OVIBench (Multiple Choice) - False Trigger: leaderboard
Metric: Answer-selection accuracy (%; choosing, among four candidate continuations, the one that handles the interruption correctly; multiple-choice subset of 4,710 samples; offline simulation of an interruption arriving while the model is generating its answer about a streaming video (3,200 videos from ActivityNet, MovieChat, QVHighlights, UCF-Crime and YouCook2); false-trigger interruptions, where the model should ignore the interruption and continue). Source: arxiv.org. Saturation forecast: Around December 2026. 6 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Flash | 56.63 |
| 2 | Qwen 2.5 VL 7B | 43.46 |
| 3 | Seed-1.6 | 30.28 |
Interactive version: theaggregate.ai/benchmark?slug=ovibench-multiple-choice-false-trigger · How It Works · Data refreshed daily, snapshot 2026-09-26.