EgoPolice - 1-Second Clips: leaderboard
Metric: Accuracy (%; 5,000 questions on 1-second clips; zero-shot five-option multiple choice (one visible police-civilian action, three actions absent from the clip and None of the above; chance 20%) on minimally edited COPA and Pasadena police body-worn camera clips; images at 1 fps (at least 8 frames) for image VLMs, author-recommended sampling for video VLMs, one shared prompt). Source: arxiv.org. Saturation forecast: Estimated already saturated. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 | 77.7 |
| 2 | Gemini 2.5 Pro | 76.9 |
| 3 | Gemini 2.5 Flash | 65.8 |
| 4 | InternVL3-8B | 64.7 |
| 5 | Qwen 2.5 VL 7B Instruct | 52.5 |
| 6 | GPT-4.1 Nano | 18.6 |
Interactive version: theaggregate.ai/benchmark?slug=egopolice-1-second-clips · How It Works · Data refreshed daily, snapshot 2026-09-29.