TraceAV-Bench - Forward Causal Reasoning: leaderboard

The Forward Causal Reasoning sub-task of TraceAV-Bench (2026), part of its audio-visual joint reasoning dimension: multiple-choice questions over long videos whose answers chain several timestamped evidence hops. The score is accuracy in percent with full audio and video input; higher is better.

Metric: Accuracy (%) on the 73 Forward Causal Reasoning (FCR) questions, 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 3.1 Pro (Preview)86.3#99
2Gemini 3 Flash83.56#175
3Gemini 2.5 Pro76.71#276
4Gemini 2.5 Flash75.34#463
5Gemini 2.0 Flash64.38#723
6Ming-Flash-Omni-2.057.53#1266
7OmniVinci-9B57.53
8Qwen 3 Omni 30B A3B43.84#847
9Gemma 4 E4B36.99#2184
10HumanOmni-7B31.51#3308
11Qwen2.5-Omni-7B30.14#1884
12video-SALMONN 230.14#2687
13MiniCPM-o-4.528.77#1252
14VideoLLaMA2.1-7B-AV17.81#4204
15Baichuan-Omni-1.515.07#3285

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-forward-causal-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-21.