FMSU-Bench - Paralinguistic Events (Chinese): leaderboard

FMSU-Bench (2026) tests fine-grained speech understanding in Chinese and English with multiple-choice questions whose distractors include subtle acoustic confusions and text-only semantic traps, plus a tagged transcription task. This board is accuracy on identifying the paralinguistic events present in the speech in the Chinese subset; percent, higher is better.

Metric: Accuracy (%) of identifying the paralinguistic events present in the speech (PE), on the FMSU-Bench Chinese subset (500 to 1,000 manually verified items per task, from audiovisual sources), multiple-choice questions whose distractors include fine-grained acoustic confusions and text-only semantic traps; higher is better. Source: arxiv.org. Saturation forecast: Around August 2027. 10 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Pro (Preview)72.1#99
2Qwen 3 Omni 30B A3B65.2#847
3Qwen2.5-Omni-7B64.7#1884
4Gemini 3 Flash63.3#175
5Kimi-Audio59.9#1735
6Gemini 2.5 Flash53.9#463
7MiMo-Audio50.6#931
8Audio Flamingo 345.3
9FMSU-Bench Step-Audio 2 (checkpoint unspecified)41.8
10Qwen2-Audio (FMSU-Bench checkpoint unspecified)14.5

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

Interactive version: theaggregate.ai/benchmark?slug=fmsu-bench-paralinguistic-events-chinese · How It Works · Data refreshed daily, snapshot 2026-11-21.