HAERAE-Vision - Transportation: leaderboard

The 35 transportation questions of HAERAE-Vision (What Users Leave Unsaid, Choi et al., ACL 2026 Findings): real Korean community questions about cars, bicycles and transport, asked with images in the users' original under-specified wording. A GPT-5-mini judge scores each answer against a 1-5 item checklist (met 1, partly met 0.5, not met 0); the score is the mean checklist credit per question, in percent, averaged over three runs; higher is better.

Metric: Checklist Score (%; original under-specified queries, GPT-5-mini judge, mean of 3 runs). Source: arxiv.org. Saturation forecast: Around February 2027. 45 models tracked.

Top models

#ModelScoreOverall rank
1Gemini 2.5 Pro57.69#262
2GPT-554.97#163
3Grok 450.43#306
4GPT-5 Mini49.22#349
5Gemini 2.5 Flash48.12#452
6Qwen 3 VL 30B A3B (Thinking)37.4#710
7Qwen 3 VL 235B A22B Instruct36.33#521
8Qwen 3 VL 32B Instruct36.02#582
9Qwen 3 VL 32B (Thinking)35.57#557
10Qwen 3 VL 235B A22B (Thinking)34.12#451
11Gemini 2.5 Flash Lite34.1#958
12Qwen 3 VL 30B A3B Instruct30.22#870
13GPT-5 Nano26.17#888
14Mistral Medium 3.125.46#852
15Qwen 3 VL 8B (Thinking)24.9#887

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=haerae-vision-transportation · How It Works · Data refreshed daily, snapshot 2026-11-12.