DramaBench: leaderboard
DramaBench evaluates screenplay continuation across six dimensions (format, narrative efficiency, character consistency, emotional depth, logic, conflict) with FDR-validated pairwise comparisons; scored by overall average.
Metric: Overall Score (%). Source: dramabench.pages.dev. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 | 96.04 | #110 |
| 2 | GLM-4.6 | 93.04 | #252 |
| 3 | Qwen 3 Max | 91.73 | #208 |
| 4 | Kimi K2 (Thinking) | 91.65 | #243 (Kimi K2) |
| 5 | DeepSeek V3.2 | 90.99 | #205 |
| 6 | Claude Opus 4.5 | 86.7 | #83 |
| 7 | Gemini 3 Pro (Preview) | 83.9 | #66 |
| 8 | MiniMax-M2 | 80.94 | #317 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=dramabench · How It Works · Data refreshed daily, snapshot 2026-10-20.