Fun-Audio-Chat-8B: benchmark results

Provider: Other.

Aggregate Rating 1556 ± 49, rank #1492 of 10516 rated models, from 18 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
DuplexSpeechBench-IFEval (IAS)47.1Instruction Adherence Score (%; deterministic real-time floo100
DuplexSpeechBench-IFEval (PAS)65.9Persona Adherence Score (0-100; GPT-4o-judged persona-consis75
PolySpeech-10052.88Accuracy (%; spoken four-option reading comprehension from B71.4
PolySpeech-100 - Chinese Dialects77.06Accuracy (%; spoken four-option reading comprehension from B71.4
PolySpeech-100 - High-Resource Languages84.82Accuracy (%; spoken four-option reading comprehension from B71.4
SpeechEQ36.5Trajectory accuracy (x 100; both forced choices of a dialogu71.4
LongShOTBench - Core Perception25.25Weighted Rubric Score (%)70.3
LongShOTBench - Information30.28Weighted Rubric Score (%)70.3
LongShOTBench - Medium Videos29.67Weighted Rubric Score (%)70.3
LongShOTBench - Speech30.12Weighted Rubric Score (%)70.3
LongShOTBench - Visual29.58Weighted Rubric Score (%)70.3
LongShOTBench - Audio27.16Weighted Rubric Score (%)68.3

Interactive version: theaggregate.ai/model?slug=fun-audio-chat-8b · How It Works · Data refreshed daily, snapshot 2026-11-21.