RefereeBench - Foul Existence: leaderboard

RefereeBench (2026) tests multimodal models as sports referees on 925 officiating videos across 11 sports with 6,475 human-annotated multiple-choice questions. This board is accuracy in percent on questions about whether a foul occurred, averaged across sports; higher is better.

Metric: Accuracy (%) on foul-existence questions (Q1), averaged across 11 sports; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Seed 1.892.5#252
2Claude Haiku 4.581#583
3Gemini 3 Flash74.2#175
4Gemini 3 Pro71#145
5Claude Sonnet 4.565.4#261
6GPT-564.2#178
7GPT-4o63#751
8Qwen 3 VL 8B50.8#917
9InternVL3.5-8B43.9#1604
10LLaVA-Video-7B37.2#2276
11VideoLLaMA3-7B33.2#2050

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=refereebench-foul-existence · How It Works · Data refreshed daily, snapshot 2026-11-21.