IndustryBench-MIPU - Single-Image: leaderboard
Metric: Image-level F1 (%) of single-image extraction (each image processed independently), structured attribute value extraction from industrial product images (specification tables, nameplates, technical drawings): property names matched exactly and values by rule-based normalization then a Qwen 3.6 Plus semantic judge; full extraction prompt, thinking enabled where supported; higher is better. Source: arxiv.org. Saturation forecast: Around April 2028. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 Plus (Thinking) | 81.3 |
| 2 | Gemini 3.1 Pro (Preview) | 77.1 |
| 3 | Qwen 3.5 397B A17B | 76 |
| 4 | Qwen 3.5 27B | 71.5 |
| 5 | Kimi K2.5 (Thinking) | 70.9 |
| 6 | Qwen 3.5 122B A10B | 70.5 |
| 7 | Claude Opus 4.6 (Thinking) | 69.4 |
| 8 | Qwen 3.5 35B A3B | 68.7 |
| 9 | GPT-5.4 (Thinking) | 66.2 |
Interactive version: theaggregate.ai/benchmark?slug=industrybench-mipu-single-image · How It Works · Data refreshed daily, snapshot 2026-09-29.