VGenST-Bench - Behavioral Trigger Identification: leaderboard
Metric: Behavioral Trigger Identification (environmental scale, exocentric, dynamic) accuracy (%) on the base multiple-choice questions of VGenST-Bench under circular evaluation (correct only if right under every cyclic permutation of the options), macro-averaged over the task's QA types, 8 uniformly sampled frames from about 100 synthesized videos per task (generated with text-to-image and image-to-video models from validated scene graphs); higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash | 96.9 |
| 2 | GPT-5.4 | 90.1 |
| 3 | Gemini 3.1 Flash Lite | 88 |
| 4 | Gemma 4 31B (IT) | 84.6 |
| 5 | Kimi K2.6 | 80.9 |
| 6 | GPT-5.4 Mini | 80.2 |
| 7 | Gemma 4 26B A4B (IT) | 79.7 |
| 8 | GPT-5.4 Nano | 78.8 |
| 9 | Qwen 3.5 27B | 64.8 |
| 10 | Qwen 3.5 9B | 63.9 |
| 11 | Qwen 3.5 4B | 58.9 |
Interactive version: theaggregate.ai/benchmark?slug=vgenst-bench-behavioral-trigger-identification · How It Works · Data refreshed daily, snapshot 2026-10-07.