DualEvasion - Vocal Confidence: leaderboard

Metric: Macro F1 (%; zero-shot classification of the response audio of 505 earnings-call answers from 60 calls in 2023-2025 as confident or unconfident, against labels from finance professionals; mean of the confident and unconfident class F1). Source: arxiv.org. Saturation forecast: Around 2028. 8 models tracked.

Top models

#ModelScore
1GPT Audio58.5
2Gemini 2.5 Flash57.9
3Gemini 3 Flash56.8
4Gemini 2.5 Flash Lite51.9
5Gemini 3.5 Flash43.8
6Qwen2-Audio-7B-Instruct28
7GPT Audio Mini27.3

Interactive version: theaggregate.ai/benchmark?slug=dualevasion-vocal-confidence · How It Works · Data refreshed daily, snapshot 2026-09-26.