ChatGPT Voice: New Model Slows Down on Command

ChatGPT's new voice mode will slow down if you tell it to. It will also acknowledge you as you speak. OpenAI is releasing two new voice models, GPT-Live-1 and GPT-Live-1 mini. These models bring major enhancements to ChatGPT's voice functionality. The previous Advanced Voice Mode had to wait for users to stop speaking before responding, leading to stilted conversations. The new GPT-Live-1 models use a duplex architecture. This means they can process your input and generate a response at the same time. This allows for a more natural back-and-forth. You can interrupt the new voice models at any time. They will also confirm they are listening with verbal acknowledgements like "mhmm." These models can also delegate tasks to other OpenAI models, including GPT-5.5, for things like web searches. ChatGPT Voice can now generate visuals for topics like weather and sports. It also includes new safeguards to steer the model toward safer responses and can notify parents in certain situations. This means a more fluid and responsive conversational experience is coming to ChatGPT.