Home / Technology / New GPT-Live: More Natural ChatGPT Voice Chats
New GPT-Live: More Natural ChatGPT Voice Chats
10 Aug
Summary
- ChatGPT's voice mode now supports simultaneous listening and speaking.
- The new GPT-Live model enhances conversational naturalness.
- GPT-Live can manage complex tasks by delegating to other models.
ChatGPT's voice conversations have reached a new level of naturalness with the introduction of the GPT-Live model. This advanced system utilizes a full-duplex architecture, enabling it to listen and speak concurrently, eliminating awkward pauses and interruptions previously experienced.
Unlike earlier versions that relied on multiple systems, GPT-Live integrates a single multimodal model for a more cohesive performance. The AI now occasionally vocalizes acknowledgments like "mhmm" during user speech, mirroring human conversational cues.
Furthermore, GPT-Live possesses the capability to delegate complex queries requiring advanced reasoning or research to other sophisticated models, such as GPT-5.5, in the background. This ensures that users receive comprehensive and accurate responses efficiently.
