Introducing GPT‑Live

OpenAI has upgraded the voice mode in ChatGPT with a new model called GPT-Live, which uses GPT-5.5 in the background for complex tasks. According to Simon Willison, who had preview access, GPT-Live marks a significant improvement over the previous voice mode, which was based on an older GPT-4o model with a 2024 knowledge cutoff. The new model can delegate harder tasks—such as web search or deeper reasoning—to GPT-5.5, returning results while maintaining conversational flow. Willison had stopped using voice mode due to the older model's limitations, but found GPT-Live much more useful as a brainstorming partner. However, he noted a peculiar bug: the model would interrupt to laugh at non-jokes, which felt condescending. OpenAI reportedly tweaked the behavior after his report. For developers building AI workflows, this update suggests that voice interfaces are becoming more capable and integrated with frontier models, enabling more natural, sustained interactions that can handle complex, multi-step tasks without breaking the conversation.