YOMI-Bench - Rhyme Selection (Hiragana): leaderboard

Metric: Accuracy (%; four-option choice of the word sharing the vowel sequence of a word written in hiragana; 120 items; mean over five paraphrased prompts; accuracy as a fraction x100). Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked.

Top models

#ModelScore
1GPT-599.8
2Claude Sonnet 4.598.4
3Gemini 2.5 Flash94.8
4Mistral Medium 3.157.4
5Llama-3.1-Swallow-8B-Instruct-v0.550.2
6Ministral-8B-Instruct-241034
7GPT-4o25.8

Interactive version: theaggregate.ai/benchmark?slug=yomi-bench-rhyme-selection-hiragana · How It Works · Data refreshed daily, snapshot 2026-09-29.