VoxSafeBench - Emotion (Chinese): leaderboard

VoxSafeBench (2026) Tier 2 safety test for speech language models with Chinese audio where the words are harmless but the audio is not: the speaker sounds angry or agitated while asking about dangerous tools. The score is the percentage of responses that warn or refuse while naming the audio cue, judged by DeepSeek-V3, higher is better.

Metric: Safety awareness rate (%), the last of the four DAR / WAR / RtA / SAR parts: VoxSafeBench Tier 2 spoken requests that are safe for a neutral adult but call for caution because of an audio cue; DeepSeek-V3 labels each response as a direct answer, a warning that names the cue, a refusal that names the cue, or skip, and the score is warnings plus refusals; Chinese audio; here the speaker sounds angry or agitated while asking about dangerous tools; higher is better. Source: arxiv.org. 7 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3 Flash37.5#180
2Gemini 3 Pro27.5#146
3Kimi-Audio2.5#1722
4Qwen 3 Omni 30B A3B (Thinking)2.5
5Qwen3 Omni 30B A3B Instruct0#928
6GPT-4o Audio0#1123
7MiMo-Audio-7B-Instruct0#1328

Interactive version: theaggregate.ai/benchmark?slug=voxsafebench-emotion-chinese · How It Works · Data refreshed daily, snapshot 2026-11-22.