HAERAE-Vision - Machinery: leaderboard
The 22 machinery questions of HAERAE-Vision (What Users Leave Unsaid, Choi et al., ACL 2026 Findings): real Korean community questions that ask what a pictured machine part, tool or piece of equipment is and what it does, 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
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 56.22 | #262 |
| 2 | GPT-5 | 52.85 | #163 |
| 3 | GPT-5 Mini | 51.68 | #349 |
| 4 | Grok 4 | 47.63 | #306 |
| 5 | Qwen 3 VL 235B A22B Instruct | 47.42 | #521 |
| 6 | Gemini 2.5 Flash | 45.86 | #452 |
| 7 | Qwen 3 VL 32B Instruct | 41.73 | #582 |
| 8 | Qwen 3 VL 30B A3B (Thinking) | 38.48 | #710 |
| 9 | Qwen 3 VL 235B A22B (Thinking) | 38.37 | #451 |
| 10 | Gemini 2.5 Flash Lite | 38.3 | #958 |
| 11 | Qwen 3 VL 32B (Thinking) | 35.66 | #557 |
| 12 | Qwen 3 VL 30B A3B Instruct | 34.18 | #870 |
| 13 | Mistral Medium 3.1 | 34.14 | #852 |
| 14 | Qwen 3 VL 8B (Thinking) | 29.73 | #887 |
| 15 | Qwen 3 VL 4B (Thinking) | 29.72 | #1303 |
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-machinery · How It Works · Data refreshed daily, snapshot 2026-11-12.