ECAC - Caption Quality: leaderboard
Overall caption quality on ECAC (2026), kindergarten daily-activity photos captioned zero-shot, scored from 0 to 100 by a Moonshot vision judge that rates the naming of toys and teaching aids, the description of expressions and actions, and the identification of the teaching area. Higher is better.
Metric: Overall caption quality score (0-100) from a moonshot-v1-8k-vision-preview judge that sees the image and rates the accuracy and professionalism of toy and teaching-aid names, the description of people's expressions and actions, and the identification of the teaching area; zero-shot captioning of ECAC test images; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 4 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 VL 7B Instruct | 85.7 | #1902 |
| 2 | Qwen 2.5 VL 3B Instruct | 84.63 | #2713 |
| 3 | GLM-4V-9B | 78.19 | #3142 |
| 4 | Gemma 3 27B | 12 | #1663 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=ecac-caption-quality · How It Works · Data refreshed daily, snapshot 2026-11-21.