BREAKING
OpenAI Launches GPT-Live Voice
From Cascaded Pipeline to Full-Duplex
1Speech to text
2LM reasoning
3Text to speech
4Full-duplex flow
Delegation Lifts GPQA Reasoning
Advanced Voice45.3
GPT-Live-184.2
BrowseComp Web Search Surges
Advanced Voice0.7
GPT-Live-175.2
Strengths and Current Limits
Praised
Natural backchannels
Smooth interruptions
Background delegation
Limits
Memory and tools
Non-native accents
No video yet
Voice May Become the Interface
AI NEWS BLITZ
OpenAI has unveiled GPT-Live, voice models that listen and speak at the same time.