HumDial-EIBench - Text Empathy (Chinese, Qwen3-Omni Judge): leaderboard

Text empathy and insight of the replies an audio language model gives in Chinese multi-turn emotional dialogues of HumDial-EIBench (2026), real human-recorded multi-turn dialogues from the ICASSP 2026 HumDial Challenge, rated from 1 to 5 by a Qwen3-Omni judge with the benchmark rubric. Higher is better.

Metric: Text empathy and insight score (1-5) of the model's empathetic spoken reply in Task 3 (Chinese multi-turn dialogues from real human-recorded dialogues from the ICASSP 2026 HumDial Challenge test set, audio language models answer from the audio), rated by a Qwen3-Omni judge with the HumDial-EIBench rubric; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-4o Audio3.77#1129
2Freeze-Omni3.69
3Gemini 2.5 Flash3.68#463
4Kimi-Audio3.62#1735
5GLM-4-Voice3.62
6Step-Audio-2-mini3.48
7Qwen2.5-Omni3.41#1588
8Doubao-realtime3.24

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=humdial-eibench-text-empathy-chinese-qwen3-omni-judge · How It Works · Data refreshed daily, snapshot 2026-11-21.