TAR-Bench - Multiple-Choice QA with Explanation: leaderboard

Metric: Scaled BERTScore F1 (%; a multiple-choice answer with an explanation, compared with the reference; 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 2031. 11 models tracked.

Top models

#ModelScore
1Qwen 3.5 9B (Non-reasoning)70.2
2Qwen 3.5 27B (Non-reasoning)55.9
3Gemma 4 31B (IT)36.7
4Gemini 3.1 Pro (Preview)36.2
5Qwen 3 VL 8B Instruct32
6Qwen 3 VL 32B Instruct27.8

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