Neo-Classic - Parallel Couplet Matching (CoT): leaderboard
Metric: Accuracy (%; select the matching line of a parallel middle couplet of a contemporary regulated verse, four options, with a chain-of-thought hint; 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. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3.2 | 91 |
| 2 | Gemini 3 Pro (Preview) | 85 |
| 3 | Gemini 2.5 Pro | 79 |
| 4 | Qwen 3 Max | 67 |
| 5 | GPT-4o | 58 |
| 6 | Yi 1.5 34B Chat | 55 |
Interactive version: theaggregate.ai/benchmark?slug=neo-classic-parallel-couplet-matching-cot · How It Works · Data refreshed daily, snapshot 2026-09-26.