CCTVBench (Video-Level Batch): leaderboard
CCTVBench (2026) quadruple accuracy (%) on real versus counterfactual traffic videos with mutually exclusive yes/no hypotheses, with all of a video's questions sent in one batch request (the paper's protocol for proprietary APIs); higher is better.
Metric: Quadruple accuracy (%): share of question pairs in which the model answers Yes to the true hypothesis on the real video and No to the other three video-question combinations, on CCTVBench's 305 contrastive traffic scenes (a real accident video paired with a world-model-generated counterfactual video) and 1,776 pairs of mutually exclusive yes/no hypothesis questions (7,104 binary instances), single-word Yes/No answers at temperature 0, all of a video's questions in one video-level batch request, each answer parsed separately (the paper's protocol for proprietary APIs); higher is better. Source: arxiv.org. Saturation forecast: Around April 2027. 4 models tracked. Updated 14 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 Mini | 25.85 | #364 |
| 2 | Gemini 2.5 Flash | 20.87 | #463 |
| 3 | Gemini 2.5 Flash Lite | 18.28 | #964 |
| 4 | GPT-5 Nano | 12.6 | #906 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=cctvbench-video-level-batch · How It Works · Data refreshed daily, snapshot 2026-11-21.