Neo-Classic - Word Cloze (Hard Distractors): leaderboard
Metric: Accuracy (%; restore a masked two-character word in a contemporary poem from four candidates whose part of speech and tone match the answer; 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 2031. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3.2 | 58.3 |
| 2 | Qwen 3 Max | 52 |
| 3 | GPT-4o | 50.2 |
| 4 | Gemini 2.5 Pro | 50 |
| 5 | Yi 1.5 34B Chat | 44 |
| 6 | Gemini 3 Pro (Preview) | 42 |
Interactive version: theaggregate.ai/benchmark?slug=neo-classic-word-cloze-hard-distractors · How It Works · Data refreshed daily, snapshot 2026-09-26.