TraceAV-Bench - Entity Tracking: leaderboard

The Entity Tracking 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 124 Entity Tracking (ET) 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)71.77#102
2Gemini 3 Flash65.32#180
3Gemini 2.5 Flash62.1#466
4Gemini 2.5 Pro60.48#280
5Gemini 2.0 Flash58.06#731
6Ming-Flash-Omni-2.047.58#1258
7OmniVinci-9B38.71
8MiniCPM-o-4.537.9#1251
9Qwen2.5-Omni-7B37.1#1888
10Gemma 4 E4B37.1#2199
11Qwen 3 Omni 30B A3B35.48#851
12HumanOmni-7B29.84#3241
13video-SALMONN 229.03#2695
14VideoLLaMA2.1-7B-AV25.81#4199
15Baichuan-Omni-1.520.97#3189

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=traceav-bench-entity-tracking · How It Works · Data refreshed daily, snapshot 2026-11-22.