Neo-Classic - Regulated Verse Line Ordering (CoT, Expert Prompt): leaderboard
Metric: Accuracy (%; restore the order of the eight shuffled lines of a contemporary regulated verse (Lyu) with an expert chain-of-thought prompt describing rhyme, parallelism and discourse checks; accuracy on a 100-question sample of the task, as reported in the paper's full results table (accuracy@IF); temperature 0.1, top-p 0.9). Source: arxiv.org. Saturation forecast: Around December 2026. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 36 |
| 2 | GPT-5 Pro | 24 |
| 3 | DeepSeek V3.2 | 13.8 |
| 4 | Qwen 3 Max | 13 |
| 5 | Gemini 2.5 Pro | 8 |
| 6 | GPT-4o | 0 |
| 7 | Yi 1.5 34B Chat | 0 |
Interactive version: theaggregate.ai/benchmark?slug=neo-classic-regulated-verse-line-ordering-cot-expert-prompt · How It Works · Data refreshed daily, snapshot 2026-09-26.