StoryCloze: leaderboard
Story Cloze Test (Mostafazadeh et al., 2016): choose the right ending for a four-sentence everyday story from two candidates. Archived Papers With Code table of accuracies as each paper reported them, with fine-tuned, zero-shot and few-shot rows; higher is better.
Metric: Accuracy. Source: huggingface.co. Saturation forecast: Rough model projection: around 2026. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | OPT-175B | 79.82 | #1507 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=storycloze · How It Works · Data refreshed daily, snapshot 2026-11-05.