SpeechEQ: leaderboard
SpeechEQ (Wu et al., 2026) tests emotional intelligence in spoken dialogue: in 2,265 three-exchange dialogues covering the 15 EQ-i 2.0 subscales, the model twice chooses the socially appropriate of two synthesized replies with identical words but different tone of voice. Trajectory accuracy counts a dialogue only when both choices are correct, in percent (higher is better; chance is 25).
Metric: Trajectory accuracy (x 100; both forced choices of a dialogue correct, between two synthesized replies with identical transcripts and different tone at turns 4 and 6, 2,265 dialogues over 15 EQ-i 2.0 subscales; chance 25). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 44.9 | #262 |
| 2 | Qwen2.5-Omni-7B | 26 | #1797 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=speecheq · How It Works · Data refreshed daily, snapshot 2026-11-12.