VNU-Bench - Narrative Reconstruction: leaderboard
Metric: Accuracy (%; question type T9, 4-option multiple choice). Source: arxiv.org. Saturation forecast: Around 2029. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 56.52 |
| 2 | Claude Sonnet 4.5 | 55.74 |
| 3 | Qwen 3 VL 8B | 47.13 |
| 4 | Gemini 2.5 Flash | 46.9 |
| 5 | InternVL3-8B | 46.08 |
| 6 | Qwen 2.5 VL 7B Instruct | 45.08 |
Interactive version: theaggregate.ai/benchmark?slug=vnu-bench-narrative-reconstruction · How It Works · Data refreshed daily, snapshot 2026-09-25.