List your company

Speech-to-Speech Question-Answering on LlamaQ

76.5AccuracyMinMo
-1.70818.59638.959.204Jan 30, 2026Feb 25, 2026Mar 23, 2026Apr 19, 2026May 15, 2026Jun 10, 2026Jul 7, 2026

Evaluation results

MethodLinks
MinMoType=AR, S→S capability=true2026.0176.5
DIFFUSPEECHType=Diff., S→S capability=true2026.0172
Llama-Omni2Type=AR, S→S capability=true2026.0170.3
MoshiType=AR, S→S capability=true2026.0168.3
Lychee-FDType=Native2026.0765.3
Fun-Audio-ChatType=Native2026.0764.3
StepAudio-2-miniType=Half-duplex2026.0762
SALMONN-omniType=Native2026.0761.7
Lychee-FD (w/o Sem-Channel)Type=Native2026.0761
MoshiType=Native2026.0754.7
VITA 1.5Type=System-level2026.0751
Freeze-OmniType=System-level2026.0750.7
SpiritLMType=AR, S→S capability=true2026.0140.1
FLM-audioType=Native2026.0736.7
Lychee-FD (w/o Param-Sep)Type=Native2026.0736
SpeechGPTType=AR, S→S capability=true2026.0121.6
dGSLMType=Native2026.071.3