TAR-Bench - Binary QA: leaderboard

Metric: Accuracy (%; yes/no questions about the anomaly, one positive and one negative per clip; zero-shot in non-reasoning (direct-answer) mode through VLMEvalKit, mean of 3 runs, on TAR-Bench: 960 human-curated annotations for 80 held-out traffic-anomaly clips from 17 public YouTube videos). Source: arxiv.org. Saturation forecast: Around 2029. 11 models tracked.

Top models

#ModelScore
1Qwen 3.5 27B (Non-reasoning)72.5
2Qwen 3.5 9B (Non-reasoning)67.3
3Qwen 3 VL 32B Instruct65
4Gemini 3.1 Pro (Preview)64.4
5Qwen 3 VL 8B Instruct59.6
6Gemma 4 31B (IT)56.5

Interactive version: theaggregate.ai/benchmark?slug=tar-bench-binary-qa · How It Works · Data refreshed daily, snapshot 2026-09-29.