MultiCaption (Zero-Shot) - English Translations: leaderboard

Metric: Accuracy (%; the same 4,920 test pairs with the claims in their English translations; zero-shot yes/no contradiction judgment at temperature 0.1, mean of five runs). Source: arxiv.org. Saturation forecast: Estimated already saturated. 5 models tracked.

Top models

#ModelScore
1Mistral 7B Instruct (v0.3)64.4
2Qwen 2.5 7B Instruct58.9
3Llama 3.1 8B Instruct53.6
4gemma-7B49.1
5Phi-4 Mini Instruct48.9

Interactive version: theaggregate.ai/benchmark?slug=multicaption-zero-shot-english-translations · How It Works · Data refreshed daily, snapshot 2026-09-26.