VNU-Bench - Temporal Development: leaderboard
VNU-Bench (Liu et al., University of Florida, 2026) question type in which a multimodal model must put event details reported by several outlets into their temporal order, given a group of YouTube news videos on one event. Each video is given as 4 selected frames and a cleaned transcript cut to 1,000 characters; zero-shot at temperature 0. Accuracy in percent on 4-option multiple-choice questions; higher is better.
Metric: Accuracy (%; question type T8, 4-option multiple choice). Source: arxiv.org. Saturation forecast: Around July 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 59.94 | #262 |
| 2 | Gemini 2.5 Flash | 53.91 | #452 |
| 3 | Claude Sonnet 4.5 | 52.47 | #244 |
| 4 | Qwen 2.5 VL 7B Instruct | 47.53 | #1842 |
| 5 | InternVL3-8B | 44.5 | #1695 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=vnu-bench-temporal-development · How It Works · Data refreshed daily, snapshot 2026-11-12.