TraceAV-Bench - Temporal Splicing Fallacy: leaderboard

The Temporal Splicing Fallacy sub-task of TraceAV-Bench (2026), in its multimodal hallucination dimension: questions that splice events from different moments into one false premise in long audio-visual videos. The score is accuracy in percent with full audio and video input; higher is better.

Metric: Accuracy (%) on the 166 Temporal Splicing Fallacy (TSF) questions of the multimodal hallucination dimension, four-option multiple choice over long audio-visual videos (10 to 140 minutes), exact match of the selected option set, each model at its official inference setting, with the full audio and video input; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 15 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash87.35#175
2Gemini 3.1 Pro (Preview)84.34#99
3Gemini 2.0 Flash77.11#723
4Gemini 2.5 Pro75.9#276
5Gemini 2.5 Flash66.87#463
6Qwen 3 Omni 30B A3B66.87#847
7Ming-Flash-Omni-2.062.65#1266
8MiniCPM-o-4.556.63#1252
9Gemma 4 E4B56.02#2184
10Qwen2.5-Omni-7B53.61#1884
11OmniVinci-9B42.17
12video-SALMONN 237.95#2687
13HumanOmni-7B28.31#3308
14Baichuan-Omni-1.522.29#3285
15VideoLLaMA2.1-7B-AV19.88#4204

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

Interactive version: theaggregate.ai/benchmark?slug=traceav-bench-temporal-splicing-fallacy · How It Works · Data refreshed daily, snapshot 2026-11-21.