepSeek-V3.2, Kimi-k2); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 15 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GLM-4.6V62#721
2GPT-560#185
3Gemini 3 Pro (Preview)58#123
4Qwen 3 VL Plus58#514
5Claude Sonnet 4.554#267
6Qwen 3 VL 32B Instruct48#607
7GPT-4o44#754
8Qwen 3 VL 30B A3B Instruct40#903
9MiniCPM-V-4.536#1451
10InternVL3.5-8B34#1618
11Gemma 3 27B (IT)24#1341
12GPT-4o Mini22#1680
13Qwen 3 VL 4B Instruct20#1348
14MiMo-VL-7B-SFT20#1097
15Llama 3.2 11B Instruct8#5213

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=mechvqa-projection-and-multi-view · How It Works · Data refreshed daily, snapshot 2026-11-22.