CASU - Entity Extraction: leaderboard
Context-aware auditory scene understanding: four-option questions on short semi-synthetic clips that mix speech, sound events and background, asking which entity in the scene explains an observation, often implied only by a sound; CASU benchmark (Zhang et al., UC Irvine, 2026). Accuracy in percent, higher is better.
Metric: Accuracy (%; four-option multiple-choice questions on semi-synthetic auditory scenes that combine speech, sound events and background; entity extraction: identify the entity in the scene that explains an observation). Source: arxiv.org. Saturation forecast: Estimated already saturated. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.0 Flash | 72.77 | #702 |
| 2 | Qwen3 Omni 30B A3B Instruct | 68.85 | #902 |
| 3 | Qwen2.5-Omni-7B | 61.1 | #1797 |
| 4 | Voxtral-Small-24B-2507 | 41.67 | #1291 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=casu-entity-extraction · How It Works · Data refreshed daily, snapshot 2026-11-12.